close

Вход

Забыли?

вход по аккаунту

?

Analysis of passive microwave data for large area environmental monitoring

код для вставкиСкачать
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality o f this reproduction is dependent upon the quality o f the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted.
Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zed> Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/321-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ANALYSIS OF PASSIVE MICROWAVE DATA FOR LARGE AREA
ENVIRONMENTAL MONITORING
A Thesis
Presented to
The Faculty o f Graduate Studies
of
The University o f Guelph
by
DUNCAN ANDREW WOOD
In partial fulfilment o f requirements
for the degree of
Master of Science
May, 1996
© Duncan Andrew Wood, 1996
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 * 1
National Library
of Canada
Bibfiotheque rationale
du Canada
Acquisitions and
Bibliographic Services
Acquisitions et
services bibliographiques
395 Wellington Street
Ottawa ON K1A0N4
Canada
395, rue Wellington
Ottawa ON K1A0N4
Canada
YourSle VetrerM renee
OurOe MetreriUrenee
The author has granted a non­
exclusive licence allowing the
National Library o f Canada to
reproduce, loan, distribute or sell
copies of this thesis in microform,
paper or electronic formats.
L’auteur a accorde une licence non
exclusive permettant a la
Bibliotheque nationale du Canada de
reproduire, preter, distribuer ou
vendre des copies de cette these sous
la forme de microfiche/film, de
reproduction sur papier ou sur format
electronique.
The author retains ownership of the
copyright in this thesis. Neither the
thesis nor substantial extracts from it
may be printed or otherwise
reproduced without the author’s
permission.
L’auteur conserve la propriete du
droit d’auteur qui protege cette these.
Ni la these ni des extraits substantiels
de celle-ci ne doivent etre imprimes
ou autrement reproduits sans son
autorisation.
0 - 612 - 24519-5
Canada
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
WOOD
860 643 560
Duncan Andrew
GRADUATE PROGRAM SERVICES
Land Resource Science
MSc
CERTIFICATE OF APPROVAL (MASTER’S THESIS')
The Examination Committee has concluded that the thesis presented by the above-named candidate in partial
fulfilment o f the requirements for the degree
Master o f Science
is worthy o f acceptance and may now be formally submitted to the Dean o f Graduate Studies.
Title:
Analysis of Passive Microwave Data
For Tjirgg Area
Environmental Monitoring
***4
Chair, Master’s Examination Committee
11.
Advisor,
111.
Date:
IV .
T. C
V-
Received by:
A c*
for Dean of Graduate Studies
Date:
7^:
U
rev. v/96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
ANALYSIS OF PASSIVE MICROWAVE DATA FOR LARGE AREA
ENVIRONMENTAL MONITORING
Duncan Andrew Wood
University o f Guelph, 1996
Advisor:
Dr. Richard Protz
Traditionally, “large area” monitoring from satellites has been accomplished using
the Advanced Very High Resolution Radiometer (AVHRR), however, the use o f this
dataset can be significantly affected by the presence o f cloud cover. The Special Sensor
Microwave Imager (SSM/I) records microwave emissions from the Earth surface.
Microwave emissions are independent of cloud cover and therefore provide a useful
dataset in areas of significant cloud cover.
The objectives if this work were to evaluate the synergism between the AVHRR
and SSM/I sensors, establish a method of scaling between the two datasets and interpret
the microwave emissions for the purpose of soil moisture monitoring.
Scaling between the two datasets was accomplished by adding together the
locational accuracies and pixel resolution of both datasets.
Microwave polarization
differences were found to be inversely related with ground temperature. Furthermore, use
of polarization differences facilitated observation o f the temporal evolution of microwave
emissions over three ecoregions.
SSM/I data recorded at 19 GHz was shown to be linearly related to surface
moisture even in the presence of vegetation.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
I wish to thank my supervisor, Dr. Richard Protz, for providing me with an
interesting project and for his helpful comments and insightful discussions during the
course of my study. I would also like to thank Drs. Terry Gillespie and Brian Brisco for
serving on my advisory and examination committee and Dr. Ralph Brown, all o f whom
provided valuable comments on a initial draft o f this thesis.
Thanks to Dr. Irene
Rubinstein (ISTS) for serving on my examination committee and providing assistance with
SSM/I data processing.
A note of thanks is also extended to the following individuals. Mr. Ken Korporal
(Statistics Canada) who graciously provided additional AVHRR data. Ms. Anne Walker
(Environment Canada) who was always willing to answer questions and provide assistance
in any way possible. Dr. Richard Raddatz (Winnipeg Climate Centre) for providing me
with soil moisture data.
Financial support for this project was provided by the Institute for Space and
Terrestrial Science (ISTS) under an Ontario Centers o f Excellence grant.
A special note of appreciation is extended to all those people that provided me
with the inspiration to complete this project, especially those who endured late nights to
help put it all together.
Finally, I wish to say thank-you to my parents for providing me with every
opportunity to learn and grow, I could not have done it without you.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS....................................................................................................i
TABLE OF CONTENTS.......................................................................................................ii
LIST OF TABLES................................................................................................................. v
LIST OF FIGURES...............................................................................................................vi
LIST OF APPENDIX TABLES.........................................................................................viii
1. INTRODUCTION...............................................................................................................1
1.1. Background................................................................................................................... 1
1.2. Research Problem......................................................................................................... 3
1.3. Remote Sensing............................................................................................................ 4
1.4. Study Objectives........................................................................................................... 6
2. LITERATURE REVIEW...................................................................................................7
2.1. Remote Sensing Principles............................................................................................7
2.2. Passive Microwave Sensors........................................................................................ 11
2.2.1. The Defence Meteorological Satellite Program (DMSP)................................... 11
2.2.2. The National Oceanic and Atmospheric Administration (NOAA) Satellites.... 13
2.3. Theory of Passive Microwave Remote Sensing........................................................ 16
2.3.1. Dielectric Properties of Materials.........................................................................19
2.3.2. Surface Roughness Effects................................................................................... 20
2.3.3. Vegetation Effects................................................................................................21
2.4. Microwave Emissions and Soil Properties.................................................................22
2.4.1. Soil Moisture Content..........................................................................................22
2.4.2. Vegetation Effects................................................................................................26
2.4.3. Surface Roughness Effects................................................................................... 28
2.5. Passive Microwave and Land Surface Processes......................................................29
2.6. Microwave Sensing of Earth Surface Processes from Satellites...............................33
2.6.1. Soil Moisture.......................................................................................................33
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.6.2. Vegetation Studies..............................................................................................37
2.7. Advanced Very High Resolution Radiometer (AVHRR)........................................ 39
2.8. Summary o f Literature...............................................................................................40
3. METHODOLOGY..........................................................................................................43
3.1. Introduction............................................................................................................... 43
3.2. Location and Description of the Study Areas.......................................................... 46
3.2.1. Mixed Wood Plain, Southern Ontario................................................................49
3.2.2. Boreal/Shield....................................................................................................... 50
3.2.3. Hudson and James Regions................................................................................. 51
3.2.4. Manitoba Study Area...........................................................................................52
3.3. Data Acquisition........................................................................................................ 53
3.3.1. SSM/I Sensor...................................................................................................... 53
3.3.1.1. SSM/I Scan Geometry.................................................................................. 54
3.3.2. AVHRR Data Processing System (GeoComp).................................................. 57
3.4. Data Preprocessing.................................................................................................... 58
3.4.1. Identification of Suitable SSM/I Study Pixels.................................................... 59
3.5. Merging AVHRR and SSM/I.................................................................................... 61
3.6. Modelled Estimate of Soil Moisture......................................................................... 65
3.7. Statistical Analysis.....................................................................................................70
3.7.1. Characterization of Reflectance D ata.................................................................70
3.7.2. Determining Earth Relationships from Satellites...............................................71
4. RESULTS AND DISCUSSION OF THE MODELLING EXPERIMENTS.............. 72
4.1. Introduction............................................................................................................... 72
4.2. Scaling Experiment....................................................................................................72
4.2.1. SSM/I - AVHRR Scaling Results....................................................................... 72
4.2.2. Discussion and Evaluation o f Observed SSM/I and AVHRR Relationships... 77
4.2.3. Summary............................................................................................................. 93
4.3. Temporal Evolution of SSM/I Signatures.................................................................95
4.3.1. Microwave Polarization Difference Index (MPDI)..........................................95
4.3.1.1. Southern Ontario.......................................................................................... 96
4.3.1.2. Boreal Region............................................................................................. 102
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.1.3. Hudson/James Bay Lowlands.................................................................. 104
4.3.2. Normalized Temperature Brightness (T nb) .......................................................105
4.3.2.1. Southern Ontario.........................................................................................105
4.3.2.2. Boreal Region............................................................................................. 107
4.3.2.3. Hudson/James Bay Lowlands.....................................................................108
4.3.3. Summary............................................................................................................109
4.4. Mapping Soil Moisture Using SSM/I......................................................................110
4.5. Summary of Analyses............................................................................................. 119
5. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK............... 121
5.1. Conclusions..............................................................................................................121
5.2. Future Research........................................................................................................125
5.2.1. Outline for Future Processing........................................................................... 125
5.2.2. Direction for Future Work................................................................................ 128
6. REFERENCES............................................................................................................... 131
IV
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table 1. Environmental products available from the SSM/I sensor............................. 13
Table 2. Characteristics o f the NOAA satellites........................................................... 15
Table 3. Comparison of remote sensing approaches for soil moisture sensing............41
Table 4. Spatial Resolution o f SSM/I channels.............................................................54
Table 5. Number of pixels sampled for each ecoregion............................................... 61
Table 6. Name and formulation o f microwave algorithms...........................................63
Table 7. Summary statistics for the 1992 SSM/I and AVHRR channel
and channel combinations for the Ontario study site.................................................... 73
Table 8. Summary statistics for 1993 SSM/I and AVHRR channel
and channel combinations for the Ontario study site.................................................... 74
Table 9. Summary statistics for 1994 SSM/I and AVHRR channel
and channel combinations for the Ontario study site.................................................... 75
Table 10. Correlations between AVHRR and SSM/I channels and channel
combinations.................................................................................................................. 76
Table 11. Regression coefficients for the respective vegetation densities..................115
Table 12. Surface moisture retreival algorithms for the respective vegetation
densities.......................................................................................................................118
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES
Figure I. Divisions o f the electromagnetic spectrum..................................................... 9
Figure 2. The interaction of incident energy with the Earth’s surface......................... II
Figure 3. Spectral distribution o f energy radiated from blackbodies of various
temperatures....................................................................................................................18
Figure 4. R ow chart o f the investigative framework................................................... 45
Figure 5. Location of the Ontario transect....................................................................47
Figure 6. Location o f the Manitoba study site.............................................................. 48
Figure 7. Instantaneous Field of View for the SSM/I operating frequencies.............. 56
Figure 8. Registration of SSM/I and AVHRR datasets............................................... 65
Figure 9. Row chart o f soil moisture modelling procedure..........................................68
Figure 10. Snow cover extent for April 2, 1992........................................................... 79
Figure 11. Snow cover extent for May 26, 1993.......................................................... 80
Figure 12. Snow cover extent for May 28, 1994.......................................................... 81
Figure 13. Relationship between the MPDI (37GH z) and NDVI for all three
years of data combined.................................................................................................84
Figure 14. Results of comparisons between NDVI and MPDI 37 for various
geographic regions......................................................................................................... 85
Figure 15. Relationship between the MPDI and NDVI for 1993 and 1994 over
Ontario............................................................................................................................87
Figure 16. Diurnal temperature range for the Ontario transect in May........................91
Figure 17. Surface temperature and water content across Ontario transect
at 7:00AM...................................................................................................................... 92
Figure 18. 37 GHz polarization difference for three different ecoregions....................98
Figure 19. Average polarization difference for three different ecoregions.................. 99
Figure 20. Polarization ratio for three different ecoregions.......................................100
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 21. 19 GHz polarization difference for three different ecoregions................101
Figure 22. Normalized temperature brightness for three different ecoregions
106
Figure 23. Relation between average polarization difference and NDVI for
September 6-12, 1993................................................................................................. 112
Figure 24. Scatterplot between soil moisture and emissivity for all vegetation
densities........................................................................................................................ 114
Figure 25. Scatterplot and regression line for medium-high density vegetation
116
Figure 26. Scatterplot and regression line for low density vegetation......................117
Figure 27. Future processing flowchart o f passive microwave data in estimating
soil moisture and vegetation biomass...........................................................................129
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF APPENDIX TABLES
Table A l. Locational Accuracies between SSM/I and AVHRR for each
ecoregion pixel (1992).................................................................................................... 149
Table A2. Locational Accuracies between SSM/I and AVHRR for each
ecoregion pixel (1993)................................................................................................... 150
Table A3. Locational Accuracies between SSM/I and AVHRR for each
ecoregion pixel (1994)................................................................................................... 151
Table A4. Meteorological data from Environment Canada climate stations
located in the Hudson / James Bay region (1988)........................................................ 152
Table A5. Meteorological data from Environment Canada climate stations
located in the Hudson / James Bay region (1990)......................................................... 153
Table A6. Meteorological data from Environment Canada climate stations
located in the Hudson / James Bay region (1992)......................................................... 154
Table A7. Meteorological data from Environment Canada climate stations
located in the Boreal region (1988).............................................................................. 155
Table A8. Meteorological data from Environment Canada climate stations
located in the Boreal region (1990).............................................................................. 156
Table A9. Meteorological data from Environment Canada climate stations
located in the Boreal region (1992).............................................................................. 157
Table A10. Meteorological data from Environment Canada climate stations
located in southern Ontario (1988).............................................................................. 158
Table Al 1. Meteorological data from Environment Canada climate stations
located in southern Ontario (1990).............................................................................. 160
Table A12. Meteorological data from Environment Canada climate stations
located in southern Ontario (1992).............................................................................. 162
Table A13. Meteorological data used in Winnipeg Climate Centre soil
moisture model (Sept. 6-12, 1993).............................................................................. 164
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 1
INTRODUCTION
1.1 Background
Agriculture and related services play an important role as a contributor to the
Gross National Product despite the fact that only seven percent of Canada’s 922 million
hectares is farmland (Dumanski et al., 1986).
In fact in terms o f Canada’s primary
industries, agriculture places second only to the mining and oil well industries, outpricing
both logging and forestry and fishing and trapping industries (Statistics Canada, 1994).
Much o f Canada’s wealth still lies in its cultivated soils.
Grain fields, pasture and
orchards, and other kinds o f farms encompass about 45 million hectares o f Canada, an
area roughly twice the land mass of Great Britain. The goods and products from this land
area account for some 10% of the Canadian economy (Statistics Canada, 1994). The
ultimate success o f most agricultural practices is largely dependent on moisture availability
which varies both in time and space. Soil moisture is an environmental descriptor that
integrates much o f the land surface hydrology and is the foundation of life on the land
surface (Engman, 1992). Utilization o f soil moisture is central to human existence as well
as biogeochemical models. It also provides a vital component as input to crop yield
models, validation o f climate models, flood and weather forecasting as well as drought
estimation.
Along with the myriad of uses of soil moisture has arisen the need to
determine it’s dynamic characteristics at a variety of scales over time. In spite of its
l
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
importance, the majority o f previously identified application areas have not been able to
adequately use soil moisture in the various models. There are two primary reasons for
this. It is a difficult variable to measure, not only at one point in time, but also in a
consistent manner and spatial basis. Furthermore, it exhibits very large spatial and
temporal variability at many different scales. As a result soil moisture has not been used
as a measurable variable in current hydrologic, climatic, agricultural, or biogeochemical
models (Engman, 1992). Protz et al. (1993) noted, however, that it is possible to derive
soil moisture through models involving precipitation and evapotranspiration but due to the
sporadic nature of rainfall, it is difficult to generally obtain a useful measure o f this
variable. In summary, the researchers believed that synoptic and frequent measurements
o f soil moisture by satellite remote sensing would be o f considerable use.
In order to maintain their competitive position in agricultural production,
Canadians must not only manage effectively the land resource on which the industry is
based but also ensure that the resource is maximized and available for future use.
Underlying effective management is a need for quick reliable information about the various
types, location and conditions o f crops under cultivation across Canada and elsewhere.
An increasingly important source of this information is the analysis o f remote
sensing data from earth observation satellites. The use o f remote sensing techniques for
agro-ecosystem monitoring applications is now being promoted (Protz et al., 1993),
however, long before the launch o f the first earth observation satellite, remote sensing
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
techniques were being proposed for agricultural (Estes, 1966) and environmental
applications (Vizy, 1974).
Data from satellites is now recognized as a valuable tool by agencies requiring
information regarding the extent and condition o f agricultural crops. This is primarily due
to their relatively low cost, large geographical coverage, and the repetitive orbits o f the
satellite which allows observation of the same land area repeatedly during the course of a
year.
1.2 Research Problem
One advantage o f passive microwave sensing systems over other remote sensing
systems operating in the near and optical region of the electromagnetic spectrum is that
the wavelength of electromagnetic radiation in the microwave region is several orders of
magnitude larger than those of the optical region. This enables longer microwave energy
(i.e. 1.4 GHz ) to be independent of atmospheric conditions and, more importantly, to
partially penetrate crop canopies. During the early part o f a growing season, before the
vegetation has matured, passive microwave sensors maybe capable of detecting near
surface soil moisture.
Wang (1985) summarized the results of a number of aircraft studies noting that
these experiments generally showed great promise for remotely estimating soil moisture.
Despite such favorable results, studies of large area soil moisture measurements have been
limited in numbers (Wang, 1985; Schmugge et al., 1977) mainly due to the limited number
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of orbiting satellites with near optimal frequencies
(1.4 GHz) for moisture studies.
Furthermore, passive microwave sensors, until recently, have been hampered by their poor
resolution (~ 150 km at 6.6 GHz) resulting in difficulty in interpreting results from such
large areas.
Presently, passive microwave data from the Special Sensor Microwave Imager
(SSM/I) is being used to monitor, among other variables, crop vigor (Teng et al., 1995),
and soil moisture (Heymsfield and Fulton, 1992) to some extent, however, the 19 and 37
GHz frequencies of the SSM/I sensor can be significantly affected by the presence of
vegetation and furthermore the resolutions remain somewhat coarse (47 km at 19.3 GHz).
Protz et al. (1993) concluded that considering no techniques are currently available,
remote or otherwise, that are capable o f providing useful information on the temporal and
spatial variability of soil moisture over large areas, even a resolution o f the order o f 25 to
50 km will probably prove useful.
Before this can be accomplished a better understanding of the complexity of “large
area” pixels as well as the physical interactions between microwave emissions and the
ground surface needs to be understood.
This will be accomplished by integrating
information received from two different satellite platforms having different resolutions,
one relatively high resolution (1.1 km) and one lower resolution (25 km)
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.3 Study Objectives
This study has evolved around the following objectives:
1. To establish a method o f scaling between two sets o f imagery having largely
different scales o f observation. Furthermore, perform data analysis to
determine information content o f microwave imagery recorded at low
resolution (25 km).
2. To analyze the temporal sequence of microwave data recorded over a variety of
different ecoregions and identify potential changes over time.
3. To test the ability o f passive microwave systems (SSM/I) to retrieve the state of
near surface soil moisture.
Chapter 2 of the thesis will discuss basic principles of remote sensing as it pertains
to passive microwave sensing and document previous research findings regarding the
analysis of passive microwave data. The methodology for the current research, including
data acquisition and statistical analysis, is outlined in chapter 3. Chapter 4 provides a
discussion and summary of the research results.
The final chapter draws relevant
conclusions from the present research and provides direction for future work.
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2
LITERATURE REVIEW
2.1 Remote Sensing
Remote sensing is a process by which the spectral response and spatial attributes
o f surface features are recorded by a sensor that does not come into contact with that
feature.
Through a variety of manual and computer-aided techniques, information
regarding the feature and it’s surrounding landscape can be extracted from the remotely
sensed data.
Remote sensing has the potential to provide an alternative technique for
monitoring environmental change over large areas and offers many advantages over in-situ
measures of the same variables. Firstly, compared to point sampling methods, remote
sensing can provide a relatively inexpensive method o f determining rates o f environmental
change as well as measuring soil moisture at a variety of scales. Furthermore, use o f
remotely sensed data has generally proven to be faster than conventional fieldwork
(Stephens et al., 1985).
It has further been proposed by some researchers that remote sensing methods
have, at a minimum, the capability to reduce the need for intensive field sampling and
ultimately the potential to fully replace fieldwork.
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Secondly, remote sensing, particularly the use o f satellite imagery, has the ability to
cover large areas repeatedly over both short and long periods o f time (Cihlar, 1987)
thereby increasing the propensity of a permanent multi-temporal sequence o f imagery.
The use of such records facilitate the evaluation o f long term trends and therefore make
remote sensing an attractive source of information for establishing various environmental
monitoring procedures.
Currently, operational use o f satellite imagery for temporal monitoring on a large
scale has focused on crop yield estimation on the Canadian Prairies and has been primarily
limited to the use o f a single sensor. As the number and use of satellites increases there
will be a need to develop methodologies for using remotely sensed data to extract
information to evaluate dynamic variables such as soil moisture. To this end remote
sensing has the potential of providing timely, inexpensive information pertaining to the
state of the environment.
2.1.1 Remote Sensing Principles
Electromagnetic radiation is a form o f energy transfer in free space, which exhibits
both wave and particle properties (Hunt, 1980).
Based on its wave properties, the
electromagnetic energy travels through space in a plane harmonic wave pattern at the
velocity of light, c = 3 x 10* m/s (Lo, 1986). A wave can be described in terms of its
wavelength (X) which is a function of the distance of separation from one wave crest to
the next (Campbell, 1987). Alternatively, a wave can be described by its frequency ( J ),
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which is the number of wave crests passing a fixed point in a given period o f time
(Campbell, 1987).
Conventionally, optical remote sensing is referred to in terms of
wavelength and for microwave remote sensing, frequency is the given terminology. The
two methods of describing a wave are related by:
Xf= c
or
/ = c /X
where: c = speed of light (m/s)
X = wavelength (m)
/ = frequency (Hz)
The electromagnetic spectrum constitutes a continuum of energy ranging in wavelength
from nanometers to kilometers (Figure 1, Sabins, 1978). Divisions of the electromagnetic
spectrum have occurred as a result of the various methods for sensing each type of
radiation and, as such, are somewhat arbitrarily defined for the purposes of remote
sensing. Remote sensing takes place in three main regions of the spectrum, i) the visible
(0.4 jim - 0.7 jim), ii) the infrared (0.7 jim -14 jim), and iii) the microwave (0.3 cm - 300
cm) regions of the spectrum. The dimension and number of the wavelength intervals in the
electromagnetic spectrum to which a sensor is sensitive determines the spectral resolution
of the sensor (Jensen, 1986). In remote sensing terminology, a wavelength interval is
known as a (wave)band or channel.
Electromagnetic radiation is a dynamic form of energy made manifest only by its
interaction with matter (Suits, 1983). When electromagnetic energy is incident on any
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 1.
Divisions o f the electromagnetic spectrum
(Taken from Lillesand and Kieffer, 1987)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
o
o
(O
o
in
O
oJ ?
✓
n
O
ar
E
co
co
o'
m
o'
•P Jr
in
<s>
•* v *
o
o'
CM
CO
o
O
>
&
o
00
«5
.<T
vS*
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
given earth surface feature or travels between two mediums o f different densities, three
fundamental energy interactions are possible. Electromagnetic energy emitted by the sun
passes through the atmosphere at which point, particularly in the shorter wavelengths,
some of the energy is absorbed by atmospheric gases, and some is reflected by clouds.
The resultant energy that reaches the earth surface is known as “incident energy” and can
either be reflected from the surface or be absorbed/transmitted by the surface (figure 2).
The various fractions o f incident energy partitioned for each type o f energy-matter
interaction is dependent on the surface characteristics.
The majority o f remote sensing
systems record the fraction of incident energy that is reflected (ER), however, there are
also sensors that are capable of detecting emitted energy (a function o f absorbed and
transmitted energy - EA and ET) from earth’s surface, these are known as radiometers.
Electromagnetic radiation that is reflected (E r) or emitted (E e) travels back
through the atmosphere upon which it is detected by the sensor. Data recorded by the
sensor are normally analog electrical signals with voltage variations related to the
measured physical variations (Jensen, 1986). Once the sensor has recorded the voltage
fluctuations, various analog-to-digital (A-to-D) conversion procedures are employed to
convert the original signals to digital numbers (DN).
Converted signals are then
downlinked to Earth receiving stations or stored on board the satellite until such time as
an appropriate ground receiving station can be reached.
Spatial resolution refers to the smallest angular or linear separation between two
objects that can be resolved by the sensor (Jensen, 1986). The basic element of remote
sensing imagery is known as the pixel (picture element). A pixel is a graphical
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 2 .
The interaction o f incident energy with the Earth’s surface
(Modified from Lillesand and Kieffer, 1987)
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
representation o f reflected electromagnetic radiation integrated over a given surface area
to provide a single value per area for each wavelength. Generally, for passive microwave
sensors the spatial resolution is a function o f frequency. Each pixel has a spectral value or
digital number associated with it. This number represents the amount o f incident radiation
reflected within the pixel, and ranges from zero (no radiation or black) to a higher number
(all radiation reflected or white) (Sabins, 1978). The sensitivity o f a particular instrument
to record various reflectance values defines the sensor’s radiometric resolution. As an
example, AVHRR data are generally recorded in a 10-bit format representing a range of
reflectance values from 0 (black) to 1023 (white).
2.2 Passive Microwave Sensors
2.2.1 The Defense Meteorological Satellite Program (DMSP)
The first Special Sensor Microwave/Imager (SSM/I) was launched June 19, 1987
aboard the Defense Meteorological Satellite Program (DMSP) Block 5D-2 F-8 spacecraft
(Hollinger, 1989).
The SSM/I is a seven channel, four frequency, linearly polarized,
passive microwave radiometric system which measures atmospheric, ocean, and terrain
microwave brightness temperatures (which will be discussed in section 2.4) at 19.3, 22.2,
37.0 and 85.5 GHz (Hollinger et al., 1990).
The Block 5D-2 F-8 spacecraft is in a circular sun-synchronous orbit, meaning that
the orbit plane proceeds about the earth at the same angular rate that the earth moves
about the sun (Paine, 1981). As a result o f the orbital characteristics, the satellite has
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approximately a 7:00 AM local ascending equatorial crossing. The SSM/I channels were
chosen primarily for their ability to measure environmental parameters as shown in table
1 (Hollinger, 1989).
Paramietiar
Ocean Surface Wind Speed
Ice - Area Covered
Ice - Age
Ice - Edge Location
Precipitation Over Land
Precipitation Over Water
Soil Moisture
Land Surface Temperature
Snow Water Content
Surface Type
Cloud Amount
......G & iiiitm
(Km)
25
25
50
25
25
25
50
25
25
25
25
Table 1. Environmental products available from the SSM/I sensor.
Therefore, as the operating frequency decreases, the pixel size increases.
Approximate pixel or footprint size for the SSM/I frequencies are 14 km at 85 GHz, 33
km at 37 GHz and 56 km at 19 GHz (Hollinger et al., 1987). The SSM/I represents a
joint Navy/Air Force operational program to obtain synoptic maps of critical atmospheric,
oceanographic, and selected land parameters on a global scale (Hollinger et al., 1987).
The SSM/I data are processed by the Naval Oceanography Command and the Air Weather
Service to obtain near real-time global maps o f cloud water, rain rates, water vapour over
oceans together with land surface type, moisture and vegetation biomass.
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.2.2 The National Oceanic and Atmospheric Administration (NOAA) Satellites
Several generations of satellites have been flown in the NOAA series. Data is
currently being archived by the Advanced Very High Resolution Radiometer (AVHRR),
one of the many sensors on board the NOAA suite o f satellites.
NOAA 11 and 12 orbit the earth at an altitude o f 833 km and complete 14.1 orbits
per day (Lillesand and Kieffer, 1987). Due to the significantly large swath width (ground
distance imaged longitudinally during one satellite pass) and orbital characteristics, the
NOAA satellites are able to provide daily coverage o f the earth surface. NOAA satellites
orbit in a near-polar, sun-synchronous manner. The even-numbered satellite makes a
north-to-south equatorial crossing at 7:30 AM, and the odd numbered satellite has a 2:30
AM north-to-south equatorial crossing time. Basic characteristics o f the NOAA satellites
are presented in table 2.
The AVHRR instrument has five spectral channels, one in the red (0.58 tun - 0.68
fim), one in the reflected infrared (0.72 (un - 1.10 tun) and three recording thermal
infrared (3.55 tun - 12.5 |un) data (Lillesand and Kieffer, 1987). The spatial resolution o f
the AVHRR instrument is 1.1 km at nadir with the resolution becoming coarser as the
viewing angle increases off-nadir. Nadir is referred to as the point on the ground vertically
beneath the perspective center of the satellite field of view (Jensen, 1986). AVHRR data
is received by NOAA at full resolution, processed and archived in two different formats.
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.
Characteristics o f the NOAA satellites
(Adapted from Lillesand and Kieffer, 1987)
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Parameter
Altitude Gan)
Period of orbit (min)
Orbit inclination
Orbits per day
Distance between orbits
Scan angle from nadir
Optical field of view (mrad)
IFOV, at nadir Gan)
IFOV, off-nadir maximum (Km)
Along track
Across track
Swath width (km)
Coverage
NOAA-6,8,10,12
NOAA-7,9,11,14
833
102
833
102
98 9 0
14.1
25.5°
+/-55.40
1.3
1.1
14.1
25.5°
+/-55.40
1.3
l .l
2.4
6.9
2400
Every 12 h
2.4
6.9
2400
Every 12 h
9 g
9 0
AVHRR spectral channels (pm)
1
2
3
4
5
0.58-0.68
0.72 - 1.10
3.55 - 3.93
10.50-11.50
10.50-11.50
0.58 - 0.68
0 .72-1.10
3.55 - 3.93
10.30-11.30
11.50-12.50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Selected data are recorded at full resolution, referred to as local area coverage
(LAC) data. All other data are resampled down to a nominal resolution of 4 km, referred
to as global area coverage (GAC) data (Lillesand and Kieffer, 1987).
The spatial
resolution is 1.1 km for each waveband unlike the Thematic Mapper sensor on board
Landsat-5 which has a spatial resolution o f 30 m for bands 1-5 & 7 and 120 m for band 6.
2.3 Theory o f Passive Microwave Remote Sensing
This section will provide the background theory o f passive microwave sensing, as
well as detail the major components influencing the amount o f electromagnetic radiation
received by the radiometer.
All matter at temperatures above absolute zero radiate electromagnetic energy.
Most natural solid and fluid media found in the terrestrial environment follow
approximately the spectral behaviour o f thermal radiation as described by the Planck
radiation law for the so-called “blackbody” (figure 3). A blackbody is an idealized body
which, when in thermodynamic equilibrium at a given temperature, radiates at least as
much energy as any other body at that same temperature (Ulaby et al., 1981).
A
blackbody is a perfect absorber, that is, it is capable of absorbing 100% of the energy
incident upon it.
Most real materials, usually referred to as “graybodies”, emit less
radiation than a blackbody does. This may be due, in part, to absorption of incident
radiation which is either lost to the atmosphere at a later time or may be used in
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
photosynthesis or plant growth. Furthermore, graybodies are unable to absorb all energy
incident upon them which, in turn, also results in a decrease in the amount o f re­
radiated/emitted energy.
The brightness (B) o f a material relative to that o f a blackbody at the same
temperature is defined as the dimensionless variable known as emissivity (e) (Ulaby et al.,
1981), such that:
6 = Tb / T
[1]
Emissivity values therefore vary between 0, for a perfectly non-emitting material, and 1 for
a perfect emitter.
For passive microwave remote sensing o f ground objects, a radiometer measures
the intensity o f emission from the ground surface. The emission is proportional to the
product of the surface temperature and the surface emissivity which is known as the
microwave brightness temperature (TB) and can be expressed as follows (Schmugge,
1986):
Tb
= t(H) * [r T sky + (1 - r)TsoaJ + T Atm
[2]
where r is the surface reflectivity and t(H) is the atmospheric transmission. The first term
(T sk y )
represents the reflected sky brightness temperature (dependent on wavelength and
atmospheric conditions); the second term ( T So i l ) is emission from the surface (1 - r = e,
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3.
Spectral distribution o f energy radiated from blackbodies o f
various temperatures
(Taken Ulaby et al., 1981)
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced
the emissivity); and the third term (T Atm) corresponds to the contribution from the
atmosphere between the earth and the receiver. A radiometer antenna pointing at an
object senses the Brightness temperature o f an object and presents the power due to the
brightness temperature as a signal to the input of the receiver which is then recorded on a
computer compatible tape. Schanda, (1976) noted that the parameters of the ground
surface that determine the amount of emitted radiation and ultimately brightness
temperature values recorded by the receiver as the complex permittivity o f the material or
mixture of materials, and the shape or roughness of its surface. System parameters also
affecting the intensity of the signal recorded by the system include frequency, polarization
and incidence angle. Irrespective of the frequency, microwave signals can be recorded in
different modes of polarization. Polarization refers to the filtering methods used to restrict
electrical wave vibrations to a single plane relative to the direction o f wave propagation.
If the electric field vector in the electromagnetic wave is perpendicular to the earth’s
surface, it is referred to as vertical (V) polarization. If the vector is parallel to the earth’s
surface, it is called horizontal (H) polarization (Lo, 1986).
Most passive microwave
radiometers are capable of recording both polarizations simultaneously.
2.3.1 Dielectric Properties of Materials
The behaviour of the reflectivity (r) of an object and hence the emissivity (s), is
related to the physical properties of the medium by means of the dielectric constant. The
dielectric constant can be defined as the ratio between the electric displacement and the
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
electric field which is related to the ability of the molecules (or microscopic components)
to align their own or induced dipole moments to the electric field (Luzi et al., 1990). The
parameter contains a real e' (permittivity), and an imaginary part e" (conductivity) which
can also be considered a “loss factor” since it is not measured directly by the radiometer.
Therefore, in the case of an isotropic homogeneous lossy medium the dielectric constant is
represented by a complex number, k = s' + e". In a non-homogeneous medium, the
dielectric constant (k) is a combination o f the dielectric values o f the individual constants,
that is, air, soil, water etc., which has led to the development o f a number o f dielectric
mixing formulas (Jackson and O’Neill, 1986; Dobson et al., 1985; Sihvola and Alanen,
1991). Due to the molecular composition of water, dipole alignment may occur quite
readily upon interaction with incident energy. Thus, the presence of liquid water will very
strongly affect the intensity of the microwave radiation upwelling from materials like soil,
vegetation canopies and snow. As an example, at 18 GHz, the dielectric constant of a wet
soil is approximately 9 compared to that of a dry soil which is on the order of 3-5 (Ulaby
etal., 1986).
2.3.2 Surface Roughness Effects
Roughening the surface of the natural material causes the emissivity to be
somewhat higher (Schmugge, 1985). The emissivity of an object is directly related to the
brightness temperature (T b) therefore any increase in the emissivity results in a
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
corresponding increase in microwave brightness temperature as recorded by the sensor.
This increase in emissivity may be attributed to the increase in the surface area that
interfaces with the air, thereby providing a larger area from which to transmit upwelling
radiation. Quantitatively, the surface roughness increases the emissivity by an amount as
follows (Choudhury et al., 1979):
Ae = rQ(1 - exp(-h))
[3]
Where r0 is the reflectivity for the smooth surface and h is an empirically determined
roughness parameter proportional to the root mean square (RMS) height variations o f the
soil surface (Engman and Chauhan, 199S).
2.3.3 Vegetation Effects
A
fairly simple model for the brightness temperature (TCan) of a scene consisting
o f a semi-infinite soil layer of physical temperature Ts, an air-soil reflectivity T,, and a layer
o f vegetation of physical temperature Tv is given by (Ulaby et al., 1981):
T can =
[1
+rsY] [1 - Y] (1 - a) Tv + [1 -rs] T T S
[4]
where a is the single-scattering albedo of the vegetation layer and T is the transmissivity o f
the vegetation layer at the refraction angle @‘ given by (Ulaby et al., 1986):
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Y (0‘) = exp( - t sec 0 ‘)
[5]
For a canopy o f height h and extinction coefficient k«, the optical thickness is x = kch.
The first term in the equation is the product o f the air-soil reflectivity and the
transmissivity o f the air-soil interface multiplied by the product of the vegetation
temperature and the emissivity. The second term corresponds to the product o f the soil
emissivity, reflectivity and the soil’s physical temperature.
Therefore, the recorded
brightness temperature of a vegetation canopy is comprised of the microwave emission
from the air-soil layer and the interactions with the vegetation layer as well as the
microwave emissions from the vegetation canopy alone.
2.4 Microwave Emissions and Soil Properties
Any portion o f incident radiation reflected or absorbed by a soil surface is primarily
a function o f vegetation cover and soil properties.
Depending on the amount of
vegetation cover, a number of soil properties such as texture, organic matter content,
surface roughness and moisture content are likely to influence the proportion of
emitted/reflected energy. The importance of each of these soil properties increases with a
decrease in vegetation cover.
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.4.1 Soil Moisture Content
A large number of studies have demonstrated the microwave sensitivity to near
surface soil moisture. However, the majority o f these studies have been performed under
regulated experimental conditions whereby various soil types o f known properties have
been successively “wetted-up” and observed at microwave frequencies (Paloscia et al.,
1993). In most studies, a significant negative correlation was reported between near­
surface soil moisture and microwave brightness temperature (T b) or emissivity (e) (Estes
et al., 1977; Newton and Rouse, 1980; Vyas et al., 1985; Jackson et al., 1984; Schmugge
et al., 1992; Jackson, 1993; Paloscia et al., 1993).
In summarizing the results from the AGRISTARS program, Schmugge et al.
(1986) noted that various researchers had found significantly high inverse correlations
between near surface soil moisture and normalized temperature brightness (emissivity) for
bare smooth fields at L-band (1.4 GHz), horizontal polarization at a variety o f different
sites.
Under such controlled conditions a number of researchers have successfully
demonstrated significant correlations between temperature brightness or emissivity and
near surface soil moisture at a number o f different frequencies and incidence angles. Estes
et al. (1977) reported a high positive correlation (r = 0.96) between image tone level and
soil moisture using a 35 GHz, H-polarization airborne scanning passive microwave
radiometer.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Using a 19.1 GHz tower-mounted radiometer, Vyas et al. (198S) reported high
sensitivity to surface soil moisture for smooth bare field conditions at incidence angles
ranging from 10° to 50°.
Newton and Rouse (1980) employed two truck mounted
radiometers operating at 1.4 GHz and 10.6 GHz to evaluate the sensitivity of microwave
radiometer measurements to soil moisture variations. Both frequencies were found to
exhibit strong negative correlations between soil moisture and microwave emissions.
Newton and Rouse (1980) further demonstrated, using a 1.4 GHz radiometer, the
increased microwave sensitivity to soil moisture in the top S cm o f the soil. The sensitivity
to near surface soil moisture was consistent for incidence angles ranging from 10 to 50
degrees using horizontal polarization. A rapid decrease in sensitivity to surface moisture
was found at the shallower incidence angles when the vertical polarization was examined.
Results from other research indicate that near surface soil moisture is significantly
related to microwave emissions recorded at specific frequencies. Research conducted
over the past twenty years has concluded the best frequency for monitoring variations in
near surface soil moisture to be L-band (1.4 GHz). Using laboratory setups, Pampaloni et
al. (1990) and Paloscia et al. (1993) observed that at L-band (1.4 GHz), soil moisture
variations could be measured to a depth of 5 cm while at X-band (10.6 GHz) the depth
decreased to < 1 cm.
Engman (1991) reported that measurement depth
for each
frequency is not constant but, rather related to the total amount o f water in the soil layer,
and thus the moisture content, as well as measurement frequency. Recently, Jackson
(1993) summarized that for the measurement o f near surface soil moisture, the optimum
frequency should be one that has the greatest depth sensitivity to soil moisture.
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Furthermore, at X-band, any day to day changes in moisture are rather small and would
often be undetectable.
The majority of the published literature has relied upon microwave sensors
mounted on truck: platforms for the development of soil moisture estimation algorithms.
Under these relatively controlled conditions, the effects o f surface parameters can be well
examined. However, complementary research has also been conducted using aircraft
sensors for the purpose of examining the effects of large area coverage, scene
heterogeneity and instrument sensitivity at the lower resolutions typical o f airborne sensors
(Schmugge et al., 1974; Jackson et al., 1984; O’Neill, 1985; Wang, 1992; Schmugge et
al., 1992; Ladson and Moore, 1992; Schmugge et al., 1994).
The majority o f earlier experiments using airborne radiometers utilized the sensor
package flown on board the NASA C-130 research aircraft, which included microwave
radiometers at 1.4 GHz and 5.0 GHz.
Jackson et al. (1984) found strong negative
correlations (R2= 0.91) between soil moisture and emissivity over an Oklahoma watershed
when using 1.4 GHz , H-polarization and 40° incidence angle. However, at 5.0 GHz the
relationship between near surface soil moisture and emissivity was much less (R2 = 0.64)
for non vegetated fields and even lower in the presence o f vegetation. More recently
Schmugge et al. (1992) and Schmugge et al. (1994) demonstrated the possibility of
mapping the spatial variation of surface soil moisture from airborne radiometers.
However, these researchers noted the dependence on using lower frequencies (i.e. 1.4
GHz) together with sparsely vegetated test sites.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Using a 1.4 GHz radiometer Ladson and Moore (1992) reported a significant
correlation (R2 = 0.67, p = 0.01) when near surface soil moisture was correlated with
emissivity over a number o f days, however, they showed that soil water content and
emissivity were not well correlated spatially on any single day.
These researchers
attributed the poor single day correlations between soil water content and emissivity to
errors in locating sample sites and to a lesser degree registration between the datasets.
2.4.2 Vegetation Effects
A vegetation canopy covering the soil will absorb and scatter some o f the
microwave radiation incident upon it. Vegetation cover over a soil surface attenuates the
radiation emitted by the soil and adds to the total radiative flux with emissions o f its own.
Researchers have shown experimentally that vegetation reduces the sensitivity o f Tb to
near surface soil moisture content and that the reduction in sensitivity depends primarily
on the type of vegetation as well as the wavelength o f observation (Newton and Rouse,
1980; Brunfeldt and Ulaby, 1986; Schmugge et al., 1986; Jackson and Schmugge, 1991).
Jackson et al. (1982) suggested that the amount of radiation attenuated by the
vegetation canopy is linearly related to the plant water content. Furthermore, the change
in soil moisture sensitivity can be expressed in the following manner (Schmugge et al.,
1986):
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[6]
de/dsm = exp(-2t) de/dsm
Where e, is the bare soil emissivity, t is the vegetation optical depth and sm is the
soil moisture. Using both a 1.4 GHz and 5.0 GHz radiometer, Ulaby et al. (1983)
determined that the presence of vegetation reduces the sensitivity o f the brightness
temperature to soil moisture, furthermore, the amount of reduction in sensitivity could be
approximated by Y = 1 - 1/L2(0), where L (0) represents the canopy loss factor.
Jackson and Schmugge (1991) noted a similar approach to quantify the effects of
vegetation by relating t, the optical depth (Kirdiashev, 1979), to influencing variables
which included vegetation dielectric properties, canopy shape and/or structure as well as
polarization and wavelength. Theis et al. (1984) showed the importance of compensating
for vegetation when remotely estimating surface moisture by comparing an overall R2 of
0.09 for a vegetated surface to an R2 of 0.71 for bare fields. These researchers were able
to significantly improve estimates of near surface soil moisture (R2= 0.09 to R2= 0.76) by
using a Perpendicular Vegetation Index (PVI - calculated as the perpendicular distance of
measured values of reflectance from the bare soil reflectance on a plot of Landsat band 5
(0.6-0.7 pin) versus band 7 (0.8-1.1 pm)) which has also been used as both a crop
classifier as well as a biomass indicator.
Barton (1978) found strong relationships between soil moisture and emissivity for
bare fields, however, the same relationships were not evident for vegetated sites even
when a 1.4 GHz airborne radiometer was employed. Newton and Rouse (1980) noted
similar results for a 2.8 cm wavelength but found that a dense canopy o f sorghum of
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
height 125 cm did not significantly impede the microwave emission at L-band for
incidence angles out to about thirty five degrees. This observation may be largely a result
o f the relatively steep incidence angles, thereby increasing the amount o f penetration
through the vegetation cover. These results are also consistent with those reported by
Ulaby etal. (1983).
Using aircraft experiments over the Konza Prairie, Schmugge et al. (1988) noted
that areas having a significant buildup o f dead vegetation in the form o f a thatch layer
completely masked the emission o f the soil itself at L-band. However, under a series of
controlled condition experiments, Jackson and O’Neill (1991) found that for truck
mounted L-and C-band radiometers, the presence of wheat and soybean residue caused
negligible attenuation of the background soil emission. The researchers also noted that as
the bulk of the crop residue began to decay, a decayed organic matter layer formed. They
concluded that this layer may serve as a highly emissive layer above the soil and, as such,
completely mask soil emissions as concluded by Schmugge (1988).
2.4.3 Surface Roughness Effects
Research has documented the effects of surface roughness and soil structure on the
microwave emission from soil (Choudhury et al., 1979; Newton and Rouse, 1980; Wang
and Choudhury, 1981; Wang et al., 1983; Wang, 1983; Coppo et al., 1991). It was
concluded from these studies that the general effect of increasing surface roughness was
an increase in the thermal emission o f the soil and a corresponding reduction in the slope
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
o f the regression between the normalized temperature brightness and the volumetric water
content. Wang (1983) used a slope reduction factor, (J 3 ), defined as a ratio of an observed
regression slope for a semi-rough field between Tub and volumetric water content and a
corresponding slope calculated from a perfectly smooth field, as a measure of the surface
roughness effect. In general, Wang (1983) concluded that the slope reduction decreased
with an increase in surface roughness, however, compared with the effects o f vegetation
on microwave brightness temperature the dependence o f f i on frequency due to surface
roughness was relatively weak. Newton and Rouse (1980) showed quantitatively that the
effect of roughening the surface of an agricultural field was quite pronounced especially at
the shallower incidence viewing angles (50°). In an electromagnetic sense, a surface can
be considered smooth if the roughness height (h) does not exceed the wavelength o f
observation. The generally accepted criteria for determining surface roughness or
smoothness in the microwave region is the Rayleigh criterion. Modification o f the
Rayleigh criterion states that a surface is “smooth”, if
h < — ^—
25 sin 6
and “rough”, if
h >
^
4.4 sin 0
where h is the vertical relief of a surface, X is the wavelength, and 0 is the incidence angle
(Sabins, 1978). This would suggest that most surface conditions would appear relatively
rougher for a 2.8 cm wavelength than for a 21.4 cm wavelength.
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.5 Passive Microwave and Earth Surface Processes
Ground-based, airborne, and satellite radiometers have been providing information
regarding the polarization characteristics of various surfaces since the launch o f SKYLAB
in 1972. Since then a number of orbiting satellites have been recording the radiation
emitted from Earth and atmosphere at selected frequencies between 1.4 and 85 GHz.
Measurements using different frequencies have enabled the identification and monitoring
of snow cover (Grody, 1991), sea ice (Campbell et al., 1987; Steffen and Schweiger,
1990; Parkinson, 1992) and precipitation over land as well as oceans (Heymsfield and
Fulton, 1992; Alliss et al., 1992; Grody, 1991;
et al., 1986).
Measurements using
different polarizations have been further used to infer wind speed, amount o f precipitation
(Barrett et al., 1988) as well as sea ice boundaries over oceans (Bailey et al., 1986;
Spencer, 1986). A number of surface retrieval algorithms (Cavalieri et al., 1984; Ferraro
et al., 1986; Hollinger et al., 1987; Neale et al., 1990; McFarland et al., 1990) have also
been developed making use of polarization difference measures over both land and oceans.
Such retrieval algorithms have served as appropriate indicators of soil moisture and
vegetation cover (Wilke and McFarland, 1986; Choudhury et al., 1987). Ferraro et al.
(1986) determined surface and precipitation features over North America using the
Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR) 18 and 37 GHz
vertical polarization channels. A classification algorithm was developed which was able to
separate six surface parameters (dry snow, new sea ice, dry land, flooded/wet land, open
water and rain) through “clustering” of 18 and 37 GHz average and difference brightness
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
temperature observations.
Neale et al. (1990) also developed land-surface-type
classification schemes which were based on statistical analysis o f SSM/I brightness
temperature combinations from several surfaces.
Neale et al. (1990) used different frequency combinations to establish land-surfacetype classification rules and corresponding brightness temperature combination threshold
values.
These threshold values were then used in determining several surface types,
including dense vegetation, rangeland and agricultural soils, deserts, snow precipitation
and surface moisture.
Similarly McFarland et al. (1990) used SSM/I microwave
brightness temperatures to derive land surface temperatures over the Central Plains of
the United States. Using the average polarization difference (85 V and 37 V) and the
channel difference (85 V - 37 V) measurements were correlated with minimum air
temperature which were assumed to be representative of the temperatures o f the
surrounding emitting surfaces at the time o f the satellite overpass.
McFarland et al.
(1990) found that regression models developed from SSM/I temperature brightness values
could be used to retrieve land surface temperatures for a variety of different surface types
consisting o f cropped land as well as moist and dry soil surfaces.
Bailey et al. (1986) and Spencer (1986) incorporated the idea that the ocean
surface has a relatively low emissivity (approximately 0.5) and rain clouds have a high
emissivity (approximately 0.9) to successfully determine rain rates over open oceans.
Spencer (1986) observed a significant correlation (r = 0.90) between rain rates derived
from ground radar measurements and SMMR 37 GHz derived rain rates for five cases of
convection over Mexico.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Passive microwave data recorded from satellite platforms have also been used in
monitoring sea ice concentrations throughout polar regions for almost two decades.
Success in sea ice concentration monitoring has resulted largely from the sharp microwave
emission contrast between sea ice and open water. In addition, the difference in emissivity
o f first year ice versus old ice at different microwave frequencies has provided a means of
separating ice types (Steffen and Schweiger, 1990). Parkinson (1992) and Drinkwater et
al. (1991) both used algorithms based on the SMMR 18 GHz polarization (H and Vpolarization) and the 37 and 18 GHz (V-polarizadon) spectral gradient ratios to establish
sea ice concentrations as well as spatial patterns of increase and decrease in length of sea
ice seasons in northern polar regions. In both cases, the retrieval algorithms performed
well in estimating both type and concentration o f ice.
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.6 Microwave Sensing of Earth Surface Processes from Satellites
2.6.1 Soil Moisture
Although the majority o f the relationships established by low altitude studies have
provided good correlations, they have not been able to address the problems associated
with monitoring the same variables from satellite sensors. These include, among others,
scene heterogeneity, a marked increase in pixel size as well as limitations regarding the
availability o f optimal frequencies.
Jackson (1993) noted that the amount of Radio
Frequency Interference (RFT) increased for wavelengths greater than 21 cm (1.4 GHz) and
therefore limited the specific regions that could be used. This becomes more of a problem
at frequencies less than 1.4 G H z.
With this in mind, Earth surface process studies from satellites have been fewer in
numbers for a number of reasons, including spatial resolution, number of available
satellites, and the problem of interpreting the heterogeneity within a large ground
resolution cell. As a result, correlations between ground variables and satellite sensors
have at times been less encouraging and somewhat daunting. Nonetheless studies have
been conducted in an attempt to extrapolate the conclusions obtained from low altitude
airborne studies to those of satellite platforms (Wilke and McFarland, 1986; Choudhury et
al., 1987; Owe et al., 1988; Choudhury and Golus, 1988; Neale et al., 1990; Kerr and
Njoku, 1993). A number of studies have explored the use of an Antecedent Precipitation
Index (API), a measure widely used in hydrology, to infer near surface soil moisture over
particularly large areas (40-100 km) (Linsley and Franzini, 1979).
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Studies by Wang (1985), Choudhury and Golus (1980), Wilke and McFarland
(1986), Choudhury et aL (1987), Owe et aL (1988) and Teng et al. (1993) have
demonstrated the usefulness o f incorporating a modelled estimate o f soil moisture over
large diverse geographical areas as an-equivalent measure o f near surface soil moisture.
Wang (1985) obtained data over a four month period in 1973 from the SKYLAB 1.4 GHz
radiometer for two regions in Texas, one exhibiting arid conditions with minimal
vegetation and the other more cultivated with a high degree o f vegetation cover. Wang
(1985) also used the API as an indicator of soil water content and obtained significant
negative correlations, r2 = 0.75 and r2 = 0.59 between Tb and API for the two regions
respectively. Owe et al. (1992) analyzed microwave data (6.6 GHz) from the Scanning
Multichannel Microwave Radiometer (SMMR) on board the Nimbus 7 satellite over a 150
km area of south central Botswana. A linear regression provided an r2 o f 0.45 between
pixel average surface soil moisture and the above canopy emissivity, however, upon using
a vegetation corrected emissivity value, the explained variance o f soil moisture from
satellite measurements was 71%. Choudhury and Golus (1988) and Wilke and McFarland
(1986) tested the applicability o f the API over the US southern Great Plains by correlating
it with SMMR 6.6 GHz frequency, horizontally polarized brightness temperature for a
number of ground resolution cells. They found that even though the study sites consisted
of a wide range o f vegetation densities, upon correlating the two variables, the API was
found to explain more than 70% (p = 0.01) of the observed temporal variability in TBhThe study also indicated that the API was significantly correlated with the Normalized
Vegetation Index calculated from the NOAA AVHRR visible and infrared wavelengths.
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Wilke and McFarland (1986) concluded that brightness temperatures measured by
microwave radiometers could provide an API time series which would facilitate
monitoring hydrological budgets on the scale of 25-50 km for the operation of large scale
crop models. Owe et al. (1988) and Teng et al. (1993) reported similar findings using
frequencies o f 18 GHz and 19 GHz respectively, however, these frequencies exhibit much
less of a response to soil moisture than the shorter frequencies as a result o f greater
atmospheric interference.
Owe et al. (1988) used measured microwave brightness temperatures together
with visible/near infrared wavelengths from the SMMR and the AVHRR sensors to
monitor soil moisture variability from spacebome platforms.
Both normalized TB and
ND VI were used in conjunction and were found to account for approximately 70% o f the
observed variability in soil moisture. The remaining variability was thought to be related
to additional surface parameters, such as surface roughness, that were not incorporated by
the model.
Teng et al. (1993) incorporated four years (1987-1990) o f SSM/I data at 19 and
37 GHz together with an API calculated from weather data in a study of the US com and
wheat belts.
The study area consisted of a more humid climate, a denser natural
vegetation cover, and different mix of agricultural crops than described by Owe et al.
(1988). The Microwave Polarization Difference Index (MPDI), an alternate indicator of
vegetation cover, based on the SSM/I’s 37 GHz channels (calculated as the difference
between the vertical and horizontal brightness temperatures) was used to derive vegetation
information. Correlations between API and Tb at 19 GHz were found to be, for the most
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
part, geographically dependent with arid areas providing stronger correlations (r = -0.60
to -0.85) and the more humid, densely vegetated areas exhibiting weaker correlations (r =
-0.30 t o -0.70). The MPDI was also used to estimate API from Tb. Estimated API values
were correlated with calculated API values and again correlations were temporally and
geographically dependent.
Heymsfield and Fulton (1992) also noticed large spatial
temperature brightness variations due to both vegetation cover and soil moisture. While
analyzing 19 GHz, horizontally polarized SSM/I data over Oklahoma in the fall o f 1987
they noticed pronounced SSM/I brightness temperature depressions in the absence of
reported precipitation. Further analysis showed that a cold anomaly was situated over a
region of bare fields which had recently received a significant rain event.
The low
brightness temperatures associated with the area were analyzed over time (15-20 days)
and found to approach values found prior to the rain event. These researchers noted that
in the absence o f vegetation, the 19 GHz horizontal polarization channel o f the SSM/I
sensor could help determine processes which deplete root zone soil moisture, notably
evaporation, transpiration and dry down. However, over areas o f denser vegetation, the
microwave temperatures were not especially well correlated with rain events and
ultimately soil moisture. This again suggests the importance o f normalizing or accounting
for the effects o f vegetation cover on microwave emissions for the purposes o f soil
moisture monitoring.
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.6.2 Vegetation Studies
While attempting to account for the effects o f vegetation cover on determining soil
moisture from satellite platforms, a number o f studies have investigated the use of
microwave indices as a surrogate measure of vegetation biomass. The 37 GHz horizontal
and vertically polarized channels of both the SMMR and SSM/I have been used for
monitoring and studying the synergy between passive microwave radiation and the NOAA
AVHRR sensor measurements (Choudhury and Tucker, 1987; Townshend et al., 1989;
Van de Griend and Owe, 1993; Kerr and Njoku, 1993; Smith and Choudhury, 1993).
Choudhury and Tucker (1987) computed the Microwave Polarization Difference
Index from two years o f SMMR data over fourteen globally distributed locations. For the
same locations they also calculated the Normalized Difference Vegetation Index (NDVI).
Correlation of the two datasets showed an extremely high inverse correlation (r = -0.97)
indicating the complementary o f the two datasets for monitoring global vegetation change.
Choudhury (1989) plotted the temporal evolution o f the MPDI for three different
areas; I) the Sahara, 2) the Sahel region, and 3) Tropical Savanna. Time series plots of
these areas found that the greater the vegetation density, the smaller the polarization
difference, however, for all areas the time series showed a remarkable sinusoidal pattern
corresponding to the maturing and senescence of the vegetation.
Furthermore, the
polarization difference decreased with increasing surface roughness and increased with
increasing soil wetness with highest values found over open water.
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Smith and Choudhury (1993) also used the MPDI to determine interannual
anomalies over Africa. SMMR 37 GHz measurements were obtained over a five year
period and their temporal evolution plotted. The researchers concluded that the anomalies
found within the MPDI values were related to year to year fluctuations in vegetation cover
and could be the possible controlling factor in the net radiation balance at the top o f the
atmosphere over North Africa.
Recently a number o f studies have sought to understand the influence o f the
atmosphere on the microwave brightness temperature as well as the polarization difference
which has been used with varying degrees of success as a measure vegetation biomass.
Kerr and Njoku (1993) noted that spectral signatures previously ascribed to terrestrial
vegetation changes, or unexplained anomalies, may be comprised of a number o f
atmospheric contributions and not so much to surface variability.
Furthermore, the
researchers empirically demonstrated through the use of a radiative transfer model that
relating temporal changes of the polarization index in terms of surface variability without
incorporating atmospheric variables such as the Integrated Liquid Water Content (ILWC)
(to which the 37 GHz signal is directly related) will be misleading. Kerr and Njoku (1993)
concluded that single frequency (37 GHz) measurements made by a satellite sensor is
unlikely to provide accurate estimates of surface variability unless ancillary measurements
of atmospheric moisture and canopy characteristics are available. Furthermore, at 37
GHz, that part of the signal related to the ground variability may be less than the part o f
the signal linked with the atmosphere.
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.7 Advanced Very High Resolution Radiometer (AVHRR)
Since the launch o f the NOAA-6 satellite in June 1979, synoptic global coverage
of the earth has been acquired on a near real time temporal frequency by the AVHRR
sensor. To this end, Ehrlich et al. (1994) stated that the global change community has
identified the AVHRR 1 km dataset as an important database for land-cover mapping and
land processes modelling at both continental and global scales. These datasets have been
used exhaustively for a number of different research applications. Ehrlich et al. (1994)
provided a brief summary o f a few of these research applications, notably land surface
temperature mapping (Collier et al., 1989), tropical forest monitoring (Smith and
Choudhury, 1990; Malingreau et al., 1989), and fire risk assessment (Lopez et al., 1991).
By far the most exhaustive use of AVHRR data has been in the monitoring o f seasonal
variation of vegetation activity and land cover mapping (Tucker and Gatlin, 1984; Tucker
et al., 1985; Goward et al., 1985; Townshend et al., 1987; Loveland et al., 1991;
Malingreau and Belward, 1992: Andres et al., 1994).
The utility of the AVHRR sensor is in its sensitivity to both red and near-infrared
parts of the spectrum. The value of these bands can be found in the sensitivity o f the red
band to absorption by chlorophyll and other plant pigments and hence to photosynthetic
activity, and the sensitivity o f the near-infrared to the mesophyll structure o f the green
leaves. Townshend et al. (1987) noted that ratios of these two bands have often been
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
related to variables such as percent vegetation cover, leaf area index (LAI) and amounts o f
green biomass present.
The Normalized Difference Vegetation Index (NDVT), which is expressed as:
NDVI = [Ch2 - Chi] / [Ch2 + Chi]
is therefore a measure of the greenness o f the vegetation and provides an indication of the
amount o f biomass present on the ground.
2.8 Summary o f Literature
Studies to date have shown the utility of radiometers for soil moisture estimation
and vegetation monitoring from ground based, airborne and spacebome platforms. A
comparison o f remote sensing approaches for soil moisture sensing is presented in table 3.
Results from past research have suggested that the lower microwave frequencies are better
for the purposes o f soil moisture estimation and the mid-frequencies, i.e. 37 GHz, can be
adequately used for large scale vegetation monitoring. The majority o f the studies have
found a significant relationship between microwave temperature brightness (T b) and
ground variables such as soil moisture and vegetation cover. Although passive microwave
research has focused on establishing algorithms for the development of models, these
studies have been conducted under controlled,
somewhat
optimal conditions.
Furthermore, passive microwave studies involving spacebome platforms have been much
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.
Comparison o f remote sensing approaches for soil moisture sensing.
(Adapted from Schmugge et al., 1981)
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SENSOR
ADVANTAGES
DISADVANTAGES
NOISE SOURCE
Thermal IR
10 pm - 12pm
Relatively high resolution
1 km
Large swath
Cloud cover, limits
frequency o f coverage
Local meteorological
conditions
Partial vegetation cover
Surface topography
Basic physics well
understood
Passive
Microwave
Independence o f
atmosphere
Moderate vegetation
penetration
Poor spatial resolution
10 - IS km ar best
Interference from man-made
radiation sources limits
operating frequencies
Vegetative cover
Soil temperature
Active
Microwave
Independence of
atmosphere
High resolution possible
Limited swath width
calibration o f SAR
Surface roughness
Vegetative cover
less common.
The success of measuring global surface parameters from spacebome
passive microwave radiometers for agro-ecosystem monitoring have been somewhat less
encouraging.
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3
METHODOLOGY
3.1 Introduction
There were three distinct phases to this investigation. The first phase comprised
research activities concerned with establishing a method of scaling from relatively high
resolution imagery (30 m to 1 km) to lower resolution imagery (25 km) over large areas
(i.e. hundreds of kilometres).
Furthermore, the first phase was used to establish a
framework for future processing o f spacebome passive microwave data and determining
the capability of the SSM7I sensor to monitor ground phenomena, such as soil moisture,
under “non-ideal” conditions as previously reported (Neale et al., 1990; Choudhury and
Tucker, 1987). Availability of data and time limitations necessitated that the study be
performed over three years and certain parameters be ignored or minimized.
The second phase of the analysis was performed to study the annual temporal
evolution of temperature brightness signatures throughout a number o f ecoregions. This
was considered important since it would provide an understanding with respect to the
change in temperature brightness over the course o f a growing season. Three consecutive
years o f data were analyzed; 1988, determined to be a “dry” year, 1990, considered to be
representative of a “normal” year and, 1992, indicative o f a “wet” year.
The third phase constituted the application stage, in which satellite derived
measurements of specific ground variables such as soil moisture were compared with
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ground or modelled estimates o f the same variables.
Furthermore, measurements o f
similar ground parameters were compared from different satellite platforms.
Also, a number o f surface retrieval algorithms were tested (1) to examine the
relationship between the temperature brightness, soil moisture and vegetation cover, (2)
to determine the magnitude o f influence of vegetation on the temperature brightness, and
(3) to modify existing retrieval algorithms for soil moisture estimation to fit the conditions
encountered across the study area.
A technological approach was used here which emphasizes the application o f
human ability for the purpose o f achieving some desired goal which is particularly suited
to investigations using remote sensing techniques (Curran, 1987). The design
methodology for remote sensing investigations requires,
.... at a minimum: (1) clear definition of the problem, (2) evaluation of the
potential for addressing the problems with remote sensing techniques, (3)
identification o f the remote interpretation procedures to be employed and the
reference data needed, and (5) identification of the criteria by which the quality of
the information can be judged (Lillesand and Kieffer, 1979,30).
Although this study is primarily investigative in nature, the above outline of the design
methodology provides an excellent method for structuring the analysis. A procedural
break down of the contents o f each chapter is presented in figure 4.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.
Flow chart o f the investigative framework
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 1
1)
Scope o f Problem
Passive Microwave images over
agricultural targets provide timely,
global scale biophysical information
such as soil mositure and vegetation
vigor/density which may not easily be
otherwise interpreted/extracted.
Chapter 2
2)
Use o f Passive Microwave
Sensing as Input to an
Agrorcosystem Monitoring
Procedure
Investigate theory behind passive
micorwave sensing. Determine the
potential o f utilizing this technology in
light of the current body of literature.
Chapter 3
3)
Investigation of Available
Methods to Model Microwave
Enissions from Vegetation and
Soils
Determination of modelling
procedures and processing steps to
evaluate specific objectives.
Chapter 4
4)
Data Interpretation
Procedures and Reference Data
Needed
Criteria for Determining the
Quality of the Information
Graph and discuss modelling results
and perform sensitivity analysis.
Results evaluated in the context of
empirical data found in literature.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2 Location and Description of the Study Areas
Due to the nature o f this investigation, that being to establish a valid method o f
scaling as well as model soil moisture, two different study areas were chosen for this
research.
The area chosen for the purpose o f establishing a method of scaling and
monitoring temporal evolution of temperature brightness was a transect extending across
Ontario from Lake Erie (approximately 43° N, 82° W) in the south to the James Bay
Lowlands in the North (approximately 55° N, 83° W) (figure 5). This area was chosen
since it comprised a variety of different ecoregions (Ecological Stratification Working
Group, 1993), namely, Mixed Wood Plain in southern Ontario represented largely by
agricultural areas, Boreal/Shield comprised mostly o f forest and lake areas, and the
Hudson and James Regions consisting primarily of low lying marsh flats and wetland
areas. Furthermore, the scale of the ecoregion map was suitable for use with SSM/I
measurements since SSM/I pixel sizes are on the order o f tens o f kilometers. The area
chosen for soil moisture analysis was an agricultural area on the Canadian Prairies located
in south-central Manitoba (figure 6). This region provided a relatively flat, homogeneous
area and had been used in other experiments (Canada Centre for Remote Sensing, 1994)
and therefore a p r i o r i knowledge of the area and vegetation characteristics could be
incorporated.
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 5.
Location of the Ontario transect
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90°
V
aor
85*
■>z
X -T S i.w * * ;,
Ontario
mt(m' ^
s v & X ■>
Quebec
50=
cJ^L . Superior
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 6.
Location of the Manitoba study site
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Manitoba
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2.1 Mixed Wood Plain, Southern Ontario
Southern Ontario, as will be described here, represents the agriculturally developed
portion of the province, which, for the most part lies south o f the Canadian Shield and
includes the south-western peninsular area bounded by the Great Lakes (Chapman and
Putnum, 1973). Southern Ontario is an area of modest relief with respect to the rest o f
Canada. The lowest areas, found to the East along the Ottawa River, are barely 50 metres
above sea level, while the highest points, on the Blue Mountain south o f Collingwood
reach almost 600 metres above sea level (Chapman and Putnum, 1973).
Successive glaciation o f Southern Ontario scoured, pulverized and dislodged vast
amounts of debris.
Erosion of the softer Paleozoic limestone material together with
successive deposition throughout Southern Ontario has resulted in deep overburden in
contrast to the much harder rocks of the Canadian Shield (Chapman and Putnum, 1973).
The high limestone and clay content of the resulting depositional till is largely responsible
for the development o f some of the most highly productive soils in Canada. Except for a
few large urban centres along Lake Ontario, such as Toronto and the surrounding cities,
Southern Ontario is largely agriculturally based.
Approximately 19,000,000 acres o f
farmland are responsible for producing almost 95% of the agricultural wealth of the
province and one-quarter of that o f the whole country (Chapman and Putnum, 1973).
Climatically, southern Ontario represents a very diverse environment with average
lows reaching -10° C and highs o f approximately +22° C. The area is characterized by a
mean annual temperature of about +4.0° C (Environment Canada, 1982).
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Annual
precipitation for the southern Ontario region ranges from 600 mm in the North around
Huron County to over 800 mm around Lake Erie (Environment Canada, 1982).
3.2.2 Boreal/Shield
Topographically, the Shield is best described as a vast “saucer-like” structure rising
outward from Hudson Bay and the Foxe Basin. It extends northward to Baffin Island,
eastward to the coast o f Labrador and Newfoundland, southward to the St. Lawrence
River and the Great Lakes and westward to the prairies (Hunt, 1974). This physiographic
region is mostly flat with local relief generally less than 80 m (Environment Canada,
1990). Despite the relatively recent departure o f the glacier, the Shield has a welldeveloped system o f rivers with many lakes occupying basins eroded in the bedrock
(Hunt, 1974). The northern part of the Shield has a polar climate with the warmest
months barely reaching above freezing temperatures. The southern part o f the Shield has
a 3-month growing season, and is much wetter than the northern part with precipitation
averaging nearly 600 mm around Lake Superior (Environment Canada, 1990). As a result
of extreme temperatures in the northern part o f the Shield, growing conditions are too
severe for tree growth and is therefore best characterized as a treeless tundra. South of
the “tundra belt” lies the boreal woodland and coniferous forest which consists of spruce
and larch as well as extensive areas of bog (Hunt, 1974). The southernmost part o f the
Shield has mixed deciduous and coniferous forest comprised of spruce, pine and hemlock
mixed with maple, beech and other northern hardwoods (Hunt, 1974).
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Soils on the Canadian Shield are young, all having been formed during the
Holocene period from glacial, fluvial, lacustrine, colluvial and beach deposits originally
derived from crystalline rocks (Hunt, 1974). In the forested southern part o f the Shield,
soils on well-drained ground are mostly Podzolic.
However, near the Ottawa River and
Great lakes, Gray Brown Podzolic soils are dominant.
Podzolic soils are well to
imperfectly drained mineral soils which have formed under acid leaching conditions
(Agriculture Canada, 1987).
3.2.3 Hudson and James Regions
The James region comprises the majority of the area west o f James Bay and
Hudson Bay and is characterized by gently rolling to broken uplands formed on the
crystalline granitic rocks of the Canadian Shield. This region is further characterized by
numerous small lakes and wetlands occurring in bedrock depressions (Environment
Canada, 1990). Surface runoff tends to be poorly organized and is typified by numerous
lakes connected by short rapid streams and rivers. The highest elevations in this region
occur in the Lake Plateau, Kaniapiskau Plateau and Mistassini Hills, all having elevations
greater than 900 m (Environment Canada, 1990).
The Hudson Region is comprised mainly of the Hudson Bay Lowland.
The
Hudson Bay Lowland is a large coastal plain adjacent to the East o f Hudson and James
Bays. Wetlands occupy between 85 and 90 percent of its surface (Riley, 1982). Ideal
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conditions for wetland development in this area include a flat topography with a gentle
slope and impermeable substrates consisting of fine grained glaciomarine clays and silts
(Environment Canada, 1990). The coastal plain rises from sea level to approximately
150 m above sea level where it meets the Canadian Shield (Riley, 1982).
Climatically, the Hudson and James Bay Lowland regions are characterized by
cold winters with mean annual temperature between -20° C and -30° C and moderately
cool summers with mean temperatures ranging from
+15° to +17° C (Environment
Canada, 1982). Annual precipitation for the region ranges from approximately 400 mm
near Moosenee to 600 mm further south (Stanford, 1977).
3.2.4 Manitoba Study Area
The Manitoba Plain is a flat-lying area consisting o f glaciolacustrine clays and silts
of glacial Lake Agassiz. The average elevation of the Manitoba Plain is about 240 m
above sea level (Hunt, 1974). The glaciolacustrine clays overlie glacial till which, in turn,
overlies sedimentary rocks (Teller and Fenton, 1980). Drainage consists of abundant lakes
and large slow-flowing meandering rivers such as the Assiniboine, Red and Saskatchewan.
Open water covers almost 50 percent of the surface of the Manitoba Plain mostly as large
lakes namely Lakes Winnipeg, Winnipegosis and Manitoba (Environment Canada, 1990).
Fine-grained sediments common to this interlake area, aiong with a flat topography,
results in poorly drained soils and the development of wetlands (Environment Canada,
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1990). Soils of the Manitoba Plain are primarily o f the grassland type (Dark-brown to
black Chernozem) in the south grading to more o f a forested soil with more acidic parent
material (podzolic) further north (Stanford, 1977).
Almost all of the Manitoba Plain
receives between 400 and 600 mm o f precipitation annually. Mean annual temperatures
during the winter months range anywhere from -20° C to -30° C, with mean annual
summer temperatures as high as 25° C (Environment Canada, 1982).
3.3 Data Acquisition
3.3.1 SSM/I Sensor
The SSM/I is a seven channel, four frequency, linearly polarized, passive
microwave radiometric system.
The sensor measures atmospheric/surface brightness
temperatures at 19.35, 22.2, 37.0 and 85.5 GHz (Heymsfield and Fulton, 1992). Data
received from the sensor are processed by the Environmental Parameter Extraction (EPE)
algorithm in place at the Fleet Numerical Oceanography Centre (FNOC), Monterey,
California, and the Air Force Global Weather Centre (AFGWC), Omaha, Nebraska to
obtain near real time precipitation maps, sea ice morphology, marine surface wind speed,
columnar integrated liquid water and soil moisture percentage.
The SSM/I data processing algorithm is composed of two major modules. The
first module ingests and processes raw satellite data and produces geo-located (latitude
and longitude) brightness temperatures (Tb) for each pixel (Hollinger et al., 1987). The
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
geolocation process involves the use o f algorithms developed by Hughes Aircraft
Company together with the satellite ephemeris data. The satellite ephemeris refers to a
tabulation o f a satellite’s position, for example, latitude, longitude, and altitude, as a
function of a prescribed time interval (Hollinger et al., 1989). The temperature brightness
values are then processed through a second module, the environmental extraction
algorithm, to produce estimates o f ocean, land, and ice parameters (Hollinger et al., 1987).
Processed data is then made available to the general scientific and industrial communities
through the METSAT archival agreement between the National Oceanic and Atmospheric
Administration (NOAA) and the Department o f Defence (DOD).
3.3.1.1 SSM/I Scan Geometry
The ground resolution cell or pixel size for the SSM/I radiometer is a function of
the operating frequency. As the frequency increases, the pixel size decreases, however, as
the frequency increases there is increased atmospheric interference thereby rendering these
frequencies much less useful for terrestrial analysis (table 4).
Frequency
GHz
19.35
22.235
37.0
85.5
Polarization
H or V
H&V
V
H&V
H&V
Resolution (km)
Along-track... ......Cross-track......
43
69
50
40
37
29
13
15
Spatial Sampling
(km)
25
25
25
12.5
Table 4. Spatial Resolution of SSM/I channels.(Adapted from Hollinger et al., 1987)
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The Instantaneous Field o f View (IFOV) o f the SSM/I sensor for each operating
frequency during a single scan over a particular area is shown in figure 7. Projection o f
the 3 dB beamwidths onto the Earth’s surface are represented with ellipses for each
frequency. The term beamwidth is used to characterize the width o f the antenna main lobe
in a given plane (Ulaby et al., 1981). The sub-satellite track o f the spacecraft moves along
in the y-direction at a velocity of 6.58 km/s which results in a separation distance between
successive A and B scans of 12.5 km along the cross-track which is almost equal to the
resolution o f the 85.5 GHz beams (Hollinger et al., 1987). Each sampling interval (Bscan) for the 85.5 GHz data takes 4.22 ms and equals the time for the beam to travel 12.5
km in the cross-track direction. Radiometric data measured at the remaining frequencies
(A-scan) are sampled every second scan with uniformly spaced samples each having an
8.44 ms interval thus providing spatial sampling every 25 km. This is the manner by which
the data becomes gridded to a 25 km cell size even though the IFOV is somewhat larger
for almost all frequencies. The satellite makes two passes per day, one at approximately
7:00 AM and the other at 7:00 PM with short periods every four to five days without
coverage due to orbit precession.
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 7.
Instantaneous Field o f View for the SSM7I operating frequencies
(Modified from Hollinger, 1989)
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
O
H
ccon
*r\
00
o\
2a
<
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3.2 AVHRR Data Processing System (GeoComp)
The GeoComp system was designed to routinely provide Canada-wide composite
images from raw AVHRR imagery in an automated and timely manner.
The primary
function o f the GeoComp system is to generate large cloud free composite images from
NOAA AVHRR imagery as well as provide geocoded, spatially corrected products
(Robertson et al., 1992). The GeoComp system is divided into two subsystems; 1) the
Image Correction Subsystem, and 2) the Composite Generation Subsystem.
The Image Correction Subsystem is responsible for removing the geometric and
radiometric effects from the raw AVHRR images.
GeoComp include spacecraft
Geometric errors corrected by
orbit and attitude, Earth rotation, Earth curvature,
panoramic distortion, as wellas first order terrain elevation.
The average absolute
geometric error for a 1.0 km resolution GeoComp single date scene product is 600 m. The
radiometric integrity of the AVHRR imagery is preserved by the GeoComp system since it
maintains the full 10-bit (0-1023) dynamic range of the AVHRR imagery throughout the
processing chain. GeoComp also employs absolute calibration methods which make use
of both pre-launch calibration as well as time dependent calibration coefficients for each
channel (Teillet et al., 1990).
The primary function of the Composite Generation
Subsystem is to combine the
precision geocoded images generated from the Image
Correction Subsystem into a single composite image o f a user specified area (Robertson et
al., 1992). To create a cloud free image, a series o f multi-temporal AVHRR scenes over a
particular area are loaded into the GeoComp system. Compositing is carried out using the
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
maximum Normalized Difference Vegetation Index (NDVI).
Here, each NDVT value is
examined individually on a pixel-by-pixel basis with the highest value being retained for
each pixel location (Holben, 1986). Accordingly, pixels for the remaining channels are
also selected thereby providing complete five channel geocoded cloud free imagery.
Although a method o f compositing is used to remove cloud, the algorithms are only
successful if cloud free images exist for a given area for each composite period. In the
event that the entire week was influenced by cloud cover, the resulting composite may
show some residual cloud effects. This effect is usually more prevalent during early spring
and late fall.
3.4 Data Preprocessing
A number of processing steps were required prior to further data analysis. Due to
orbit precession of the DMSP satellites together with possible sensor start problems,
complete coverage o f the study area was not always obtained for the exact dates of
interest. Furthermore, comparison of pixel brightness temperatures over time could not be
accurately performed as a result of satellite precession which is not corrected for in the
archived data, therefore data was not tied to a similar latitude and longitude. This meant
that SSM/I pixel centres were not consistently located thereby making absolute pixel
comparisons difficult. In order to overcome the problem o f incomplete coverage o f the
study area during the analysis, SSM/I imagery from two consecutive morning passes were
merged together to ensure a complete coverage o f the study area.
To observe the
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
temporal evolution o f SSM/I pixels, the April 1988 SSM/I imagery was arbitrarily chosen
as the base image to which all other dates of SSM/I imagery were merged onto. In this
process, the latitude and longitude values of successive SSM/I passes were resampled to
match those of the April image. Each successive SSM/I overpass was resampled and
mapped to the April image using an equal weighting resampling algorithm (GADS Version
3.0 User’s Manual, 1995). Resampling is the process o f sampling one or more input
pixels to create one output pixel (PCI, 1995). An equal weighting resampling algorithm
allows all data values within the radius o f interest to have the same amount of influence on
the average weighting o f the resulting grid point. Initially, the radius of influence was set
at 12.5 km which corresponded to the pixel size of the 85.5 GHz frequency, however, this
resulted in a number o f missing datapoints. Since the 19 and 37 GHz channels were of
interest, a radius of influence o f 25 km or one regridded SSM/I pixel was used. This
provided a final image with relatively few missing data points and further allowed for
observation of the temporal evolution o f temperature brightness for a particular SSM/I
pixel.
3.4.1 Identification of Suitable SSM/I Study Pixels
A transect across Ontario was identified for development of an appropriate scaling
methodology. The transect was established from Lake Erie in the south to Hudson Bay in
the north. This transect was identified for two reasons: firstly because it spanned a large
enough area for the analysis o f SSM/I imagery and secondly it covered a variety of
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
different and relatively homogeneous physiographic regions. Although the SSM/I records
data twice daily, only the morning data were analyzed here in order to minimize any
atmospheric and diurnal effects. The reason for this is that the atmosphere is considered
most stable during the morning when the ground surface is colder, on average, than the air
above and a temperature inversion occurs (Cole, 1970). Consideration o f the time of
image acquisition is important since Kerr and Njoku (1993) demonstrated that
atmospheric effects such as humidity and temperature could be significant when compared
to effects of surface variability, especially at 37 GHz.
Identification of SSM/I pixels involved a pseudo random stratification process
whereby SSM/I pixels were randomly identified. Each SSM/I brightness value was given a
number from 1 to n where n represented the total number of pixels for each scene. A total
of 100 numbers were randomly generated from the total number of pixels in each scene.
The random numbers were then used to successively identify SSM/I pixels which were
located in the respective ecoregions. A pixel was considered not useable if it was in close
proximity with, or bounded, large water bodies, major urban centres or simply were
located too close to the edge of the available imagery. It was recognized that within any
SSM/I pixel there was likely to be some heterogeneity such as small bodies o f water as
well as urban landscapes, however, at the scale of observation (tens of kilometers) it was
decided that the influence of such phenomena would be small and for the most part could
be neglected. Moreover, the adoption of complex mixing formula to account for this
heterogeneity is being developed but was beyond the scope of this study.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Initially four SSM/I pixels were identified for each ecoregion.
However, as a
result o f a variety o f farming and land uses practices within southern Ontario a rather
dissected landscape has developed. Accordingly, it was decided to increase the number of
pixels identified thereby providing a more representative signature for this specific
ecoregion.
Consideration o f the locational accuracy together with pixel selection criteria
made it difficult to identify more than ten pixels for southern Ontario due to the overall
size of this ecoregion.
The final number o f pixels sampled for each ecoregion are
presented in table 5.
Ecoregions
Number o f SSM/I pixels
Hudson/James Bay Lowlands
4
Boreal
6
10
Southern / Central Ontario
------------------------------------------------------
Table 5. Number o f pixels sampled for each ecoregion.
3.5 Merging AVHRR and SSM/I
Ten day composited Advanced Very High Resolution Radiometer (AVHRR)
imagery for May 21-31, 1992, May 21-31, 1993 and May 21-31, 1994 was acquired for
the Ontario study area. The images were purchased system-corrected (radiometrically and
spatially).
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
AVHRR imagery was chosen since it provided an excellent intermediary step
between high resolution imagery (Le. 30 m - Landsat Thematic Mapper) and the much
coarser resolution SSM7I imagery (25 km). Furthermore, AVHRR data were collected
daily over both study sites. Daily coverage o f each area for a period o f ten days ensured
that a cloud free composite image could be created. This was especially important for this
analysis since the area of interest covered large extents (hundreds o f kilometres) and
therefore it was unlikely that single-date cloud-free imagery would be available for the
entire area on any given date.
EASI/PACE image analysis software (PCI Inc., 1993, version 5.3) was used to
analyze the AVHRR imagery. Non systematic distortions in satellite imagery are a result
of variations in the sensor platform’s altitude and attitude (roll, pitch and yaw) (Jensen,
1986), and make the imagery non planimetric. In order to ensure that area and distance
measurements were accurate, the imagery was rectified to a map base.
consisted o f relating pixel co-ordinates to
This process
their respective latitude and longitude co­
ordinates on the ground. To do this the four comers of the image were registered using
decimal degrees and projected using a Lambert Conic Conformal projection.
Composited AVHRR imagery acquired in late September and early October of
1993 by Statistics Canada over the Manitoba study site was supplied for the soil moisture
analysis. Since Statistics Canada receive preprocessed GeoComp data some further
preprocessing needed to be completed in order to remove the effects o f residual clouds. A
simple algorithm was written for the NDVI channel which gave those pixels having a
digital number o f 13 or less, a value o f zero.
This value was determined to be the
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
threshold below which pixels in channel one were affected by cloud contamination
(Brown, 1994).
Calculation o f a Normalized Vegetation Index involved comparing channels one
and two.
This was performed using simple image Boolean operators found within
EASI/PACE.
Processing o f the latitude, longitude tagged SSM/I data was performed using a
specific software package, GADS (Geographic Analysis and Display System) (PhJD.
Associates, 1993, version 1.0). Using the GADS software a number of algorithms were
calculated based on current literature (Tucker and Choudhury, 1987; Neale et al., 1990;
Kerr and Njoku, 1993; Hollinger et al., 1991). The names and formulation o f each
respective algorithm used in this study are outlined in table 6.
Name
Algorithm
Microwave Polarization Difference Index
MPDI = V ( f ) - H (/)
Average Polarization Difference
POLDIF = [(V19 + V37)/2] - [(H19 + H37)/2]
Normalized Brightness Temperature
T n b = H 1 9 /V 3 7
Polarization Ratio
PR = \ V ( f ) - H(/)] / [V(/) + H (/)
Where (f ) represents 19 and 37 GHz
Table 6. Name and formulation of microwave algorithms
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to compare the two sets o f imagery, a method of appropriately scaling
from AVHRR to SSM/I needed to be developed. Although this method o f scaling has
been used here for AVHRR and SSM/L, it could conceivably be used to scale between
imagery obtained from other sources. First, the geolocational accuracy o f each SSM/I
pixel was determined. Initial estimates of the locational accuracy for the SSM/I sensor
were on the order o f 20 to 30 km, however, approximately half of this error was attributed
to the use of a poor estimate of the satellite ephemeris (Hollinger et al., 1989).
Corrections to the processing software reduced the geolocation error to less than
the DMSP SSM/I accuracy specifications o f +/- 7 km, however, Wentz (1994) noted that
such an accuracy was optimistic and suggested using +/- 7 km as the amount o f error
associated with locating each pixel.
Output from the GeoComp software provides a sub pixel locational accuracy of
approximately 600 m. In order to compare measurements made by the two sensors, it was
important to be as certain as possible that both sensors were recording data from the same
approximate areas. Both the SSM/I and AVHRR locational accuracy’s along with the
SSM/I pixel size were incorporated into the scaling methodology.
The size o f the
resulting ground cell that could be contributing to the SSM/I brightness temperature
measurements is shown in figure 8. By adding both the SSM/I and AVHRR locational
accuracy to the initial SSM/I pixel resolution, a pixel size of 40 km x 40 km resulted.
Therefore an area o f
approximately 40 x 40 pixels on the AVHHR imagery had to be
exactly located to correspond with each SSM/I pixel.
Since two different software
packages were used to analyze the data sets, it was difficult to precisely locate the pixel
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 8.
Registration of SSM/I and AVHRR datasets
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[w a i V w w w m w w a v !;
<■■•■
xi^w^ x^, >v
* '
-■*.
x#
/V fT
v
*
v
'; ■&?' W-3
*
jj> * ' V
~
‘
V,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
centre on the AVHRR imagery. Coincident pixel centres were located as near as possible
and the difference between SSM/I and AVHRR pixel centre values were recorded. Both
the AVHRR and SSM/I datasets were referenced by latitude and longitude. In order to
ensure that the locational accuracy was not greater than 500 m each latitude and longitude
values were converted to a Universal Transverse Mercator (UTM) projection.
Since the
UTM projection is given in metres the difference between the two UTM co-ordinates
provided a measure o f the accuracy o f co-location between SSM/I and AVHRR for each
pixel.
For all three years, the locational error o f each pixel centre was less than 600 m
(+/- 0.5 AVHRR pixel) in any direction (Appendix). Using the respective pixel centres as
a starting reference, areas of forty kilometers by forty kilometers were delineated on the
AVHRR image and average reflectance values extracted for each newly defined pixel.
Upon completion of this process it became evident that some o f the previously identified
SSM/I pixels could no longer be used since they had become too close to water bodies
upon consideration of the respective locational accuracies.
Therefore the process of
selecting SSM/I had to be performed again based on criteria established earlier.
3.6 Modelled Estimate of Soil Moisture
In-situ measurements of soil moisture, while available, do not provide synoptic
information on a large scale. In an attempt to derive large scale soil moisture maps
empirical relations have been established in the form of Antecedent Precipitation Indices
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Saxton and Lenz, 1967). API measures are based upon the fact that soil moisture
depletion can be expressed as an exponential decay function o f the moisture input to the
profile on any given day. The models also take into consideration the amount o f
precipitation for each day and a running measure of soil wetness is maintained. Large scale
estimates of soil moisture were required for this analysis in order to determine possible
relationships between soil wetness and SSM/I temperature brightness.
The Winnipeg
Climate Centre, Environment Canada, provides a set of weekly Agrometeorological
Bulletins for the Canadian prairies.
Included in these bulletins are estimates of modelled
weekly actual evapotranspiration, moisture stress (i.e., available water minus the crop
demand) and a modelled estimate o f available soil moisture values. Here, a modelled
estimate of soil moisture is given as the available soil moisture reserve in millimetres along
with “percent of normal” calculated for the period 1951-1980 (Raddatz, 1989). The soil
moisture model (figure 9) implemented at the Winnipeg Climate Centre is similar to the
Versatile Soil Moisture Budget (Baier et al., 1979) which allows the calculation o f
available soil water to be performed for climatological sites which only observe maximum
temperature, minimum temperature and daily precipitation. There are approximately one
hundred and fifty sites across the prairies which either collect the data in a manual or
automated manner and forward it to the Winnnipeg Climate Centre. Approximately forty
five weather stations distributed throughout Manitoba were used to determine modelled
estimates of soil moisture. The model regularly provides measures of soil moisture (in
millimetres) per 120 cm of soil depth but was recently updated to provide measures of soil
moisture (in millimetres) for the 0-10 cm depth. Prior to operational implementation of
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 9.
Flow chart of soil moisture modelling procedure
(Modified from Raddatz, 1989)
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Initial available
soil moisture
Precipitation
Potential
Evapotranspiration
Crop moisture
consumption
Potential plant
water use
Precipitation >
Demand
Demand met by:
- precipitation
Moisture stress
Plant water-use
Demand > Precipitation
Demand met by:
- precipitation
- soil moisture
Available soil
moisture
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the 0-10 cm moisture model, data from ten weather stations were used to test the
modelled output against ground measured values. This initial test data was further made
available by Environment Canada for the purpose o f this study.
In order to convert the
depth of water in the top 10 and 120 cm soil column to a percent moisture value, an
average bulk density for each depth needed to be determined. Using an average bulk
density of 1.1 g/cm3 and 1.3 g/cm3 for the 10 and 120 cm depth respectively, the weight of
soil for each depth was calculated (156 g for 120 cm). In order to calculate percent soil
moisture the specific gravity of water (1.0 g/cm3) was multiplied by the modelled estimate
of soil water (depth in cm) and percent soil moisture for each depth calculated.
Soil moisture retrieval algorithms developed as part of the SSM/I
Calibration/Validation report (Hollinger, 1989) provide estimates of near surface soil
moisture (in percent) for three different vegetation densities, low, medium, and mediumhigh.
The vegetation density can be approximated by the amount of polarization
difference between 19 and 37 GHz vertical channels and the 19 and 37 GHz horizontal
channels (see Hollinger et al., 1989) as well as from the NDVI values.
A corresponding
SSM/I pixel was located for each weather station. Again, SSM/I pixel centres could not
be precisely anchored to the exact co-ordinates o f each weather station, however, in all
cases the chosen SSM/I pixel included the respective weather station.
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.7 Statistical Analysis
3.7.1 Characterization of Reflectance Data
All data were entered into an EXCEL spreadsheet (Microsoft Corporation, 1994,
Version 4.0). Data included AVHRR reflectance values for channels I to 5 as well as a
number o f different vegetation indices and temperature brightness values for the respective
SSM/I channels. Algorithm processing for the SSM/I data was performed using the
spreadsheet associated with the GADS software and later exported to an EXCEL
spreadsheet. The 85 GHz channel o f the SSM/I sensor was excluded due to significant
atmospheric interference (Hollinger et al., 1991). Soil moisture information (0-10 cm and
0-120 cm) derived from modelled estimates were also entered for the 42 weather
reporting stations throughout Manitoba (Environment Canada, 1995).
All data was subsequently imported into the STATISTIC A statistical software
package (StatSoft Inc., 1994, Version 4.0) for further analysis. Summary statistics were
calculated for each individual SSM/I frequency/polarization combination as well as each
AVHRR channel. Correlation matrices were further generated to analyze the relationship
between the SSM/I and AVHRR sensors. Correlation coefficients were calculated for all
possible combinations of SSM/I and AVHRR channels. The correlation coefficient is a
ratio o f the extent to which two channels vary together to the overall variability in the two
channels (Matthews, 1981). It was hypothesized that soil moisture would be significantly
related to a microwave frequency and/or combination of microwave frequencies used by
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the SSM/I sensor.
Based upon the results of the simple linear regression, models for
estimating soil moisture using the SSM/I sensor were derived.
3.7.2 Determining Earth Relationships from Satellites
Remote sensing research seeks to understand the dynamics of earth surface
parameters by measuring the surface radiation and comparing them with actual surface
values. By doing so, useful techniques are developed which utilize reflectance information
to predict, as well as monitor changes in, surface parameters such as vegetation density or
near surface soil moisture.
The results o f several field experiments have generally shown a linear relationship
between normalized temperature brightness and soil moisture expressed either on a
gravimetric or percent of field capacity basis (Schmugge, 1980; Wang et al., 1983).
Therefore, it was hypothesized for this study that the best correlation between soil
moisture and SSM/I variables would be a simple linear equation.
brightness temperatures are influenced by the physical temperature
Since microwave
of the emitting
surface, the apparent emissivity should be a more accurate indicator o f surface moisture
since it removes some variability in the observation due to changes in the surface physical
temperature.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4
RESULTS AND DISCUSSION OF THE MODELLING EXPERIMENTS
4.1 Introduction
The results o f the analyses are presented, discussed, and evaluated in the context
of the current literature in sections 4.2, 4.3, and 4.4 o f this chapter. The three main
objectives of the research, 1) establishing a valid scaling methodology, 2) monitoring the
temporal evolution o f temperature brightness, and, 3) mapping soil moisture using satellite
data are dealt with as separate entities initially, and then brought together in section 4.5.
In each section the results are presented, discussed and evaluated with an examination of
available literature.
The results o f the study were evaluated both quantitatively and qualitatively using
graphs, and tables, and imagery, as methods of data presentation and examination.
4.2 Scaling Experiment
4.2.1 SSM/I - AVHRR Scaling Results
The results presented in this section depict the measured microwave brightness
temperature
(T
b)
observed by the SSM/I sensor over a variety o f different surface types
against measured radiance values observed by the AVHRR sensor for the same surface
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
types. Summary statistics for each microwave channel and channel combinations together
with AVHRR thermal (3, 4, and 5) and NDVT channels are presented in tables 7 through
9.
The digital numbers listed in tables 7, 8 and 9 for the thermal channels on the
AVHRR sensor represent digital counts on a 10-bit scale (0 to 1023), however, the NDVI
measurement represents a number from 0 to 1. Conversely, microwave temperatures are
recorded as apparent brightness temperatures with no constraints on the intensity o f the
number.
Channels and combinations
SSM/I
V19
H19
V37
H37
PR37
PR19
MPDI19
MPDI37
POLDEF
AVHRR
Chi
Ch2
Ch3
Ch4
Ch5
NDVI
N
Mean
Minimum
Maximum
Std. Dev.
20
20
20
20
20
20
20
20
20
256.488
244.788
248.905
241.504
0.0153
0.0236
11.7
7.401
9.551
243.4
222.1
215.8
203.6
0.0043
0.0059
3.1
2.1
2.75
264.75
260.4
265.15
258.67
0.0291
0.0458
21.3
12.2
16.75
7.0378
10.931
16.843
17.526
0.007
0.01
5.057
2.89
3.80
20
20
20
20
20
20
97.2243
172.668
670.421
425.754
379.857
0.274
79.736
118.983
446.351
353.701
305.85
0.183
124.38
213.164
964.516
461.556
417.396
0.392
12.6
31.78
120.944
34.734
36.027
0.0697
Table 7. Summary statistics for the 1992 SSM/I (April 2) and AVHRR (May 21 -31)
channel and channel combinations for the Ontario study site.
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Channels and combinations
V19
H19
V37
H37
PR37
PR19
MPDI19
MPDD7
POLDIF
AVHRR
Chi
Ch2
Ch3
Ch4
Ch5
NDVI
N
Mean
Minimum
Maximum
20
20
20
20
20
20
20
20
20
260.857
250.922
260.111
253.132
0.0136
0.0196
9.937
6.98
8.458
250.7
234.31
242.00
239.10
0.004
0.009
4.71
1.90
3.35
271.04
266.33
270.24
267.32
0.029
0.039
19.21
14.57
16.89
50.53
8.416
6.908
7.261
0.007
0.009
4.516
3.270
3.717
20
20
20
20
20
20
160.464
207.789
642.524
513.043
465.512
0.135
88.606
120.664
494.790
404.628
245.449
0.033
249.454
302.943
776.176
643.746
628.561
0.251
37.087
44.423
97.990
61.566
82.559
0.054
Std. Dev.
Table 8. Summary statistics for 1993 SSM/I (May 26) and AVHRR (May 2 1 -3 1 ) channel
and channel combinations for the Ontario study site.
Correlations between microwave measurements and optical/thermal measurements
for the three dates of interest are presented in table 10.
Since the various channel
combinations showed the most consistency over time they will be discussed in more detail.
Tucker and Choudhury (1987) and more recently Teng et al. (1995) have presented an
inverse relationship between NDVI and MPDI37, despite the difference in study areas.
Tucker and Choudhury (1987) studied a variety of locations around the world ranging
from desert to dense rainforest where as Teng et al. (1995) observed the relationship for a
densely cropped area in the US com belt. However, results obtained from this study do
not entirely substantiate this relationship perhaps for reasons which will be discussed in
section 4.2.2.
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Channels and combinations
SSM /I
V19
H19
V37
H37
PR37
PR19
MPDI19
MPDI37
POLDIF
AVHRR
Chi
Ch2
Ch3
Ch4
NDVI
N
Mean
Minimum
Maximum
Std. Dev.
20
20
20
20
20
20
20
20
20
267.834
258.856
268.558
261.836
0.013
0.017
9.206
6.873
8.040
260.79
247.67
261.96
250.33
0.004
0.004
3.92
2.3
3.11
276.61
269.85
275.45
269.35
0.024
0.031
15.56
12.25
13.55
4.35
6.07
4.22
4.83
0.01
0.01
3.36
2.56
2.76
20
20
20
20
20
76.448
125.53
734.208
453.152
0.238
67.552
97.016
627.575
409.737
0.167
89.764
152.081
805.571
499.804
0.340
6.69
19.89
40.16
18.92
0.056
Table 9. Summary statistics for 1994 SSM/I (May 28) and AVHRR (May 21-31) channel
and channel combinations for the Ontario study site
A further interesting result o f the analysis is the consistent relation between the
AVHRR thermal channels (3, 4 and 5) and the SSM/I microwave channel combinations
(M PD I19 & 37, PR 19 & 37, POLDIF as well as the Normalized Brightness temperature,
T nb).
For all three years a positive relationship between the Normalized Brightness
temperature and the thermal channels (3 and 4) can be seen. This relationship suggests
that as the thermal temperature o f the ground surface decreases, as a result of an increase
in near surface soil moisture, the Normalized Brightness temperature also decreases.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 10.
Correlations between AVHRR and SSM/I channels and channel
combinations for the Ontario transect
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Channel 3
Channel 4
Channel 5
V19
0.03
-0.07
-0.06
H19
0.16
0.02
0.03
V37
-0.18
-0.22
-0.21
H37
-0.11
-0,18
-0.17
H19/V37
0.47*
0.39 **
0.39
PR37
-0.29
-0.13
-0.13
PR19
-0.28
-0.13
-0.14
-0.14
-0.15
NDVI
0.70*
0.68 *
0.66 *
.68*
-0.37
-0.41 **
-0.53 *
-0.50 *
-0.31
V19
0.32
0.16
V37
0.1
H37
0.32
-0.4
0.46*
0.03
0.06
0.37
0.01
H19/V37
0.49*
0.42 **
-0.03
PR37
-0.52*
-0.23
0.13
PR19
-0.66 *
-0.50 *
-0.39
0.03
-0.17
0.42
MPDI19
-0.67 *
-0.50 *
-0.15
0.41
MPDI37
-0.51 *
-0.21
0.15
0.02
POLDIF
-0.63 *
-0.40 *♦
-0,02
0.26
-0.74 *
0.44
-0.23
H19
.55 *
0.37
0.34
-0.36
V19
-0.14
0.01
H19
0.26
0.32
V37
-0.34
-0,15
H37
-0.1
0.01
H19/V37
0.58*
0.49*
PR37
-0.36
-0.24
PR19
-0.59 *
-0.50 *
MPDI19
-0.61 *
-0.51 *
MPDI37
-0.38
-0.29
POLDIF
-0.55 *
-0.45 *
NDVI
-0.11
-0.1
0.56*
0.18
0.68*
0.53*
-0.33
0.1
0.34
0.45 ♦
0.12
0.33
MPD137
-0.37
-0.19
-0.19
POLDIF
-0.34
-0.17
-0.17
NDVI
0.29
0.22
I
MPDI19
-0.31
o
IS)
*
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1992
1993
Channel 3
Channel 4
Channel 5
NDVI
NDVI
-0,36
1994
Channel 3
Channel 4
Channel 5
NDVI
* Marked correlations are significant at p < 0.05
** Marked correlations are significant at p < 0.1
Channel 5 data not provided for 1994
While there is no significant relationship between the thermal channels for 1992 and the
microwave algorithms that utilize the polarization difference as a measure o f surface
variable change, the observed inverse relationship is seen through observation of the
remaining two years o f data (1993 and 1994).
For both the 1993 and the 1994 datasets, a significant inverse relationship exists
between the average polarization difference (POLDIF) and channel 3 (r = -0.63 and
r = -0.55 respectively). A similar relationship can also be seen between channel 4 and
POLDIF (r = -0.40 and r = -0.45 respectively) although for 1994 the relationship is
somewhat weaker than that obtained for the previous year. Observation o f the remaining
microwave channel combinations (MPDI19 and 37 as well as PR19 and 37) further
demonstrate the inverse relationship between the microwave frequencies and the thermal
channels, particularly channel 3.
4.2.2 Discussion and Evaluation of Observed SSM/I and AVHRR Relationships
As mentioned in the previous section, a significant relationship between NDVI and
M PDI37 or POLDIF was generally not observed for the three years o f interest except for
perhaps the 1992 period. The inverse relationship obtained for 1992 can be considered
rather fortuitous for the following reason.
In all cases both the AVHRR and SSM/I
datasets were acquired as close together as possible, however, due to time and budgetary
constraints, daily AVHRR data could not be obtained.
Hence, a
maximum value
composited image could not be created as described by Holben (1986).
Therefore, a
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
commercially available weekly composite had to be obtained from the Manitoba Remote
Sensing Centre (MRSC). The centre routinely creates weekly composites as input to an
operational Crop Information System (Korporal et al., 1989).
However, the first
composite dataset is not required until mid May, therefore the earliest available composite
date for the 1992 season was May 21-31. Since the SSM/I data had already been acquired
it was not feasible to re-order the data for this concurrent time frame. Data acquisition for
the 1993 and 1994 time period was therefore planned accordingly to remedy this
shortcoming. Plots of the 1992 data suggest there is an inverse relationship between
NDVI and the average polarization difference (POLDIF) and a similar relationship
between the Polarization Ratio (PR37) and NDVI. However, in light o f the problems
obtaining coincident datasets, these results are somewhat misleading. Figures 10 through
12 show SSM/I images representing snow covered pixels for the study area as derived
from a snow mapping algorithm proposed by Nagler and Rott (1992). For the 1992 time
period the majority o f central and northern Ontario was heavily influenced by snow cover
thereby causing the measures of polarization difference to increase in a manner similar to
that of bare soil. McFarland et al. (1987) found that the 37 GHz polarization difference
increased linearly with increasing snow depth up to about IS cm, then the polarization
difference remained between 20 and 30 Kelvin with further increases in snow depth.
Thus, in the high northern latitudes the polarization difference is likely to be as high as 25
Kelvin during winter months. Other researchers mapping snow area, as well as snow
depth, have also noted increases in polarization differences as high as 1 0 - 3 0 Kelvin in the
absence o f vegetation (Walker, 1995).
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 10.
Snow cover extent for April 2, 1992
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 11.
Snow cover extent for May 26, 1993
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 12.
Snow cover extent for May 28, 1994
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the classic relationship of NDVI versus MPDI37 presented by Tucker and Choudhury
(1987) polarization differences as high as 25 Kelvin were reported for the Sahara Desert,
an area characterized by little or no vegetation, however, as the proportion o f vegetation
increased in the pixel, the polarization difference showed a corresponding decrease. As
can be seen from figure 10, there is a small area in the southern portion of the image that
appears to be snow free. Since the polarization difference measure is much higher for
snow covered areas than those without snow cover, it is likely that the area in southern
Ontario has yielded low measures of polarization difference and high NDVI measures.
Meanwhile the snow covered areas have provided a greater polarization difference. For
this reason the appropriate ranges of polarization difference and NDVI measures noted by
Tucker and Choudhury (1987) have been shown here. Furthermore, calculation o f the
NDVI at a later date resulted in higher values than would be expected for the
corresponding SSM/I acquisition date.
Therefore in conjunction with the increased
polarization difference in the north and elevated NDVI values to the south, an inverse
relation has resulted.
This would explain why a weak, but significant inverse correlation
exists between the NDVI and the POLDIF and also PR37 (r = -0.45 and r = -0.41
respectively) despite the fact the dates o f acquisition are almost two months apart and the
surface cover types would be quite different. Correlations between the two 19 GHz
algorithms (MPDI19 and PR19) and the NDVI measure are perhaps a little more difficult
to explain although the relationship appears to be similar to that o f the 37 GHz algorithms.
It is possible that the lower frequencies of the SSM/I sensor (19 GHz) are scattered less
by a typical snowpack (Hollinger et al., 1991) and as a consequence microwave emissions
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
are less affected by the surface and more information is obtained with respect to the
internal characteristics of the snowpack.
Since a detailed understanding o f snowpack
characteristics was not the objective o f this study, suffice it to say that the internal nature
of the snowpack results in a further depression in temperature brightness values (Chang et
al., 1987) and therefore provides an inverse correlation with NDVI as noted above.
Analysis o f the 1993 and 1994 data shows no inverse correlation between NDVI
and MPDI37 or PR37. In fact the inverse relationship is not seen at all. One possible
reason for this may be that the dates o f image acquisition were chosen to reflect the
beginning o f the growing season when vegetation density was relatively low across all
ecoregions.
This was performed to increase the possibility of monitoring surface soil
moisture. Since 19.3S GHz microwave emissions can be substantially affected by the
presence o f a vegetative cover, early spring was the best time to look at this possibility.
As a result o f the decision to acquire imagery in early spring, the range o f NDVI and
polarization difference is relatively low across the transect which is the most likely reason
for the poor relationship between the two variables.
As previously noted, the geographic environments chosen for this study were not
as diverse as those analyzed by Choudhury and Tucker (1987).
The NDVI versus
MPDI37 relationship with all three years o f data combined is shown in figure 13. Despite
the erratic nature of the datapoints, the results can possibly be considered in agreement
with those presented by Teng et al., (1995) (figure 14) for the following reason.
Curve
B represents data points collected from one geographic region under different
meteorological conditions and crop phenology with the data being averaged every five
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 13.
Relationship between the MPDI (37 G H z) and NDVI for all three years o f data
combined
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
18
16
14
++
12
+ +
10
8
6
++
++
4
2
0
0
0.1
0.3
0.2
0.4
0.5
NDVI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 14.
Results o f comparisons between NDVI and M PD I37 for various geographic regions
(Taken from Teng et al., 1995)
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A (Choudhuiy and Tucker; 1987)
B (Tengetal.1995)
Central Iowa, 1988
5 - day averages
U
X
u
G
a
u
G N
bi Z£.
a ^
Q rn
lasi
Is
II
15
sO |!^.
0.0
0.1
0.2
0.3
0.4
0.5
Normalized Difference Vegetation Index
(NDVI from AVHRR)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
days throughout the growing season. In comparison data points for this analysis were
extracted from more diverse geographic regions as well as different meteorological
conditions for a single date during the spring o f each year. The range of NDVI values
were similar for both curves (0.03 to 0.4), despite the difference in time o f year (early
spring versus complete growing season). This similar range in NDVI may have resulted
from the Boreal region where, even during winter months, microwave emissions may still
be somewhat indicative of any vegetative cover. Likewise, the ranges in MPDI values
were also similar (3 to 14 Kelvin). Further analysis of curve B shows that MPDI varies
more at the lower vegetation indices (0.05 to 0.15) than the higher indices (> 0.25) which
is consistent with previous studies (Teng et al., 1993). The large range o f MPDI at lower
NDVI is likely related to the SSM/I pixel heterogeneity resulting from a mixture of
vegetated/bare soil surfaces and surface water, and to a lesser extent atmospheric
interference, which could not be accounted for, however, was minimized.
The poor
correlation between NDVI and MPDI37, PR37 and POLDIF can therefore be explained in
the following manner. Teng et al. (1995) compared the relationship of NDVI and MPDI37
for a temperate, densely cropped area in central Iowa with the characteristic curve
produced by Choudhury and Tucker (1987) collected from many diverse geographical and
climatic regions. Indeed, for vegetation densities encountered for this present analysis
(0.10 and 0.25), Teng et al. (1995) showed a weaker inverse relationship between NDVI
and MPDI37.
The results obtained from this analysis for both 1993 and 1994 are
presented in figure 15. These results are in agreement with the results o f earlier studies
(Becker and Choudhury, 1988; Townshend et al., 1989) all of whom noted the difficulty in
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 15.
Relationship between the MPDI and NDVI for 1993 and 1994 over Ontario
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
18
16
14
12
10
8
6
4
2
0
0.05
0.1
0.3
0.15
0.35
0.4
0.45
0.5
NDVI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
distinguishing denser vegetation using the seasonal variability o f MPDI. These studies
also confirmed the overall decrease in MPDI as geographic location changed from arid to
more vegetated. Results of the analysis performed by Teng et al. (1995) showed that the
seasonal inverse relationship between NDVI and MPDI for similar geographic areas was
quite subtle. From this comparison it is apparent that comparing single dates o f SSM/I and
AVHRR imagery with one another does not allow observation o f the potential trends
which have been shown by other researchers (Choudhury, 1989; Choudhury, et al., 1987;
Choudhury, 1990).
A further consistent result shown in the correlation matrices (table 10) is the
relationship between the Normalized Brightness temperature (T nb) and the thermal
channels (3 and 4). This positive relationship was observed for all three years of data.
Although the channel 4 data shows less o f a relationship with the Normalized Brightness
temperature, it will be discussed first since the use of this channel has been favoured over
channel 3 and is therefore better understood at this time. Again, despite the disparity in
dates of image acquisition for 1992, it is clear that a relationship exists for all years. Two
possible explanations are presented here for the relationship between Tnb and thermal
infrared data.
Since thermal-IR data respond to evapotranspiration cooling which is
related to the photosynthetic activity o f plants (Toure et al., 1991),
any increase in
vegetation evapotranspiration accompanying increased plant photosynthetic activity during
initial growth stages would cause
a corresponding decrease in the surface canopy
temperature. With this in mind, a corresponding inverse relationship between the NDVI
and the thermal-IR data should exist since a denser vegetation canopy (higher NDVI)
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
would be more photosynthetically active and therefore exhibit a lower thermal
temperature.
Ignoring the data presented for 1992 for reasons discussed previously,
table 10 shows that indeed there is an inverse relationship between the thermal-IR and the
NDVI for both 1993 and 1994. Similar inverse results between crop canopy temperature
and NDVI were shown by Toure et al., (1991) albeit for wheat fields in western Canada.
Furthermore, Price (1989) found that NDVI and thermal-IR data were inversely correlated
throughout the early part of the growing season when the vegetation was in a growth
stage, however, toward the end of the season the inverse relation was not as strong due to
plant maturation and senescence. This relationship could not be fully tested here due to
availability and timing o f data acquisition. Therefore, the positive relationship between
channel 4 and Tnb might therefore be explained in terms o f evapotranspiration. Increases
in vegetation density provide a corresponding increase in plant photosynthetic activity and
amount o f moisture transpired by the plant. Consequently an increase in vegetative cover
is seen as a decrease in surface temperature and ultimately a decrease in Normalized
Brightness temperature.
A second explanation is that the surface conditions become
wetter across the transect thereby causing a depression in thermal temperatures and the
Normalized Brightness temperature (Hollinger, 1991).
As mentioned, data was acquired for the most part during late May o f each year,
therefore plant evapotranspiration may not be the most logical explanation for the inverse
relationship between the two measures. It is more likely that this relation is related to the
amount of surface and/or vegetation water content in each pixel as well as the diurnal
temperature ranges across the Ontario transect.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The generalized patterns o f diurnal temperature for the three different ecoregions
examined here are presented in figures 16 and 17.
Each diurnal temperature pattern
shows the effect o f changing latitude such that the James Bay Lowland areas reach a
lower peak thermal temperature later in the day than southern Ontario. Surface
temperature together with surface vegetation water content are shown as a function of
location throughout the transect.
Clearly, as latitude increases, thermal temperature
decreases, conversely as latitude changes from southern Ontario through to James Bay,
the amount o f water present in each SSM/I EFOV also increases. As the amount o f water
increases in each pixel the Tnb also increases (Hollinger, 1991) and thermal temperature
decreases. The reason for a stronger correlation between channel 3 and the Normalized
Brightness temperature is not completely understood at this time due to the limited use of
channel 3 for land surface analyses. Use of channel 3 data has been hampered by sensor
problems (Fedosejevs, 1995) as well as the difficulty in estimating atmospheric attenuation
and topographic effects (Collier et al., 1989; Cihlar, 1987). However, it would seem
logical that correlations between channel 3 and Normalized Brightness temperature are
related in a manner similar to those presented for channel 4. Perhaps the best explanation
is that both sensors measure brightness temperature, a function of the ground surface
temperature and condition and the surface material emissivity. Therefore, it is likely that
the two measurements are related through amount o f surface water, temperature and
amount of evapotranspiration. This would seem the most logical explanation in light of
the work by Toure et al. (1991) and the results presented here. Although this research
was unable to confirm whether or not channel 3 provided a better measure of surface
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 16.
Generalized diurnal temperature range for the Ontario transect in May
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
Hudson/James Bay Lowlands
0
12
Boreal Region
e.
0
12
Southern Ontario
0
4A M
12 n o o n
Time o f Day
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4A M
Figure 17.
Generalized surface temperature and water content across the
Ontario transect at 7:00 AM
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
12
Water content o f
Vegetation and
Temperature o f
Surface at 6:30AM
W a te r
Temperature
Southern Ontario
Hudson/James Bay
Lowlands
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
temperature than did channel 4 it may provide a rationalization for a stronger correlation
between channel 3 and TnbFinally, examination o f the correlation matrix (table 10) shows that the most
consistent relationship between the AVHRR and the SSM/I sensors is between the thermal
(Channel 3 and Channel 4) and the microwave algorithms (POLDIF, MPDI and PR). In
all cases there is an inverse relationship between the respective channels o f each sensor.
Although this relationship was not explicitly tested, the results can be explained in the
following manner for both 19 and 37 GHz. In the same manner that Tnb responded to the
proportion o f water in each SSM/I pixel, the respective measures o f polarization
difference respond also. Becker and Choudhury (1988) showed an increase in polarization
difference over open water, however, the influence of vegetation caused the polarization
difference, at 37 GHz, to decrease. An increase in surface water will result in a decrease
in thermal temperature and an increase in any polarization difference measure across the
transect.
The results from this study show that the polarization difference for 37 GHz are
affected by the presence of vegetation, however, MPDI 19 and PR19 appear to be less
affected, due to an increased penetration depth, as shown by the stronger correlations.
4.2.3 Summary
In general, the observed relationships between data received by the SSM/I and the
AVHRR sensors appear to be somewhat consistent with previously documented results
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
which have shown an inverse relationship between NDVT and MPDI and a direct
relationship between the Normalized Brightness temperature and the thermal Infrared
wavelengths.
However, to date, the majority o f previous experiments have been
conducted in very different geographic regions with extreme climatic differences such as
Cameroon (tropical rainforest) and the Sahara Desert. Recent research has focused on
observing the relationship under less extreme conditions and, as a consequence, the
relationships have been somewhat less evident, although nonetheless encouraging.
Although the data presented here do not overwhelmingly support the previously
documented results, they do show similar trends and hint at potential relationships tinder
less varied geographic conditions unlike those studied by Choudhury and Tucker (1987).
As is often the case when dealing with such large scales of observation the data may or
may not be correlated from year to year as a result o f the seasonality of small scale
weather patterns and fluctuations in overall meteorological and atmospheric conditions.
After analyzing five years of microwave data Owe et al. (1988) found their results to be
highly variable and often somewhat inconsistent with correlations between microwave and
ground data much poorer for time periods greater than six months.
The relationships found between the thermal and microwave data might be a little
more difficult to explain at present and undoubtedly will require more detailed studies to
understand the exact physical relationships. However, plausible explanations have been
presented here and indicate that there may be a further link between crop canopy
temperature and evapotranspiration as well as the amount of surficial water measured by
the each sensor. This would be in agreement with the results obtained by Choudhury
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1991) in which a strong relationship between polarization difference and a ratio o f actual
evaporation to potential evaporation was established. The method used here to establish a
scaling mechanism has shown relative success and suggests the need to adopt a more
robust approach to scaling approach. Ground surface mixture modelling approaches
continue to be developed and tested (Dobson et al., 198S) and are likely to assist in
understanding the complexity o f heterogeneous “large area” pixels. The results obtained
are encouraging and provide the rationale to continue studying the synergy between the
two sensors for the purposes o f global change monitoring.
4.3 Temporal Evolution of SSM/I Signatures
By merging SSM/I data from three different time periods over three years, 1988,
1990 and 1992, the temporal evolution of microwave emissions for a variety of ecoregions
could be observed. No coincident AVHRR data was available for this part o f the study
and therefore all data examined are from the SSM/I sensor. Figures 18 through 22 show
temporal plots of various microwave channels and channel combinations for the three
years of interest. Each o f the respective plots will be discussed individually and then
brought together in a final summary section.
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.1 Microwave Polarization Difference Index (MPDI)
A number o f variations on MPDI have also been analyzed here, those being the
polarization ratio (PR) and the average polarization difference (POLDIF). Analysis o f the
POLDIF and PR algorithms showed that in each instance the resulting graphs were not
significantly different from the trend shown by MPDI.
Therefore the remaining two
measures o f polarization difference are discussed in context with MPDI with any
significant differences between them noted.
4.3.1.1 Southern Ontario
The temporal evolution of microwave emissions for southern Ontario are shown in
figures
18 to 21.
As previously mentioned 1988 was assumed to be a dry year.
However, observation of monthly weather means recorded at ten meteorological stations
did not entirely substantiate this thought (Appendix). Perhaps because data were acquired
only for the months of interest (four dates throughout the entire year which makes it
difficult to see fluctuations and trends in the data), this hypothesis could not be entirely
confirmed. The weather data recorded for 1992 did in fact reflect that, in general, 1992
was much wetter than normal. For the purpose o f this analysis, 1990 was considered to
be representative o f a normal year. The temporal curve for 1988 can be explained in the
following manner.
At the beginning of April, the majority o f agricultural land is
characterized by bare soil except for areas where a winter crop was planted. Therefore, a
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
higher polarization difference measure (MPDI, PR, POLDIF) results from a landscape in
which vegetation biomass is at a minimum and contribution from bare soil is at a
maximum. In fact, at this time, the majority o f deciduous trees would only be starting to
come into bud, therefore the influence o f any vegetation at this time is likely to be small.
Choudhury (1989) noted similar results stating that as the amount o f vegetation cover
increased, a much smaller polarization difference resulted. Therefore, as the crop canopy
develops a corresponding decrease in polarization difference should generally be observed.
This decrease in POLDIF can easily be seen for 1990, however, for 1988, the decrease in
polarization difference is less dramatic. One possible reason for this slight decrease is that
due to a somewhat drier spring (Environment Canada, 1993) crop growth and
development was delayed. As a result, vegetation biomass would not have reached a
maximum as in a normal year and therefore there is a much larger soil component
contributing to the microwave emission and therefore a less dramatic decrease in
polarization difference than observed in 1990.
During plant senescence and harvest, the
polarization difference increases as a result o f decreased vegetation water content. A
further infuencing factor affecting the polarization difference is that during this time of
year harvesting would almost be complete providing a greater soil contribution to the
emission of each SSM/I pixel. With the onset o f snow, polarization difference measures
increase to values slightly greater than those o f bare soil which is consistent with the
findings o f Walker (1995).
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 18.
37 GHz polarization difference for three different ecoregions
(a) Southern Ontario
(b)
Boreal
(c) Hudson/James Bay Lowlands
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
i
—♦ —1988
—■ —1990
1992
20-Jul
b)
i
14
12
10
—♦ —1988
- ■ —1990
1992
8
6
4
2
0
20-Feb
«
11-Apr
i— 31-May
i
i
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
28-Oct
17-Dec
5-Feb
C)
14
12
10
8
6
4
2
0
20-Feb
i
i
11-Apr
31-May
i
■■ — . — i —
20-Jul
8-Sep
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 19.
Average polarization difference for three different ecoregions
(a) Southern Ontario
(b) Boreal
(c) Hudson/James Bay Lowlands
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
e
=
14
12
—♦ —1988
- • —1990
1992
£
20-Feb
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
b)
12
-♦ -1 9 8 8
—■ —1990
1992
£
20-Feb
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
C)
16
14
12
10
—♦ —1988
—• — 1990
1992
8
6
4
2
0 --20-Feb
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 20.
Polarization Ratio for three different ecoregions
(a) Southern Ontario
(b) Boreal
(c) Hudson/James Bay Lowlands
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
0.0300
p
5t
0.0250
0.0200
—♦ —1988
-•-1 9 9 0
1992
0.0150
I
0.0100
0.0050
0.0000
20-Fob
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Doc
5-Feb
b)
0.0300
P
0.0250
0.0050 >
0 .0 0 0 0
-----------------
20-Feb
—
I-----------------•-----------------1
I
•-----------------1
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
C)
0.0300
P
0.0250
§ 0.0200
Ig :
0.0150
fi
0.0100
£.
0.0050
0.0000
20-Feb
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 21.
19 GHz polarization difference for three different ecoregions
(a) Southern Ontario
(b) Boreal
(c) Hudson/James Bay Lowlands
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
! a)
—♦ —1988
- ■ —1990
1992
o 1
20-Feb
•
.
.
.
.
.
.
11 -Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
b)
18 ■
16 •
14 •
12
■
10
•
8 •
6 ■
4 •
2 •
0 --20-Feb
5-Feb
20-Jul
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Weather data substantiated the initial hypothesis that 1992 was a wetter year than
normal with monthly mean rainfall values almost twice that recorded for 1990. The
temporal sequence of polarization difference shows the effect o f a wetter year on
microwave emissions.
In general, wetter conditions serve to depress the microwave
brightness temperature and cause a decrease in polarization difference. For April 1992,
the polarization difference is almost 4 degrees below that observed for both 1988 and
1990, however, this is to be expected in light o f twice the amount o f rainfall for the same
period in 1992.
For July, the polarization difference is slightly higher than would be
expected in a normal year. The exact reason for this is not quite clear since weather data
showed this to be the wettest of the three years. It is quite possible that the canopy has
not completely developed
as a result o f a wetter spring and therefore polarization
difference values are not as low as would be expected for a fully developed canopy. This
idea could not be confirmed since an alternative measure of biomass was not obtained.
4.3.1.2 Boreal Region
The temporal evolution of the microwave emissions for the Boreal region are
presented in figures 18 to 21 for the three dates o f interest. Walker (1995) suggested that
the polarization difference should not change much throughout the year for this region,
however, a difference is evident from figure 18. The figure shows an opposite trend
compared to what would be expected based on knowledge of how the polarization
difference responds to changes in vegetation phenology.
As a first consideration, this
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
opposite trend may be related to calibration and precision as well as locational accuracies.
All three years of data appear to follow a similar trend which suggests that the temporal
response is either related to the precision o f the instrument or is a function o f the ground
surface itself and not a result o f atmospheric effects. Although the weather data gathered
for this region show that meteorological conditions were quite different for each year, it is
difficult to determine this from the any of the polarization difference signatures. For both
April and December, polarization difference measures are lower than the remaining two
months.
Choudhury et al. (1990) have shown that polarization difference values are
substantially affected by the woody component of a forest. It is therefore possible that the
polarized snow emissions resulting from the ground are being attenuated and the
depolarized emission from the woody component o f the forest increases, thereby causing
an overall lowering o f the polarization difference for 37 GHz.
Wigneron et al. (1995)
also documented a decrease in polarized surface emissions resulting from an increase in
depolarized vegetation emissions. For the remaining months (July and October) the
polarization difference remains relatively stable around 7 K.
During this time the
polarization difference is largely representative of the forest canopy which is, for the most
part, green during this time period. Observation o f the polarization difference for the 19
GHz shows less variability throughout the growing season as suggested by Walker (1995)
with polarization differences for the growing season about 8 K.
The difference in the
temporal response for each polarization combination may be due, in part, to precision and
calibration issues which were assumed to be stable for this study.
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.1.3 Hudson/James Bay Lowlands
Figures 18 to 21 again show the temporal evolution of the microwave emissions
for the Hudson/James Bay lowlands for all three years.
Again, these curves can be
explained as a function of change in vegetation phenology.
During the spring
the
polarization difference ranges from nine to eleven Kelvin for all three years. This would
suggest that for all three years surface conditions were similar. Mean temperatures and
snow cover for April o f each year, indeed confirmed that surface conditions were similar.
For all years the polarization difference values infer that the majority o f the pixels chosen
were snow covered during this time. Shortly after the snow has melted, the vegetation
begins to “green-up” and the polarization differences decrease for all years.
For July
1992, meteorological data indicates that on average, this month was wetter than the same
month for the previous two years of data. Therefore, increases in the fraction o f the
amount of standing surface water in a SSM/I pixel would be the main reason for the slight
increase in polarization difference. Giddings and Choudhury (1989) showed how areas
seasonally inundated by water in South America could easily be delineated and monitored
using passive microwave data. With the onset o f winter, vegetation begins to senesce and
the propensity of snow increases, the polarization difference is seen to increase
accordingly until it reaches a maximum in December when snow and ice dominate the
microwave response.
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.2 Normalized Temperature Brightness (T nb)
Since microwave emissions are a combination of surface material emissivity and
surface temperature, any fluctuation in surface temperature will cause the microwave
response to change due to solar heating throughout the day. Such surface temperature
change can yield quite different emission responses for the same land surface type.
Therefore a Normalized Brightness temperature (T19H/T37V) was calculated to mitigate
against the influence o f solar heating.
In the absence of a direct measure of surface
temperature Hollinger (1991) showed that T37V could be used as a surrogate measure of
skin surface temperature due to its smaller penetration depth.
4.3.2.1 Southern Ontario
The relative change in Normalized Brightness temperature over the course of a
year is shown in figure 22. Analysis of these curves show the difficulty o f determining
surface parameters for an heterogeneous area from microwave emissions early in the year.
In 1988 and 1990, Tnb values were low in comparison to the rest of the year and also in
comparison with 1992. Although 1992 had a wet spring, the majority o f the precipitation
did not occur until mid April (after the SSM/I acquisition). Analysis o f the monthly
records for March showed that 1992 was indeed drier than the previous two years. This
result was further substantiated
by the higher
T
nb
for early April which is further
indicative o f drier conditions.
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 22.
Normalized temperature brightness for three different ecoregions
(a) Southern Ontario
(b) Boreal
(c) Hudson/James Bay Lowlands
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a)
1.09
1.07
1.05
1.03
- ♦ —1988
- • —1990
1992
1.01
0.99
0.97
0.95
0.93
0.91
20-Feb
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
b)
1.09
1.07
- ♦ —1988
- • —1990
1992
1.05
1.03
1.01
0.99
0.97
0.95
0.93
0.91
20-Feb
11-Apr
31-May
20-Jul
8-Sep
28-Oct
17-Dec
5-Feb
8-Sep
28-Oct
17-Dec
5-Feb
C)
1.09
1.07
1.05
1.03
1.01
0.99
0.97
0.95
0.93
0.91
20-Feb
11-Apr
31-May
20-Jul
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The rest of the temporal evolution appears to approximate the characteristic decline in
microwave response with increase in vegetation cover and steady increase as vegetation
begins to senesce.
Overall, the Tnb for this region does not fluctuate very much
throughout the year. Although it could not be confirmed here since it was beyond the
scope of this research the small change in microwave response is probably related to the
problem of mixed pixels and the difficulty in establishing an accurate ground-cover
composition for each SSM/I pixel..
4.3.2.2 Boreal Region
The temporal evolution of microwave signatures for this region are presented in
figure 22. In early April, microwave emissions are representative o f a combination of
snow and vegetative cover and therefore Tnb is at its highest value. The slight differences
in Tnb from year to year are likely related to changes in meteorological conditions and not
from changes in surface type since the ecoregion is relatively stable. As the vegetation
canopies develop, microwave emissions become increasingly more indicative o f the
surface vegetation and less o f the underlying surfaces. As is the case with polarization
difference measures any increases in vegetation cover are complimented by steady
decreases in Tnb such as those seen through July and into October. Changes in the general
shape of temporal evolution o f microwave emissions are much less pronounced which is a
direct result of the small changes in surface cover throughout the year.
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.2.3 Hudson/James Bay Lowlands
Analysis o f the temporal change in T nb for this region shows remarkable
consistency in general shape of the curves (figures 18 to 22) suggesting that the
microwave response associated with this ecoregion does not change significantly over
time. Explanation of these characteristic curves is as follows. In April, the majority of this
low lying area is frozen and covered in snow which has not begun to melt. The Tnb for
this time o f the year is relatively high resulting from dry snow and little vegetation
influence. Differences in Tnb for the three years of interest are quite small and are most
likely related to variations in weather conditions and perhaps sensor calibration and
precision.
It is encouraging though that despite any differences in meteorological
conditions the microwave response does not change significantly for the three years of
data examined. This demonstrates the possibility of utilizing microwave data to record
surface change in this ecoregion over time.
As conditions become warmer and vegetation cover again dominates the response
a decrease in microwave emissions is seen up until the onset of snow when
T
nb
increases
slowly as a result of frozen conditions. Since this region is primarily made up of lowland
marsh flats the decrease in microwave response can be considered a result of increasing
vegetative cover together with increases in surface water. This is evident by the lower
T
nb
measures (< 0.93). Once conditions become colder, vegetation senesces and any
open water freezes thereby causing an increase in T n b -
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.3.3 Summary
While only data from specific months were available for this temporal analysis,
they do appear to indicate the state of the ground surface conditions for these specific
ecoregions which is encouraging for monitoring changes in these respective locations.
This is especially important for the Boreal and Hudson Bay ecoregions since extensive
research is currently being conducted under the Boreal Ecosystem-Atmosphere Study
(BOREAS) (Cihlar et al., 1995; McDermid et al., 1995) to develop algorithms capable of
transferring our understanding o f energy exchange processes from local to regional scales,
for input into numerical models of environmental change. In general, the temporal
evolution of SSM/I signatures associated with all algorithms showed remarkable
consistency despite any differences in weather conditions, however, in some instances the
effect of different weather conditions can be seen (i.e. southern Ontario, 1992). The initial
hypothesis that 1988 was a dry year, 1990 a normal year and 1992 a wet year is really only
evident for the southern Ontario ecoregion based on microwave response. Unfortunately
large scale studies such as this are hampered by the problem of adequately characterizing
meteorological conditions. Most large scale measures o f soil moisture, for example, are
determined from models and may not be sensitive to small scale variations at such a large
scale. Undoubtedly more robust methods of measuring/estimating dynamic variables such
as soil moisture need to be established if measurements o f surface conditions are to be
routinely made from space.
Many researchers have shown the geographic dependency of
microwave emissions, and indeed the results of this analysis appear to support this
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
statement with characteristically different curves established for each ecoregion. Kerr and
Njoku (1993) proposed the use of a polarization ratio to minimize the effect o f variable
surface temperature instead of using the traditional polarization difference. However, the
recommendations were based on interpretation o f the Scanning Multichannel Microwave
Radiometer (SMMR) data which had a midday equatorial crossing. The results presented
here show the similarity in general
shape o f the average polarization difference,
polarization difference and the polarization ratio. It is therefore evident that temperature
brightness values recorded by the SSM/I sensor during a morning pass are less affected by
solar heating. Furthermore the use of the average polarization difference is preferred since
it incorporates polarization differences from both 19 and 37 GHz channels and does not
seem to be affected by atmospheric interactions, however, any of the other algorithms
measuring polarization difference would suffice.
Also, at a minimum, four well timed
dates will suffice to understand temporal nature o f a given area. It is, however, advisable
if the data are to be used to quantify differences in the respective environments that
ground and satellite data be acquired on a biweekly or monthly basis
4.4 M apping Soil Moisture Using SSM/I
Microwave emissions recorded from a bare soil surface at any location are known
to be affected by the presence of vegetation cover. Therefore, accurate estimation of near
surface soil moisture requires that the vegetative component of microwave emission be
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
removed o r minimized to ensure microwave responses are indicative of moisture
conditions in the soil. In the previous two sections the relationship between NDVI and
average polarization difference (POLDIF) has been established and further demonstrated
as a useful method o f monitoring temporal variations due to changes in biomass density.
Furthermore, the relationship between Tnb and soil wetness have been eluded to in the
previous two sections. The relationship between NDVI and the average polarization
difference as established from twenty SSM/I and corresponding AVHRR pixels is shown
in figure 23. A significant inverse relationship (r = - 0.742) was found between POLDIF
and NDVI which further supports the results o f Hollinger (1991).
The results of figure 23 show that despite the fact the images were acquired in
early September, there was still a considerable amount of vegetation present on the ground
as demonstrated by the moderately high NDVI (> 0.2). Furthermore, average polarization
difference values between 2 K and 6 K have been shown by Hollinger (1991) to be
indicative o f medium density vegetation covers.
A scatterplot of soil moisture versus Normalized Brightness temperature for all
vegetation densities encountered at each weather station is presented in figure 24.
It
would appear that no relationship exists between the two measures o f surface conditions.
Analysis o f the vegetation relationship shows that the average polarization difference
measure can be separated into three classes (Hollinger, 1991; Neale et al., 1990). The
medium-high density class was determined to range between 3 K and 5 K.
This is
consistent with the values established for medium-high density vegetation by Neale et al.
Ill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 23.
Relation between average polarization difference and NDVI
for September 6-12, 1993
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
12
y = -61.S83x+ 17.143
r = 0.7422
o
o
co
ba
10
£
a
C
O
a
8
J5
6
u
«>
2u
>
<
4
o
a.
2
0
0.00
0.05
0.10
0.15
0.25
Normalized Difference Vegetation Index (NDVI)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1990) in which they studied forested areas to establish the range o f polarization difference
associated with this class. The major difference in range of polarization difference values
established in this study compared to that o f Neale et al. (1990) are most likely a result of
differences in vegetation type, density, and phenology as well as geographic location.
Based on the scatterplot (figure 24) together with the categories established by Hollinger
(1991) the medium density class was set between 5 K and 7 K, however, due to a lack of
pixels identified within this range, it was difficult to further evaluate this class and
determine appropriate equations for the purpose of retrieving soil moisture measurements.
A number o f pixels showed average polarization differences o f less than 3 K.
These areas corresponded to data points located in the Boreal region o f the transect
where a significant portion of the ground covered by each SSM/I pixel is heavily forested.
Neale et al. (1990) evaluated the average polarization difference for a variety o f different
surface
types
and
consequently
noted
relatively low
polarization
differences
(approximately 0.7 K) for reasonably dense forested vegetation. Determination o f soil
moisture under areas o f dense vegetation cover is difficult due to the small penetration
depth of the 19.35 GHz frequency. For this reason a retrieval algorithm for estimating
surface moisture was not considered for such a vegetation density class.
Again, based on the scatterplot and the study by Hollinger (1991), the low density
vegetation class was determined as being > 7 K.
Each SSM/I pixel was co-located with on of the synoptic weather stations for
which soil moisture data was available (Appendix). Due to the size of the area covered by
the SSM/I sensor it is entirely possible that the point source weather data are not
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 24.
Scatterplot between soil moisture and Tnb for all vegetation densities
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.02
1
++
0.98
0.96
0.94
0.92
0.9
0.88
0
2
4
6
8
10
12
14
16
Soil Moisture (%) (0-120cm)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
indicative of small scale frontal storms, however, since no other timely measure o f large
scale soil moisture was available, this data was used. To determine soil moisture values in
the presence of surface vegetation, the data were separated to correspond to differences in
vegetation density (i.e. high, medium, and low). The majority o f the pixels appeared to
fall within the medium-high density class as indicated by their average polarization
difference measure (3 K - 5 K). Figures 25 and 26 present the relationship between soil
moisture and the Normalized Brightness temperature
(T
nb)
for the medium-high and low
density vegetation classes. Despite the fact that the number of points is quite small (n =
17), an inverse relationship between Tnb and soil moisture is evident (r = 0.7465). Owe et
al. (1992) found similar results between emissivity and volumetric soil moisture using 6.6
GHz SMMR data over an area in the Republic o f Botswana. A similar relationship was
also observed between Tnb and soil moisture for the low density class. Although only
eleven of the SSM/I pixels fell into this density class, however, an inverse relationship ( r =
0.6059) was again noted and is further consistent with the results of Hollinger (1991).
The regression equations for each vegetation density class are presented in table
11.
VegetationDensity Glass;
Slopd;
Intercept
Low Density
Medium Density
Medium High Density
-0.0028
*
0.9827
*
0.6059
-0.0007
0.9997
0.7465
Model T19H/T37V = a + B * (SM)
* T o o fe w p ix e ls
Table 11. Regression coefficients for the respective vegetation densities
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 25.
Scatterplot and regression line for medium-high density vegetation
over southern Manitoba (Sept 7, 1993)
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.998
+ +
0.997
y = -0.0007x ♦ 0.9997
r = 0.7465
0.996
0.995
0.994
0.993
0.992
0.991
0.99
0.989
0.988
0.987
0
5
10
15
20
Soil Moisture (%) (0-120cm)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 26.
Scatterplot and regression line for low density vegetation
over southern Manitoba (Sept. 7, 1993)
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.965
y = -0.0028X + 0.9827
r = 0.6059
0.96
0.955
ft
0.95
&
ON
0.945
0.94
0.935
0.93
0
5
10
15
20
Soil Moisture (%) (0-120cm)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The equations are shown in table 12 in their operational form, as inversions o f the
equations developed On the previous table).
VegetahonDensrty Class
Low Density
Medium Density
Medium High Density
TEST
B1
-357.14
*
350.96
*
-1428.57
1428.14
SM = Al +B1 * (T19H/T37V)
* Too few Pixels for analysis
rbl>7K
5 < [b] < 7 K
3 < [b] < 5 K
[b] (19V +37V)/2 - (19H - 37H)/2
Table 12. Surface moisture retrieval algorithms for the respective vegetation densities
Unfortunately the equations developed here for the purposes of soil moisture
retrieval were not further tested due to lack of time and difficulty in obtaining a synoptic
measure of soil moisture, however, the appropriate methodology has been outlined and
showed the potential o f the SSM/I instrument for determining surface moisture providing
the vegetation cover can be accounted for. Unfortunately due to the lack o f weather
stations reporting soil moisture for a 0-10 cm soil depth, results were not as favourable as
those presented for the 0-120 cm. Correlations between the 0-10 cm soil moisture model
and Tnb were made but showed no relationship. This result further accentuates the need
for an alternative method o f producing synoptic maps o f soil moisture and also the
importance of accounting for vegetation effects on microwave emissions from soil.
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.5 Summary of Analyses
In spite o f the differences between each o f the individual studies, a common theme
appeared throughout.
Results of the scaling analysis documented the difficulty of
correlating data from two different sensors for a single date due, in part, to locational
difficulties.
Furthermore, comparison o f the results obtained here with the results
presented by Choudhury and Tucker (1987) showed that this type o f relation while
important in showing the link between the two satellites does not necessarily inversely
related for homogenous geographic areas. While the results obtained here have been
explained with respect to the findings o f Teng et al., (1995) they do not appear to
substantiate the relation obtained by the researchers between NDVI and MPDI37. This is
not surprising since the objective at the outset o f the study was to determine the potential
of the SSM/I sensor to monitor soil moisture.
Therefore data was acquired when
vegetation densities were at a minimum. Undoubtedly a more distinct relationship would
have resulted if imagery was available for the same region for an entire year. Although the
focus of literature has been directed toward the NDVT-MPDI relation, not much attention
has been given to the possible connection between the thermal channels and the
polarization difference.
Preliminary explanations for the relation between these two
variables have been put forth, however, the exact reasons for the relations are not well
understood and should be the focus o f more carefully planned studies. The results o f the
scaling analysis have shown that if a detailed interpretation of the land surface is not
required, then the scaling procedure developed here may be applied. As mentioned
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
previously a number of dielectric mixture models have been developed and tested under
laboratory conditions (Dobson et al., 1985) and may provide for a better understanding of
“large area” pixels which would certainly prove beneficial for future studies. Unfortunately
the range o f materials encountered throughout an SSM/I pixel would undoubtedly require
a detailed understanding of the dielectric properties o f each individual material.
Nonetheless use of mixture models will most likely be the answer to establishing links
between ground phenomena and the information recorded by the passive microwave
sensors.
Again, results from the temporal evolution of microwave signatures and the soil
moisture analysis suggest the need for a larger annual dataset.
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTERS
CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK
5.1 Conclusions
Passive microwave data from SSM/I were used in conjunction with optical and
thermal data acquired by AVHRR to evaluate the synergy between the two sensors for the
purposes of monitoring surface change over large areas. The primary objective was to
determine the capability o f the SSM/I sensor to provide timely information as input to an
agro-ecosystem monitoring procedure.
One specific agro-ecosystem monitoring system
proposed by Protz et al. (1993) identified information obtained by the SSM/I sensor as a
primary data source for the purpose of monitoring surface soil moisture, snow cover and
soil profile moisture. Under the main objective, three specific goals were identified 1) to
establish a method of appropriately scaling between different sensor resolutions; 2) to
evaluate the capability o f the SSM/I sensor to monitor temporal change at an ecoregion
level; and 3) to examine the possibility of mapping soil wetness from space using the
SSM/I instrument.
5.1.1 Scaling Analysis
The results of the scaling analysis showed a significant negative relationship
between energy recorded in the AVHRR thermal wavebands (channels 3 and 4) and the
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
combinations o f SSM/I frequencies and polarizations (POLDIF, MPDI 19 & 37,
T nb) .
Despite differences in dates o f image acquisition for 1992, and difficulty in merging the
two datasets, the inverse trend found in 1993 and 1994 was also observed in 1992. This
inverse relationship between the SSM/I and AVHRR sensors suggests that information
recorded by the respective sensors is a function o f surface temperature and is largely
controlled by the water content of both the vegetation and the soil. To date, however,
similar relationships between AVHRR thermal data and SSM/I for the purpose o f ground
surface monitoring have not been shown in the literature.
The very distinct inverse relationship between the Normalized Difference
Vegetation Index and the Microwave Polarization Difference algorithms shown in the
literature (Choudhury and Tucker, 1987) was not observed in this research. Instead, this
work showed a much less dramatic inverse correlation between MPDI37 and NDVI which
was similar to the findings o f Teng et al., (1995). The results presented here substantiate
the difficulty in utilizing the polarization difference as a direct indicator o f vegetation
density over similar geographic areas. The range of 37 GHz polarization difference values
was found to be much less in the presence of a vegetative cover and therefore estimation
of vegetation biomass and density from the SSM/I sensor was difficult in the three
ecoregions chosen here since the changes in vegetation biomass was quite subtle.
Conversely, use of a Normalized Brightness temperature
(T
nb),
as a measure o f surface
material emissivity, showed a much more consistent relationship with the thermal channels
of the AVHRR sensor.
Although not statistically significant, Tnb
consistent relation with NDVI.
also showed a
This indicates that surface material emissivity (e), as
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approximated by Tnb in conjunction with thermal measurements, may allow for a more
detailed analysis o f surface conditions on a single date than a polarization difference
measure. However, the utility o f the polarization difference as a surrogate measure o f
vegetation density and biomass should not be underestimated in light of the recent results
presented by Teng et al., (1995) but should be examined in greater detail over a much
longer time frame. As shown recently in the literature, the use o f a five year dataset
provided a much stronger relationship between NDVI and MPDI37 than comparison of
these variables on a single date.
The 19 GHz polarization difference (MPDI19) and ratio (PR19) measures showed
significant correlations with the thermal channels (3 and 4) o f the AVHRR sensor. The
exact physical reason for a better relationship between these variables is not clear from this
present analysis but is most likely a result of a lower attenuation of ground emissions by
the vegetation canopy and therefore a larger polarization difference.
This hypothesis
should be further tested using ground based radiometers under more controlled conditions.
Indeed, from this analysis it would appear that use of the 19 GHz channels
provides better relationships between the AVHRR sensor and the SSM/I instruments,
however, these conclusions may be geographically dependent and therefore not similar for
other environments.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.1.2 Temporal Evolution
The temporal curves of the polarization difference algorithms (POLDIF, M PD I19
& 37, PR 19 & 37) hint at the possibility of using passive microwave sensors such as the
SSM/I to differentiate between various ecoregions such as forest, wetlands and
agriculture. By establishing a time sequence history for each ecoregion, a “normal” curve
for a specific region could be determined. Using a “normal” curve for each ecoregion as a
baseline, possible long term deviations could be identified and further investigated. This
will have important implications for global change monitoring which is currently the focus
of a number of ongoing research activities (Cihlar, 1995; McDermid, 1995). Although the
results obtained have proven to be useful in establishing general trends with respect to
microwave signatures for particular regions, the use of a daily data set to create weekly
composites would undoubtedly provide for a more detailed understanding o f the temporal
changes in microwave emissions associated with each region. Furthermore, the use of
weekly composites would help to minimize any atmospheric ambiguities as well as be
more consistent with currently available meteorological data which is reported on a
weekly basis.
Results from the analysis o f the temporal profiles show, that for an
ecoregion scale (1:500 000), we are able to use data acquired by the SSM/I sensor to
delineate relatively homogeneous areas such as the Boreal forest and Hudson/James Bay
Lowlands.
Conversely, heterogeneous regions such as southern Ontario, which are
comprised o f many different surface types, make interpretation of a 25 km pixel difficult
without the use of complex surface mixture models and more robust methods of scaling.
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Therefore, research activities directed at understanding the relationship between surface
type and microwave emissions o f complex regions, such as southern Ontario, may have to
wait for future higher resolution sensors such as ESTAR to adequately make use o f
passive microwave information in an operational framework.
An additional conclusion resulting from the data analysis o f the previous two
sections is that results o f the Polarization Ratio (Kerr and Njoku, 1993) do not appear to
be significantly different than the results obtained by using the polarization difference. The
relatively small difference between the polarization difference temporal profile and that of
the polarization ratio suggests that the morning SSM/I orbits are less affected by
variability in surface temperature resulting from solar heating (section 4.3, figures 18 and
20) than the Scanning Multichannel Microwave Radiometer, upon which the derivation of
the Polarization Ratio was based.
5.1.3 Soil Moisture M apping
Results of the soil moisture modelling section showed the influence of surface
vegetation on microwave emissions from soil surfaces. Vegetation depolarizes soil
emissions and contributes polarized emissions of its own to the overall temperature
brightness recorded at the sensor. Therefore, in order to adequately assess the state of
soil moisture using the SSM/I instrument, the vegetation component of surface microwave
emissions needs to be minimized or accounted for. This research has shown that both a
single sensor (SSM/I) and a multi-sensor (SSM/I & AVHRR) approach could be used to
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
obtain measures o f vegetation biomass and density for the purposes o f soil moisture
monitoring. Estimation of soil wetness using the SSM/I instrument is, however, only
possible when vegetation densities are low as indicated by the respective average
polarization difference values (> 7 IQ. Furthermore, in the absence o f vegetation, broad
scale soil moisture measures may be adequately determined using the SSM/I sensor.
Using a simple linear model which accounts for vegetation effects, surface soil moisture
can be estimated using a normalized temperature brightness (19H/37V).
Furthermore, this research has provided evidence that the even the lowest
frequency (19 GHz) used by the SSM/I sensor is not optimal for determining near surface
soil moisture as a result of the small penetration depth and its sensitivity to vegetation.
Jackson (1993) noted if we are to successfully obtain operational estimates o f surface soil
moisture from space, lower frequencies (i.e. 1.4 GHz, X = 21 cm) need to be used.
Limited data sets have been available for determining soil moisture from space using 1.4
GHz, however, the work presented here has again demonstrated the problems associated
with measuring soil moisture using 19 GHz and further accentuated the need for lower
frequencies. Jackson (1993) also cited the problem o f increased ground resolution when
using 1.4 GHz with nominal resolutions o f SO - 100 km, which is too coarse for all but
small scale applications. Therefore, if the use o f 1.4 GHz radiometers to monitor soil
moisture is to be successful there will be a need to understand the heterogeneous nature of
large area pixels. Methods such as those identified here for scaling between datasets of
different resolution will provide an important base for understanding the complexity of
information recorded over large areas.
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Again, results of the soil moisture analysis reveal the importance o f analyzing low
resolution data sets over a number o f years to establish relevant trends which are not
related to short term annual fluctuations but rather a measurable dynamic variable such as
soil moisture.
While in all cases this research has shown weak relationships the use of these low
resolution sensors is only likely to increase with global awareness. Commensurate with
this increase in their use will be a further advancement in sensor design technology and
ultimately increases in spatial resolution.
By identifying potential relationships using
datasets such as SSM/I and AVHRR we put in place methodologies to be in a position to
take full advantage o f this advancement o f technology. The results presented here are an
initial analysis o f the SSM/I passive microwave radiometer data and, as such, a number of
parameters have been ignored and assumptions made which may not behave well under
scrutiny. Nonetheless this exploratory study has hinted at the possibility of using the
SSM/I sensor in synergy with other Earth observation platforms for monitoring Earth
surface processes.
5.2 Future Research
5.2.1 Outline for Future Processing
While the results presented here are by no means conclusive they have served to
identify shortcomings in the initial processing methodology. A number of areas, which
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
would undoubtedly expedite data processing and therefore provide much stronger
relationships in the future, are now presented. A proposed processing methodology to
analyze large area estimates o f regionalized variables, such as snow extent, soil moisture
or estimates o f vegetation biomass, is shown in figure 27.
One substantial
recommendation that can be made as a result o f this work is that large area studies should
be performed, at a minimum, for an entire year with SSM/I daily data composited into
monthly observations. The use of a single year of data is even questionable as a result o f
short term fluctuations in meteorological and growing conditions. Therefore, if possible,
studies such as these should be performed for a duration o f at least five years.
Furthermore, incorporation of a measure of atmospheric water content is needed to
account for any atmospheric contributions to Brightness temperature. This could most
likely be obtained from the GOES meteorological satellites. This may be o f importance
since Kerr and Njoku (1994) noted that up to 40 % of the variability in Brightness
temperature may be associated with the atmosphere, however, their results were based on
SMMR data which had a midday equatorial crossing. This may not be as important for
SSM/I data providing morning overpasses are utilized. Operational use o f both AVHRR
and SSM/I in conjunction would also require they both be georeferenced to a similar and
consistent coordinate system.
This is especially important for the SSM/I data since it is
affected by small amounts of drift in the orbital cycle of the satellite. Currently, this is not
accounted for thereby making precise pixel comparisons over time somewhat difficult.
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 27.
Future processing flowchart o f passive microwave data in estimating soil
moisture and vegetation biomass
(Updated from Jackson, 1993)
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Estimates of
Atmospheric Water
Content
ILWC
Weekly Composited,
Geolocated Passive
Microwave Data
Ancillary Image
Data
i.e. NOAA, TM
Calibrated
Brightness
Temperature
Ground Ancillary
Data
Meteorological
data
Antecedent
Precipitation
Index
Land Cover
Examination
Surface Temperature
Calculate Pixel
Emissivity (T ^)
Soil Texture
Vegetation Biomass
Estimation
Correction for
Vegetation
Estimation of
Vegetation Type and
Biomass
Vegetation
Water
Content
Soil
Roughness
Effect
Soil Moisture
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As a result o f this research it is recommended that AVHRR data be acquired on a
weekly basis.
Presently, a number of commercially available composited datasets exist
from the AVHRR sensor, notably the Pathfinder dataset (Agbu and James, 1995) and the
GeoComp data (Robertson et al., 1992), however, no such data products are routinely
produced for the SSM/I sensor. Therefore, a protocol for SSM/I data processing should
include, at a minimum, daily image acquisitions regridded to a common datum (UTM or
Latitude/Longitude) with computation o f weekly and biweekly temperature brightness
averages for each frequency. As previously mentioned, weekly and biweekly composites
would assist in mitigating against local frontal rainstorms which may have a significant
influence on pixel temperature brightness and are often undetected by weather stations.
Furthermore, this compositing procedure would help to ensure that variability in
temperature brightness is primarily a function of surface change.
Although available data only permitted a cursory study of the capability of the
SSM/I sensor to obtain soil moisture measurements there are a few recommendations that
can be made if we hope to use SSM/I for this purpose. Data should be acquired when
vegetation densities are low (this becomes less important as the operating frequency
decreases), notably spring and autumn. Vegetation density for each pixel should then be
established and segmented into low, medium, and high density classes. Estimation o f soil
moisture for the high density vegetation class should not be attempted since the underlying
soil emissions will be completely masked by the vegetation cover (Hollinger et al., 1991).
Ground validation data should provide soil wetness information pertinent to the top 10 cm
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of the soil profile and should be modelled using an Antecedent Precipitation Index (Saxton
and Linz, 1967).
5.2.2 Direction for Future W ork
One o f the major stumbling blocks on the use of passive microwave sensing has
been the problem o f low resolution data (-150 km) associated with the use of shorter
frequencies (1.4 GHz) passive microwave sensors and has therefore been the focus of
much research.
A promising solution has been developed which is based on alternative
antenna technology called Electronically Scanned Thinned Array Radiometry (ESTAR)
(Le Vine et al., 1989; Jackson et al., 1995).
This approach achieves better ground
resolution by synthesizing a larger filled array antenna with an array of distributed
elements that are of low volume and mass, thus overcoming the launch and construction
constraints, o f traditional antennas (Jackson, 1993). Based on known design constraints
researchers (Blyth, 1993; Jackson, 1993) note that a ground resolution cell o f 10 km could
easily be achieved in the near future for the purposes of obtaining moisture measurements
from space.
A number o f other new data sources are becoming commercially available. The
recent launch o f RADARSAT with its many different operating modes, in particular
ScanSAR (100 m resolution), offers the potential to explore the synergy between active
and passive sensors. A number of these such studies have recently been initiated on a
ground level (Wegmuller, 1994), however, use of RADARSAT data in conjunction with
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the passive microwave would extend these studies into space and provide an excellent
intermediary step between TM, AVHRR and SSM/L
Recently, NOAA/NASA made available a global archive of AVHRR imagery
dating from 1982 to present (Agbu and James, 1995).
This provides a tremendous
database of the entire world which could easily be used in combination with the SSM/I
and SMMR data sets which also date back to the early eighties. Analysis of this long term
data set would undoubtedly provide further impetus for global change studies.
Finally, for the purposes of agro-ecosystem monitoring, a multi-sensor, multi­
frequency approach would provide the necessary tools to be able to take full advantage of
data received from orbiting satellites. This is especially important if the progression to
operational monitoring is to be achieved.
The data analyzed here represents only one
component of an agro-ecosystem monitoring procedure. Each component will need to be
analyzed and protocols established for merging the many datasets and maximizing the
information extraction.
Microwave sensors, both active and passive, will enable temporal changes in
surface features to be studied in a way which has previously been impossible. The full
benefit of remote sensing is likely to come from the integration o f knowledge gained from
a range of sensor types.
This will pose new challenges in the use of geographic
information systems (GIS) which will be essential to relate these low spatial and high
temporal data sets to ground validation information, and in the development o f inversion
models which can handle a range of electromagnetic frequencies to enable variables such
as soil moisture to be estimated.
132
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
MTTMi
REFERENCES
Agbu, P.A. and James, M.E. 1995. The NOAA/NASA Pathfinder AVHRR Land Data Set
User’s Manual. Goddard Distributed Active Archive Center, NASA, Goddard
Space Flight Center, Greeenbelt.
Agriculture Canada Expert Committee on Soil Survey. 1987. The Canadian System of
Soil Classification, second edition. Agriculture Canada Publication 1646. 164 pp.
Alliss, R.J., Raman, S and Chang, S.W. 1992. “Special Sensor Microwave /Imager
(SSM7I) Observations o f Hurricane Hugo (1989)”. M o n th ly W e a th e r R e v ie w , 120:
2723-2737.
Andres, L., Wiliams, S.A. and Skole, D. 1994. “Fourier Analysis o f Multi-temporal
AVHRR Data Applied to a Land Cover Classification”. In te r n a tio n a l J o u r n a l o f
R e m o te S e n s in g , 1 5 ,(5 ): 1115-1121.
Baier, W., Dyer, J,A., and Sharp, W R . 1979. The Versatile Soil Moisture Budget.
Technical Bulletin No. 87, Agriculture Canada, 52 pp.
Bailey, J.O., Barrett, E.C. and Kidd, C. 1986. “Satellite Passive Microwave Imagery and
it’s Potential for Rainfall Estimation Over Land”. J o u r n a l o f th e B r itis h
In te r p la n e ta r y S o c ie ty , 3 9 : 527-534.
Barrett, E.C., Kidd, C. and Bailey, J.O. 1988. “The Special Sensor Microwave Imager A
New Instrument with Rainfall Monitoring Potential”. In te r n a tio n a l J o u r n a l o f
R e m o te S e n s in g , 9,(12): 1943-1950.
Barton, I. J. 1978. “A Case Study o f Microwave Radiometer Measurements Over Bare
and Vegetated Surfaces”. J o u r n a l o f G e o p h y s ic a l R e s e a r c h , I: 3515-3517.
Becker, F. and Choudhury, B.J. 1988. ‘Relative Sensitivity of Normalized Difference
Vegetation Index and Microwave Polarization Difference for Vegetation and
Desertification Monitoring”. R e m o te S e n s in g o f E n v iro n m e n t, 2 4 : 297-311.
Blanchard, B.J., McFarland, M.J., Schmugge, T.J. and Rhodes, E. 1981. “Estimation of
Soil Moisture with API Algorithms and Microwave Emission”. W a te r R e s o u rc e s
B u lle tin , 1 7 ,(5 ): 767-774.
Blyth, K. 1993. “The Use of Microwave Remote Sensing to Improve Spatial
Parameterization of Hydrological Models”. J o u r n a l o f H y d r o lo g y , 1 5 2 : 103-129.
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Brown, R.J. 1994. Research Scientist, Canada Centre for Remote Sensing. Personal
Commnication.
Brunfeldt, D.R,. and Ulaby, F.T. 1986. “Microwave Emission from Row Crops”. IE E E
T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 2 4 ,(3 ): 353-359.
Canada Centre for Remote Sensing. 1994. Altona Soil Moisture Experiment Database
Report- Hydrology Group. 26pp
Campbell, W.J., Gloersen, P., Josberger, E.G., Johannessen, O.M., Guest, P.S., Mognard,
N., Shuchman, R., Bums, B.A., Lannelogue, N. and Davidson, K.L. 1987.
“Variations o f Mesoscale and Large-Scale Sea Ice Morphology in the 1984
Marginal Ice Zone Experiment as Observed by Microwave Remote Sensing”.
J o u r n a l o f G e o p h y s ic a l R e s e a r c h , 9 2 ,(C7): 6805-6824.
Campbell, JJR. 1987. Introduction to Remote Sensing. The Guilford Press, New York.
551 pp.
Cavalieri, D.J., Gloersen, P. and Campbell, W J. 1984. “Determination o f Sea Ice
Parameters with Nimbus 7 SMMR”. J o u r n a l o f G e o p h y s ic a l R e s e a r c h , 8 9 : 53555369.
Cihlar, J. 1987. “Environmental Factors Influencing Daytime and Nighttime Thermal
Infrared Images”. C a n a d ia n J o u r n a l o f R e m o te S e n s in g , 1 3 ,(1 ): 31-38.
Cihlar, J. and Huang, F. 1994. “Effect of Atmospheric Correction and Viewing Angle
Restriction on AVHRR Data Composites”. C a n a d ia n J o u r n a l o f R e m o te S e n s in g ,
2 0 ,(2 ): 132-137.
Cihlar, J., Hall, F., Goodison, B., Ranson, J., Sellers, P., Baldocchi, D., Margolis, H., den
Hartog, G., Crill, P., Kelly, R., Lettenmeier, D. and Ryan, M. 1995. “The Boreal
Ecosystem-Atmosphere Study (BOREAS)”. 1 7 th C a n a d ia n S y m p o siu m o n R e m o te
S e n s in g , Saskatoon, Saskatchewan, pp 168.
Chang, A.T.C., Foster, JX., and Hall, DiC. 1987. “NIMBUS-7 SMMR Derived Global
Snow Cover Parameters”. A n n a ls o f G la c io lo g y , 9 : 39-44
Chapman, L.J. and Putnum, D.F. 1984. The Physiography of Southern Ontario. 2nd
Edition. Ontario Research Foundation: University of Toronto Press. 386 pp.
Choudhury, B.J., Schmugge, T.J. Chang, A and Newton, R.W. 1979. “Effect of Surface
Roughness on the Microwave Emission from Soils”. J o u r n a l o f G e o p h y s ic a l
R e s e a r c h , 8 4 ,(C9): 5699-5706.
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Choudhury, B J., Owe, M., Goward, S.N., Golus, R_E., Onnsby, J.P., Chang, A T . and
Wang, JJR. 1987. “Quantifying Spatial and Temporal Variabilities o f Microwave
Brightness Temperature Over the U.S. Southern Great Plains”. In te r n a tio n a l
J o u r n a l o f R e m o te S e n s in g , 5,(2): 177-191.
Choudhury, B J. and Tucker, C.J. 1987. “Monitoring Global Vegetation using Nimbus-7
37 GHz Data Some Empirical Relations". In te r n a tio n a l J o u r n a l o f R e m o te
S e n s in g , 5,(7): 1085-1090.
Choudhury, B.J. and Golus, R-E. 1988. “Estimating Soil Wetness Using Satellite Data”.
In te r n a tio n a l J o u r n a l o f R e m o te S e n sin g , 9 ,(7 ): 1251-1257.
Choudhury, B.J. 1988. “Microwave Vegetation Index: A new Long-Term Global Data Set
for Biospheric Studies”. In te r n a tio n a l J o u r n a l o f R e m o te S e n sin g , 9,(2): 185-186.
Choudhury, B.J. 1989. “Monitoring Global Land Surface Using Nimbus-7 37 GHz Data
Theory and Examples”. In te r n a tio n a l J o u r n a l o f R e m o te S e n sin g , 70,(10): 15791605.
Choudhury, B.J., Wang, JR ., Hsu, AY., and Chen, Y.L. 1990. “Simulated and Observed
37 GHz Emission over Africa”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g
Choudhury, B.J. 1990. ‘Monitoring Arid Lands Using AVHRR-Observed Visible
Reflectance and SMMR-37 GHz Polarization Difference”. In te r n a tio n a l J o u r n a l
o f R e m o te S e n s in g , 77,(10): 1949-1956.
Choudhury, B. 1991. “Passive Microwave Sensing Contribution to Hydrological
Variables”. S u r v e y s in G e o p h y s ic s, 1 2 : 63-84.
Choudhury, B.J. and DiGirolamo, N.E. 1994. “Relations Between SMMR 37GHz
Polarization Difference and Rainfall, and Atmospheric Carbon Dioxide
Concentration”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 75,(18): 3643-3661.
Cole, F.W. 1970. Introduction to Meteorolpv. John Wiley and Sons, Toronto. 388pp.
Collier, P.A., Runacres, M.E. and McClatchey, J. 1989. ‘Mapping Very Low Surface
Temperature in the Scottish Highlands using NOAA AVHRR Data”. In te r n a tio n a l
J o u r n a l o f R e m o te S e n s in g , 1 0 : 1519-1529.
Coppo, P., Luzi, G., Paloscia, S., and Pampaloni, P. 1991. “Effect of Soil Roughness on
Microwave Emission: Comparison Between Experimental Data and Models”.
Proceedings, IGARSS ‘91, Espoo, Finland. 1167-1170.
Curran, P.J. 1987. “Review Article: Remote Sensing Methodologies and Geography”.
In te r n a tio n a l J o u r n a l o f R e m o te S e n sin g , 5,(5): 1255-1275.
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dobson, M.C., Kouyate, F. and Ulaby, F.T. 1984. “A Reexamination o f Soil Textural
Effects on Emission and Backscattering”. IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d
R e m o te S e n s in g , 2 2 ,(6 ): 530-536.
Dobson, M.C., Ulaby, F.T., Hallikainen, M.T. and El-Rayes, M.A. 1985. “Microwave
Dielectric Behaviour o f Wet Soils II: Dielectric Mixing Models”. IE E E
T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 23,(1): 35-46.
Drinkwater, MJL, Crawford, J_P., and Cavalieri, D J. 1991. “Multi-Frequency, MultiPolarization SAR and Radiometer Sea Ice Classification”. P r o c e e d in g s IG A R S S
‘9 1 , Espoo, Finland, pp. 107-111.
Dumanski, J., Coote, D R., Lucuik, G. and Lok, C. 1986. “Soil Conservation in Canada”.
J o u r n a l o f S o il a n d W a te r C o n se rv a tio n , 4 4 : 431-434.
Ecological Stratification Working Group. 1993. Ecoregions of Canada. Provisional Map.
Scale 1:7500000. Agriculture Canada Research Branch, Centre for Land and
Biological Resources Research and Environment Canada. State o f the
Environmnet Reporting, Ottawa.
Ehrlich, D., Estes, J.E. and Singh, A 1994. “Applications of NOAA-AVHRR 1km Data
for Environmental Monitoring”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 75,(1):
145-161.
Engman, E.T. 1992. “Soil Moisture Needs in Earth Science”. P r o c e e d in g s IG A R S S ’ 9 2 ,
Houston, Texas, pp. 477- 479.
Engman, E.T. 1991. “Applications of Microwave Remote Sensing of Soil Moisture for
Water Resources and Agriculture”. R e m o te S e n s in g o f E n v ir o n m e n t, 3 5 :213-226.
Engman, E.T and Chauhan, N. 1995. “Status o f Microwave Soil Moisture Measurements
with Remote Sensing”. R e m o te S e n s in g o f E n v iro n m e n t, 5 1 : 189-198.
Environment Canada. 1990. Eastern Canadian Boreal and Sub-Arctic Wetlands: A
Resource Document. 169 pp.
Environment Canada. 1951-1980. Canadian Climate Normals: Temperature and
Precipitation; Ontario. 254 pp.
Environment Canada. 1951-1980. Canadian Climate Normals: Temperature and
Precipitation; Prairie Provinces. 429 pp.
Eom, H.J. 1992. “A Thermal Microwave Emission Model for Row-Structured
Vegetation”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 73,(16): 2975-2982.
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Estes, J. E. 1966. “Some Applications o f Aerial Infrared Imagery”. A n n a ls o f th e
A s s o c ia tio n o f A m e ric a n G e o g ra p h e rs , 5 6 ,(4 ): 673-691.
Estes, JE ., Mel, M R . and Hooper, J .0 .1977. “Measuring Soil M oisture with an Airborne
Imaging Passive Microwave Radiometer”. P h o to g r a m m e tric E n g in e e r in g a n d
R e m o te S e n s in g , ■/.?,(10): 1273-1281.
Fedosejevs, G. 1995. Research Scientist, Canada Centre for Remote Sensing. Personal
Communication.
Ferraro, R JL, Grody, N.C and Kogut, J.A. 1986. “Classification o f Geophysical
Paramaters Using Passive Microwave Satellite Measurements”. IE E E
T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 2 4 ,(6 ): 1008-1013.
Giddings, L. and Choudhury, B.J. 1989. “Observation of Hydrological Features with
Nimbus-7 37 GHz Data, Applied to South America”. In te r n a tio n a l J o u r n a l o f
R e m o te S e n s in g , 1 0 : 1673-1686.
Goward, S.N., Compton, C.J. and Dye, D.G. 1985. “North American Vegetation Patterns
Observed with the NOAA-7 Advanced Very High Resolution Radiometer”.
V e g e ta tio , 6 4 : 3-14.
Grody, N.C. 1991. “Classification o f Snow Cover and Precipitation Using the Special
Sensor Microwave Imager”. J o u r n a l o f G e o p h y s ic a l R e s e a r c h , 9 6 ,(D4): 74237435.
Hall, F.G, Townshend, J.R. and Engman, E.T. 1995. “Status o f Remote Sensing
Algorithms for Estimation of Land Surface State Parameters”. R e m o te S e n s in g o f
E n v ir o n m e n t, 5 1 : 138-156.
Heymsfield, G M . and Fulton, R. 1992. ‘M odulation o f SSMZI Microwave Soil Radiances
by Rainfall”. R e m o te S e n s in g o f E n v ir o n m e n t, 3 9 : 187-202.
Holben, B.N. 1986. “Characteristics o f Maximum-Value Composite Images from
Temporal AVHRR Data”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 7,(11): 14171434.
Hollinger, J., Lo. R., Poe, G., Savage, R., and Peirce, J. 1987. Special Sensor
Microwave/Imager User’s Guide. Naval Research Laboratory, Washington, D.C.
120 pp.
Hollinger, J. 1989. DMSP Special Sensor Microwave/Imager Calibration/Validation Final
Report Volume I. Naval Research Laboratory, Washington, D.C.
Hollinger, J. 1991. DMSP Special Sensor Microwave/Imager Calibration/Validation Final
Report Volume II. Naval Research Laboratory, Washington, D.C.
137
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Hollinger, J.P., Pierce, J i . and Poe, G.A. 1990. “SSM/I Instrument Evaluation". IE E E
T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 2 8 ,(5 ): 781-789.
r unt, C.B. 1974. Natural Regions o f the United States and Canada. W JL Freeman and
Company, San Francisco. 725 pp.
Hunt, G.R. “Electromagnetic Radiation: The Communication Link in Remote Sensing”. In
Remote Sensing in Geology. Siegal, B.S. and Gillespie, AJL [Eds.]. Toronto: John
Wiley and Sons.
Jackson, T.J., Schmugge, T.J. and Wang, JJL 1982. “Passive Microwave Sensing of Soil
Moisture Under Vegetation Canopies”. W a te r R e s o u r c e s R e s e a r c h , 1 8 ,(4): 11371142.
Jackson, T.J., Schmugge, T J and O’Neill, P.E. 1984. “Passive Microwave Remote
Sensing o f Soil Moisture from an Aircraft Platform ”. R e m o te S e n s in g o f
E n v ir o n m e n t, 1 4 : 135-151.
Jackson, T.J. and O’Neill, P.E. 1986. “Microwave Dielectric Model for Aggregated
Soils”. IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 2 4 ,(6 ) : 920-929.
Jackson, T.J., and O’Neill, P.E. 1987. “Temporal Observations o f Soil Moisture Using a
Passive Microwave Sensor” . R e m o te S e n s in g o f E n v ir o n m e n t, 2 1 : 281-296.
Jackson, T.J. and O’Neill, P.E. 1990. “Attenuation o f Soil Microwave Emission by Com
and Soybeans at 1.4GHz and 5 GHz”. IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d
R e m o te S e n s in g , 2 8 ,(5 ): 978-980.
Jackson, T.J. and O’Neill, P_E. 1991. ‘Microwave Emission and Crop Residues”. R e m o te
S e n s in g o f E n v ir o n m e n t, 3 6 : 129-136.
Jackson, T.J. and Schmugge, T.J. 1991. “Vegetation Effects on the Microwave Emission
of Soils”. R e m o te S e n s in g o f E n v iro n m e n t, 3 6 : 203-212.
Jackson, T.J., Kostov, K.G. and Saatchi, S.S. 1992. “Rock Fraction Effects on the
Interpretation o f Microwave Emission from Soils”. IE E E T ra n sc a tio n s o n
G e o sc ie n c e a n d R e m o te S e n sin g , 3 0 ,(3 ): 630-616.
Jackson, T.J. 1993. “Measuring Surface Soil Moisture Using Passive Microwave Remote
S e n s in g '. H y d r o lo g ic a l P ro c e s se s , 7: 139-152.
Jackson, T.J., Le Vine, D M ., Swift, C.T., Schmugge, T.J., and Schiebe, R. 1995. “Large
Area Mapping o f Soil Moisture Using the ESTAR Passive Microwave Radiometer
in Washita ‘92”. R e m o te S e n s in g o f E n v iro n m e n t, 5 3 : 27-37.
138
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Jensen, JJL Introductory Digital Image Processing: A Remote Sensing Perspective.
Prentice-Hall, Toronto. 379 pp.
Kerr, Y U . and Njoku, E.G. 1993. “On the Use o f Passive Microwaves at 37GHz in
Remote Sensing o f Vegetation”. In te r n a tio n a l J o u r n a l o f R e m o te S e n sin g ,
1 4 ,(1 0 ): 1931-1943.
Klassen, R.W. 1969. Quaternary Stratigraphy and Radiocarbon Chronology in
Southwestern Manitoba. Paper 69-27, Department o f Energy, Mines amd
Resources.
Korporal, K., Dobbins, R., and Hillary, N M . 1989. “The Statistics Canada Crop
Condition Assessment Program”. Proceedings, IG A R S S ‘8 9 /1 2 * C a n a d ia n
S y m p o siu m o n R e m o te S e n sin g , Vancouver, Canada. 2006-2009.
Ladson, A R. and Moore, IT). 1992. “Soil Water Prediction on the Konza Prairie by
Microwave Remote Sensing and Topographic Attributes”. J o u r n a l o f H y d ro lo g y ,
1 3 8 : 385-407.
Le Vine, D M ., Wilheit, T.T., Murphy, BLE., and Swift, C.T. 1989. “A Multifreqency,
Microwave Radiometer for the Future". IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d
R e m o te S e n s in g , 2 7 : 193-199.
Lillesand, TM . and Kieffer, R.W. 1987. Remote Sensing and Image Interpretation. John
Wiley and Sons. New York. 721pp
Linsley, RJC. and Franzini, J.B. 1979. Water Resources Engineering. McGraw-Hill,
Toronto. 716 pp.
Lo, C.P. 1986. Applied Remote Sensing. Longman Group, New York. 393pp.
Loveland, T iL , Merchant, J.W., Ohlen, D.O. and Brown, J.F. 1991. “Development o f a
Land-Cover Characteristics Database for the
Conterminous U.S”.
P h o to g r a m m e tr ic E n g in e e r in g a n d R e m o te S e n s in g , 57,(11): 1453-1463.
Lopez, S., Gonzales, F., Llop, R. and Cuevas, JM . 1991. “An Evaluation of the Utility of
NOAA AVHRR images for Monitoring Forest Risk in Spain”. In te r n a tio n a l
J o u r n a l o d R e m o te S e n sin g , 1 2 : 1841-1851.
Luzi, G., Paloscia, S. and Pampaloni, P. 1990. Microwave Radiometry for Monitoring
Agricultural Crops. Irr Applications of Remote Sensing in Agriculture. MT>.
Steven and J.A. Clark [Eds.]. Toronto: Butterworths.
Malingreau, J.P. and Tucker, C.J., 1988. “Large Scale Deforestation in the Southeastern
Amazon Basin o f Brazil”. A m b io , 1 7 : 49-55.
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Malingreau, JJP. and Belward, A S . 1992. “Scale Considerations in Vegetation Monitoring
Using AVHRR Data”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , J5,(12):22892307.
Malingreau, J.P., Tucker, C.J. and Laporte, N. 1989. “AVHRR for Monitoring Global
Tropical Deforestation”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 1 0 : 855-867.
McDermid, G.J., Peddle, D R ., LeDrew, EJF. 1995. “An Overview o f BOREAS Remote
Sensing Research at the Earth-Observation Laboratory, University o f Waterloo”.
1 7 th C a n a d ia n S y m p o siu m o n R e m o te S e n s in g , Saskatoon, Saskatchewan.
McFarland , M.J., Wilke, G.D. and Harper, P.H. 1987. “Nimbus-7 SMMR Investigation
o f Snowpack Properties in the Northern Great Plains for the Winter o f 1978-79”.
TF .F .F T r a n sc a tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 2 5 : 35-46.
McFarland, M.J., Miller, RX. and Neale, C M . 1990. “Land Surface Temperature Derived
from SSM/I Passive Microwave Brightness Temperatures”. IE E E T r a n s a c tio n s o n
G e o sc ie n c e a n d R e m o te S e n s in g , 2 8 ,(5 ) : 839-844.
Microsoft Corporation. 1994. Microsoft Excel Version 4.0: User’s Guide. Microsoft
Corporation, Toronto.
Nagler, T. and Rott, H. 1992. “Development and Intercomparisons o f Snow Mapping
Algorithms Based on SSM7I Data”. P ro c e e d in g s , IG A R S S ‘9 2 , Houston, Texas.
812-814.
Neale, C.M., McFarland, M J. and Chang, K. 1990. “Land-Surface-Type Classification
Using Microwave Brightness Temperatures from the Special Sensor
Microwave/Imager”. IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g ,
2 8 ,(5 ): 829-837.
Newton, R W . and Rouse, J.W. 1980. “Microwave Radiometer Measurements o f Soil
Moisture Content”. IE E E T r a n s a c tio n s o n A n te n n a s a n d P r o p a g a tio n , 2 8 ,(5 ):
680-686.
Nghiem, S.V., Veysoglu, M X., Kong, J.A., Shin, R T ., O’Neill, K.O. and Lohanick, A.W.
1991. “Polarimetric Passive remote Sensing of a Periodic Soil Surface: Microwave
Measurements and Analysis”. J o u r n a l o f E le c tr o m a g n e tic W a v e s a n d
A p p lic a tio n s , 5,(9): 997-1005.
Njoku, E.G. and Kong, J. 1977. “Theory for Passive Microwave Remote Sensing o f NearSurface Soil Moisture”. J o u r n a l o f G e o p h y s ic a l R e s e a r c h , 8 2 ,(20): 3108-3118.
O’Neill, P.E. and Jackson, T.J. 1990. “Observed Effects o f Soil Organic M atter Content
on the Microwave Emissivity o f Soils”. R e m o te S e n s in g o f E n v ir o n m e n t, 3 1 : 175182.
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
O’Neill, P E . “Microwvae Remote Sensing o f Soil Moisture: A Comparison o f Results
from Different Truck and Aircraft Platform s”. In te r n a tio n a l J o u r n a l o f R e m o te
S e n s in g , 6 ,(7 ): 1125-1134.
Owe, M., Chang, A. and Golus, R.E. 1988. “Estimating Surface Soil Moisture from
Satellite Microwave Measurements and a Satellite Derived Vegetation Index”.
R e m o te S e n s in g o f E n v ir o n m e n t, 2 4 : 331-345.
Owe, M., Van De Griend, A A and Chang, A T . 1992. “Surface Moisture and Satellite
Microwave Observations in Semiarid Southern Africa”. W a te r R e s o u r c e s
R e s e a r c h , 2 8 ,(3 ): 829-839.
Paine, D.P. 1981. Aerial Photography and Image Interpretation for Resource
Management. John Wiley and Sons, New York. 571 pp.
Paloscia, S. and Pampaloni, P. 1992. “Microwave vegetation Indexes for Detecting
Biomass and W ater Conditions o f Agricultural Crops”. R e m o te S e n s in g o f
E n v ir o n m e n t, 4 0 : 15-26.
Paloscia, S., Pampaloni, P., Chiarantini, L., Coppo, P., Gagliani, S. and Luzi, G. 1993.
“Multifrequency Passive Microwave Remote Sensing of Soil Moisture and
Roughness”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 74,(3): 467-483.
Pampaloni, P., Paloscia, S., Chiarantini, L., Coppo, P., Gagliani, S., and Luzi, G. 1990.
“Sampling Depth o f Soil Moisture Content by Radiometric Measurements at 21
cm Wavelength: Some Experimental R esults”. In te r n a tio n a l J o u n r a l o f R e m o te
S e n s in g , 1 1 : 1085-1092.
Parkinson. C.L. 1992. “Spatial Patterns of Increases and Decreases in the Length o f the
Sea Ice Season in the North Polar Region, 1979-1986”. J o u r n a l o f G e o p h y s ic a l
R e s e a r c h , 9 7 ,(C 9): 14377-14388.
PCI Inc. 1993. EASI/PACE Versions 5.2 and 5.3. PCI Inc., Toronto.
Ph.D. Associates Inc. 1995. Geographic Analysis and Display System version 3.0: User’s
Guide. Ph.D. Associates Inc., North York.
Price, J. 1989. “Estimating Evapotranspiration Over Large Areas". P ro c e e d in g s, IG A R S S
‘8 9 /1 2 th C a n a d ia n S y m p o siu m o n R e m o te S e n s in g , Vancouver, Canada. 21382141.
Protz, R., Wood, D., Duke, C. and McNaim, H. 1993. A comprehensive Agricultural
Monitoring System - The Need for Soil M ositure Measurements. Proceedings o f
the ISTS Workshop on Ground Based Microwave Radiometry for Snow Cover
and Soil Moisture. University of Toronto Institute for Aerospace Studies. North
York, Ontario.
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Raddatz, RX. 1989. “An Operational Agrometeorological Information System for the
Canadian Prairies”. C lim a to lo g ic a lB u lle tin , 2 3 ,(3 ): 83-97.
Raddatz, RX. 1993. “Prairie Agroclimate Boundary-Layer Model: A Simulation o f the
Atmosphere/Crop-Soil Interface”. A tm o sp h e re -O c e a n , 57,(4): 399-419.
Riley, JX. 1982. “Hudson Bay Lowland Floristic Inventory, Wetlands Catalogue and
Conservation Strategy”. N a tu r a lis te C a n a d a , 1 0 9 : 543-555
Robertson, B., Erickson, A., Friedel, J., Guindon, B., Fisher, T., Brown, R., Teillet, P.,
D’lorio, M., Cihlar, J. and Sanz, A. 1992. “GEOCOMP, a NOAA AVHRR
Geocoding and Compositing System”. P ro c e e d in g s , IS P R S C o n fe r e n c e
C o m m issio n 2 , Washington, D.C. pp.231-228.
Sabins, F.F. 1978. Remote Sensing Principles and Interpretation. W.H Freeman and
Company, San Fransisco. 426pp.
Saxton, KX. and Lenz, A.T. 1967. “Antecedent Retention Indexes Predict Soil
Moisture”. J o u r n a l o f th e H y d ra u lic s D iv is io n , P r o c e e d in g s o f th e A m e r ic a n
S o c ie ty o f C iv il E n g in e e r s , 9 3 , (HY4): 223-241.
Schanda, E. 1976. Passive Microwave Sensing. In Remote Sensing for Environmental
Sciences. E. Schanda [Ed.] Springer-Verlag. New York. 187-256.
Schmugge, T.J., Meneely, J.M., Rango, A., and Neff, R. 1977. “Satellite Microwave
Observations of Soil Moisture Variations”. W a te r R e s o u r c e s B u lle tin , 1 3 : 265.
Schmugge, T.J., Jackson, T.J. and McKim, HX . 1980. “Survey of Methods for Soil
Moisture Determination”. W a te r R e s o u rc e s R e s e a r c h , 1 6 ,(6 ): 961-979.
Schmugge, T.J. 1980. “Effect o f Texture on Microwave Emission from Soils”. IE E E
T r a n sa c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 1 8 ,(4): 353-361.
Schmugge, T.J. 1985. Remote Sensing of Soil Mositure. In: Anderson, M.G. and Burt,
T P. [Eds.] Hydrological Forecasting. John Wiley and Sons, New York.
Schmugge, T.J., O’Neill, P.E. and Wang, JR . 1986. “Passive Microwave Soil Moisture
Research”. IE E E T r a n s a c tio n s o n m G e o sc ie n c e a n d R e m o te S e n s in g , 2 4 ,(1): 1222.
Schmugge, T.J., Wang, JR ., Asrar, G. 1988. ‘R esults from the Push Broom Microwave
Radiometer Flights Over the Konza Prairie in 1985”. IE E E T r a n sa c tio n s o n
G e o sc ie n c e a n d R e m o te S e n sin g , 2 6 : 590-596.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Schmugge, T.J. and Jackson, T.J. 1992. “A Dielectric Model o f the Vegetation Effects on
the Microwave Emission from Soils”. IE E E T r a n sa c tio n s o n G e o sc ie n c e a n d
R e m o te S e n s in g , 50,(4): 757-760.
Schmugge, T.J., Jackson, T.J., Kustas, W.P. and Wang, JR . 1992. “Passive Microwave
Remote Sensing o f Soil Moisture: Results from HAPEX, FIFE and MONSOON
90”. IS P R S J o u r n a l o fP h o to g r a m m e tr y a n d R e m o te S e n s in g , 4 7 : 127-143.
Schmugge, T.J., Jackson, T.J., Kustas, WJP., Roberts, R., Parry, R., Goodrich, D.C.,
Amer, S.A and Weltz, M A 1994. “Push Broom Microwave Radiometer
Observations o f Surface Soil Moisture in Monsoon c90”. W a te r R e s o u rc e s
R e s e a r c h , 30,(5): 1321-1327.
Shuman, C .A , Alley, R.B., Anandakrishnam, S. and Steams, C.R. 1995. “An Empirical
Technique for Estimating Near-Surface Temperature Trends in Central Greenland
from SSM/I Brightness Temperatures”. R e m o te S e n s in g o f E n v ir o n m e n t, 5 1 : 245252.
Sihvola, A H. and Alanen, E. 1991. “Studies o f Mixing Formula in the Complex Plane”.
IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 2 9 ,(4 ): 679-687.
Smith, R.C. and Choudhury, B.J. 1990. “On the Correlation o f Indices o f Vegetation and
Surface Temperature over South-Eastern Australia”. In te r n a tio n a l J o u r n a l o f
R e m o te S e n s in g , 1 1 : 2113-2120.
Smith, E .A and Choudhury, B.J. 1993. “The Relationship Between Anomlies of
Interannual Net Radiation and 37 GHz Polarization Difference Over North
America”. C lim a tic C h a n g e , 2 3 : 141-154.
Spencer, R 1986. “A Satellite 37-GHz Scattering-based Method for Measuring Oceanic
Rain Rates”. J o u r n a l o f C lim a te a n d A p p lie d M e te o r o lo g y , 2 5 : 754-766.
Stanford, Q. 1977. The New Canadian Oxford Atlas: Mean Annual Precipitation. Oxford
University Press, Don Mills. 172pp.
Statistics Canada. 1994. Canada Yearbook. Supply and Services Canada, Ottawa.
StatSoft Inc. 1994. Statistica Version 4.0: User’s Guide Volume 1. StatSoft Inc.
Oklahoma.
Steffen, K. and Schweiger, A J. 1990. “A Multisensor Approach to Sea Ice Classification
for the Validation o f DMSP-SSM7I Passive Microwave Derived Sea Ice
Products”. P h o to g ra m m e tric E n g in e e r in g a n d R e m o te S e n s in g , 56,(1): 75-82.
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Stephens, P., MacMillan, J., Daigle, J., and Cihlar, J. 1985. “Estimating Universal Soil
Loss Equation Factor Values with Aerial Photography”. J o u r n a l o f S o il a n d W a te r
C o n s e r v a tio n , 4 0 : 293-296.
Suits, G.H. 1983. “The Nature o f Electromagnetic Radiation”. In Manual o f Remote
Sensing. R.N. Colwell [Ed.]. Falls Church, Virginia: American Society o f
Photogrammetry and Remote Sensing, Volume 1, pp37-60.
Teillet, P_M., Slater, P.N., Ding Y., Santer, R_P., Jackson, R.D., and Moran. M.S. 1990.
“Three Methods for the Absolute Calibration of NOAA AVHRR Sensors in
Flight”. R e m o te S e n s in g o f E n v ir o n m e n t, 3 1 : 105-120.
Teller, J.T. and Fenton, M F. 1980. “Late Wisconsin Glacial Stratigraphy and History o f
Southeastern Manitoba”. C a n a d ia n J o u r n a l o f E a r th S c ie n c e , 1 7 : 19-35.
Teng W l., W ang J.R. and Doraiswamy, P.C. 1993. “Relationship Between Satellite
Microwave Radiometric Data, Antecedent Precipitation Index, and Regional Soil
Moisture”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 1 4 ,{ 13): 2483-2500.
Teng, W l., Doraiswamy, P.C. and Wang, J.R. 1995. “Temporal Variations o f the
Microwave Polarization Difference Index and its Relationship to the Normalized
Difference Index in a Densely Cropped Area”. P h o to g r a m m e tric E n g in e e r in g a n d
R e m o te S e n s in g , 6 1 ,{8): 1033-1040.
Theis, S.W., Blanchard, B.J. and Newton, R.W. 1984. “Utilization o f Vegetation Indices
to Improve Soil Moisture Estimates Over Agricultural Lands”. IE E E T r a n s a c tio n s
o n G e o s c ie n c e a n d R e m o te S e n s in g , 2 2 ,(6): 490-495.
Theis, S.W. and Blanchard, A J. 1988. “The Effect o f Measurement Error and Confusion
from Vegetation on Passive Microwave Estimates of Soil Moisture”. In te r n a tio n a l
J o u r n a l o f R e m o te S e n sin g , 9,(2): 333-340.
Toure, A , Condial, A R . and Brown, R.J. 1991. “Analysis of NOAA-9 Vegetation Index
and Temperature Time Series o f Wheat Fields in Western Canada”. C a n a d ia n
J o u r n a l o f R e m o te S e n sin g , 77,(1): 21-26.
Townshend, J.R., Justice, C.O. and Kalb, V. 1987. “Characterization and Classification of
South American Land Cover Types Using Satellite D ata”. In te r n a tio n a l J o u r n a l
o f R e m o te S e n s in g , 5,(8): 1189-1207.
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Townshend, JR ., Choudhury, B.J., Giddings, L., Justice, C.O., Prince, SJ). and Tucker,
C.J. 1989. “Comparison o f Data from the Scanning Multifrequency Microwave
Radiometer (SMMR) with Data from the Advanced Very High Resolution
Radiometer (AHVRR) for Terrestrial Environment Monitoring: An Overview”.
In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 70,(10): 1687-1690.
Tucker, C.J. and Gatlin, J.A. 1984. “Monitoring Vegetation in the Nile Delta with NOAA6 and NOAA-7 AVHRR Imagery”. P h o to g r a m m e tr ic E n g in e e r in g a n d R e m o te
S e n s in g , 50,(1): 53-61.
Tucker, C.J., Townshend, R.G. and Goff, T.E. 1985. “African Land-Cover Classification
Using satellite Data”. S c ie n c e , 227,(4685): 369-375.
Tucker, C.J. and Choudhury, B.J. 1987. “Satellite Remote Sensing of Drought
Conditions”. R e m o te S e n s in g o f E n v iro n m e n t, 2 3 : 243-251.
Ulaby, F.T., Moore, RiC. and Fung, AJC. 1981-1986. Microwave Remote Sensing:
Active and Passive: Vol. 1-3. Artech House, Norwood, Massachussets.
Ulaby, F.T., Razani, M. and Dobson, M.C. 1983. “Effects o f Vegetation Cover on the
Microwave Radiometric Sensitivity to Soil Moisture”. IE E E T r a n sa c tio n s o n
G e o sc ie n c e a n d R e m o te S e n sin g , 27,(1): 51-61.
Van De Griend, A.A. and Owe, M. 1993. “Determination o f Microwave Vegetation
Optical Depth and Single Scattering Albedo from Large Scale Soil Moisture and
Nimbus/SMMR Satellite Observations”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g ,
74,(10): 1875-1886.
Vyas, A.D., Trivedi, A.J., Calla, O.P., Rana, S.S. and Raju, G. 1985. ‘Tassive Microwave
Remote Sensing o f Soil Moisture”. In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g ,
6,(7): 1153-1162.
Vyas, A.D., Trivedi, A.J., Calla, O.P., Rana, S.S. and Sharma, S B. 1988. “Remote
Sensing o f Soil Moisture Content Over Bare Soil at Microwave Frequencies”.
In te r n a tio n a l J o u r n a l o f R e m o te S e n s in g , 9,(2): 341-347.
Vizy, K.N. 1974. “Detecting and Monitoring Oil Slicks with Aerial Photos”.
P h o to g r a m m e tr ic E n g in e e r in g , 4 0 ,(6 ): 697-708.
Walker, A. 1995. Research Scientist, Environment Canada. Personal Communication.
Wang, J.R., Newton, R W . and Rouse, JJL 1980. ‘Tassive Microwave Remote Sensing of
Soil Moisture: The effect of Tilled Row Structure”. IE E E T r a n sa c tio n s o n
G e o sc ie n c e a n d R e m o te S e n sin g , 1 8 ,(4 ) : 296-301.
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Wang, J.R. and Choudhury, B J. 1981. “Remote Sensing o f Soil Moisture Content Over
Bare Field at 1.4GHz Frequency”. J o u r n a l o f G e o p h y s ic a l R e s e a r c h , 8 6 ,(C6):
5277-5282.
Wang, J.R. 1983. “Passive Microwave Sensing o f Soil Moisture Content: The Effects o f
Soil Bulk Density and Surface Roughness”. R e m o te S e n s in g o f E n v ir o n m e n t, 1 3 :
329-344.
Wang, JR ., O’Neill, P.E., Jackson, T.J. and Engman, E. 1983. “Multifrequency
Measurements o f the Effects o f Soil Moisture, Soil Texture, and Surface
Riughness”. IE E E T r a n s a c tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 27,(1): 4450.
Wang, J.R. 1985. ‘E ffect o f Vegetation on Soil Moisture Sensing Observed from Orbiting
Microwave Radiometers”. R e m o te S e n s in g o f E n v ir o n m e n t, 1 7 : 141-151.
Wang, J.R. 1987. “Microwave Emission from Smooth Bare Fields and Soil Moisture
Sampling Depth”. IE E E T r a n sc a tio n s o n G e o sc ie n c e a n d R e m o te S e n s in g , 25,(5):
616-622.
Wang, J.R. 1992. “An Overview o f the Measurements o f Soil Moisture and Modelling o f
Moisture Flux in FIFE”. J o u r n a l o f G e o p h y s ic a l R e s e a r c h , 9 7 ,(D17); 1895518959.
Wegmuller, V. 1993. “Signature Research for Crop Classification by Active and Passive
Microwaves”. In te r n a tio n a l J o u r a n l o f R e m o te S e n s in g , 1 4 : 871-883.
Wentz, F.J. 1994. Research Sceintist. Personal Communication. May 1994.
Wigneron, J.P., Chanzy, A., Calvet, J.C. and Bruguier, N. 1995. “A Simple Algorithm to
Retrieve Soil Moisture aand Vegetation Biomass Using Passive Microwave
Measurements over Crop Fields”. R e m o te S e n s in g o f E n v iro n m e n t, 5 1 : 331-341.
Wilke, G.D. and McFarland, M.J. 1986. “Correlations Between Nimbus-7 Scanning
Multichannel Microwave Radiometer and an Antecedent Precipitation Index”.
J o u r n a l o f C lim a te a n d A p p lie d M e te o r o lg y , 2 5 : 227-238.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX TABLES
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A1
Locational accuracies between SSMT and AVHRR
for each ecoregion pixel (1992)
1992
P ix el#
1
2
3
4
5
6
Accuracy - X
(m)
51.43
497.95
636.29
1.96
506.58
65.98
Accuracy - Y
(m )
41.38
90.54
492.16
389.50
44.23
318.34
Jam es Bay
Lowlands
1
2
3
4
92.01
282.37
250.45
327.31
190.63
62.22
597.35
644.70
Southern O ntario
1
2
3
4
5
6
7
8
9
10
266.60
446.65
262.67
332.01
273.41
269.23
84.94
261.95
258.29
145.45
200.67
524.12
276.46
537.72
614.64
32.82
70.34
621.48
360.94
4.45
Boreal
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A2
Locational accuracies between SSM/I and AVHRR
for each ecoregion pixel (1993)
Pixel#
1
2
3
4
5
6
Accuracy - X
(m)
270.00
298.33
177.80
230.24
506.58
2.10
Accuracy - Y
(m)
214.56
222.71
405.72
148.61
44.23
293.25
Jam es Bay
Lowlands
1
2
3
4
326.65
540.08
436.45
437.13
371.01
156.40
531.16
267.24
Southern Ontario
1
2
3
4
5
6
7
8
9
10
34.88
168.56
391.17
212.69
176.53
391.45
10.82
236.18
112.10
377.48
361.62
277.30
59.35
223.15
254.46
334.04
476.19
237.16
45.32
552.40
1993
Boreal
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A3
Locational accuracies between SSM/I and AVHRR
for each ecoregion pixel (1994)
1994
B oreal
Pixel#
1
2
3
4
5
6
Accuracy - X Accuracy - Y
(m)
(m)
270.00
214.56
298.33
222.71
177.80
405.72
649.28
12.86
227.17
75.00
40.50
447.45
Jam es Bay
Lowlands
1
2
3
4
326.65
440.71
436.45
218.38
371.01
381.48
531.16
713.13
Southern O ntario
1
2
3
4
5
6
7
8
9
10
34.88
168.56
391.17
212.69
176.53
391.45
10.82
236.18
112.10
377.48
361.62
277.30
59.35
223.15
254.46
334.04
476.19
237.16
45.32
552.40
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A4
Meteorological data for Environment Canada climate stations located
in the Hudson / James Bay Region (1988)
Moosenee 1988
Dec
April
Jul
Oct
Great W hale 1988
Dec
April
Jul
Oct
Earlton 1988
Dec
April
Jul
Oct
Churchill 1988
Dec
April
Jul
Oct
Means Dec
Means Apr
Means Jul
Means Oct
Mean Max Mean M in
Tem p(°C) Tem p(°C )
-24.6
-12.9
-6.5
4.1
23.9
8
-2.5
6.3
Mean
Total Rain Total Snow Total Prerip
(mm)
(mm)
(cm)
Tem p(°C )
44.9
38.1
-18.8
0
8.4
64.5
-1.2
34.2
0
116
16
116
78.1
1.9
45.2
372
Mean Max Mean M in
Temp (° C) Temp (° Q
-21.6
-13.3
-6.5
2.5
16.6
6.5
5.8
-0.5
Mean
Total Rain Total Snow Total Prerip
(mm)
Temp (° C)
(mm)
(cm)
-17.5
37.5
34.2
8.2
16.4
-2
8.2
11.6
0
52
52
2.7
75
51
24.2
Mean Max Mean M in
Tem p(°C) Temp (° C)
-7.1
-20.2
-2.8
7.3
12.8
26.8
-0.7
7.9
Mean
Total Rain Total Snow Total Prerip
Temp (° C)
(cm)
(mm)
(mm)
49.9
-13.6
2.8
48.1
2.3
93.6
13.9
110.5
0
34.6
19.8
34.6
3.6
55.5
10.2
65.4
Mean Max Mean M in
Temp (° C) Temp (° C)
-19.6
-27.9
-12.8
-5.4
6.1
16.8
-4.8
0.9
Mean
Total Rain Total Snow Total Precip
Temp (° C)
(mm)
(cm)
(mm)
-23.8
25.5
17.4
0
-9.2
0
66.8
65.2
11.5
0
48.4
48.4
69.6
-2
6.2
65.6
-13.23
2.13
21.03
5.23
-23.58
-7.15
8.35
-2.13
-18.43
-2.53
14.73
1.55
34.00
62.75
37.48
39.00
24.33
0.00
36.30
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34.90
64.15
62.75
72.03
Table A5
Meteorological data for Environment Canada climate stations
located in the Hudson / James Bay Region (1990)
M oosenee 1990
Dec
April
Jul
Oct
Great W hale 1990
Dec
April
Jul
Oct
Earlton 1990
Dec
April
Jul
Oct
Churchill 1990
Dec
April
Jul
Oct
M eans Dec
M eans Apr
M eans Jul
M eans Oct
Mean Max Mean Min
Tem p(°C) Temp (° C)
-10.5
-22.9
3.4
-8.8
9.9
23.8
7.1
-1.2
Total Rain Total Snow Total Precip
Mean
(mm)
Temp (° C)
(cm)
(mm)
-16.7
58.8
43.8
27
-2.7
13
36.2
16.8
66.5
0
66.5
37.4
3
23.8
612
Mean Max Mean Min
Temp (° C) T em p(°Q
-13
-22.7
-5
-15.9
16.3
5.6
-1.3
3.2
Total Rain Total Snow Total Precip
Mean
(mm)
(cm)
(mm)
Temp (° C)
36.9
-17.9
36.7
-10.4
4.2
26.2
27.9
0
10.9
108.2
108.2
26.3
38.2
63.9
1
Mean Max Mean Min
Temp (° C) Temp (° C)
-4.1
-22.5
10.4
-2.2
24.3
11.4
8.9
-0.5
Total Rain Total Snow Total Precip
Mean
(mm)
(cm)
Temp (° C)
(mm)
2.8
-10.5
40
42.2
4.7
4.1
27
29.7
17.9
58.4
0
58.4
4.2
118.5
4.7
122.2
Mean Max Mean Min
Temp (° C) Temp (° C)
-24
-32.8
-6.4
-15.2
18
8.4
-5.8
-0.2
Total Rain Total Snow Total Precip
Mean
Temp (° C)
(mm)
(cm)
(mm)
-28.4
0
42.3
33.5
0.6
-10.8
30.2
27.2
13.2
59
0
59
-3
32.6
67.6
91.8
-12.90
0.60
20.60
4.75
-25.23
-10.53
8.83
-2.20
-18.38
-4.95
14.70
1.30
9.13
73.03
53.70
44.50
24.10
0.00
33.58
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39.05
30.25
73.03
84.78
Table A6
Meteorological data for Environment Canada climate stations located
in the Hudson / James Bay Region (1992)
M oosenee 1992
Dec
April
Jul
Oct
Great W hale 1992
Dec
April
Jul
Oct
Earlton 1992
Dec
April
Jul
Oct
Churchill 1992
Dec
April
Jul
Oct
Means Dec
Means Apr
M eans Jul
Means Oct
Mean Max Mean Min
Temp (° C) Temp (° C)
-10.6
-20
2.1
-10.6
19.1
5.1
4.4
-1.6
Total Rain Total Snow Total Precip
Mean
(mm)
(mm)
(cm)
Temp(° C)
14.4
3.6
44.6
-15.3
43
9.6
51.2
-4.3
127.6
0
127.6
12.1
67.4
38.8
29
1.4
Mean Max Mean Min
Temp (° C) Temp (° C)
Total Rain Total Snow Total Precip
Mean
(cm)
(mm)
(mm)
Temp (° C)
-6.2
14.6
3.5
-17.9
4.3
-0.9
-12.1
9.5
1.3
5.2
76.8
25.9
19.4
0
34.2
23.4
76.8
56.3
Mean Max Mean Min
Temp (° C) Temp (° C)
-5
-15.6
6.6
-4.2
21.1
8.8
7.8
-1.1
Total Rain Total Snow Total Precip
Mean
(mm)
(mm)
(cm)
Temp (° C)
54.4
52.9
-10.3
2.8
47.4
49.6
2.2
1.2
103.8
0
103.8
15
5.1
54.1
49
3.3
Mean Max Mean Min
Temp (° C) Temp (° C)
-19.6
-26.7
-8.6
-18.2
12.6
2.3
-0.3
-5.7
Mean
Total Rain Total Snow Total Precip
(mm)
Temp (° C)
(mm)
(cm)
0
18.4
10.2
-23.1
17
11.8
-13.4
1.4
0
20
20
7.5
19.6
3.8
19.4
-3
•8.80
-1.53
16.85
3.85
-15.58
-12.73
5.13
-2.33
-12.18
-7.15
11.03
0.75
24.25
82.05
29.38
25.47
12.05
0.00
21.93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26.93
34.00
82.05
49.35
Table A7
Meteorological data from Environment Canada climate stations located
in the Boreal Region (1988)
Timmins 1988
Dec
April
Jul
Oct
Sudbury 1988
Dec
April
Jul
Oct
North Bay 1988
Dec
April
Jul
Oct
Kapus leasing 1988
Dec
April
Jul
Oct
Means Dec
Means A pr
Means Jul
Means Oct
Mean Max
Temp (° C)
-8.6
-7.2
26.8
7.1
Mean Min
Temp (° C)
-21.8
-3.9
12
-1.8
Mean
Temp (° Q
-15.2
1.7
19.5
2.7
Total Rain
(mm)
0.2
42.5
44
55.4
Total Snow
(cm)
68.9
17.6
0
19.1
Total Precip
(mm)
66.3
46.2
44
73.7
Mean Max
Temp (° C)
-5.2
8.3
28.3
7.7
Mean Min
Temp (° C)
-15.4
-1.7
15.5
0.4
Mean
Temp (° C)
-10.3
3.3
21.9
4.1
Total Rain
(mm)
16
66.6
29.8
121.4
Total Snow
(cm)
64
12
0
2.6
Total Precip
(mm)
66.2
77
19.8
124
Mean Max
Temp (° C)
-4.7
8.7
26.5
7.8
Mean Min
Temp (° C)
-15.4
-0.9
15
0.8
Mean
Temp (° C)
-10.1
3.9
20.8
4.3
Total Rain
(mm)
12.7
67.2
70
119.1
Total Snow
(cm)
73.4
12
0
10.6
Total Precip
(mm)
65.2
77
70
128.7
Mean Max
Temp (° C)
-11.1
7.3
26.1
6.6
Mean Min
Temp (° C)
-22
-4.6
11.8
-2.1
Mean
Temp (° C)
-16.6
1.4
19
2.3
Total Rain
(mm)
misg
26.6
57.6
40.3
Total Snow
(cm)
66
17.4
0
22.2
Total Precip
(mm)
61.6
42.8
57.6
60.9
-7.40
4.28
26.93
7.30
-18.65
-2.78
13.58
-0.68
-13.05
2.58
20.30
3.35
50.73
50.35
84.05
68.08
14.75
0.00
13.63
64.83
60.75
47.85
96.83
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A8
Meteorological Data from Environment Canada climate stations located
in the Boreal Region (1990)
Tim mins 1990
Dec
April
Jul
Oct
Sudbury 1990
Dec
April
Jul
Oct
North Bay 1990
Dec
April
Jul
Oct
K apuskasing 1990
Dec
April
Jul
Oct
M eans Dec
M eans Apr
M eans Jul
M eans Oct
Mean Max Mean Min
T em p(°Q Temp (° Q
-5.5
-19.1
9.7
A
24.1
10.7
7.5
-L I
Mean
Total Rain Total Snow Total Precip
(mm)
(cm)
(mm)
T em p(°C )
32.6
-12.3
73
63
8.6
20
28
2.9
174.2
17.4
174.2
0
158
128
3.2
30.4
Mean Max Mean Min
Temp (° Q Temp (° C)
-3.1
-12.6
10.5
-0.7
24.6
13
9.8
0.4
Mean
Total Rain Total Snow Total Precip
(mm)
Temp (° C)
(mm)
(cm)
54.4
-7.9
4.9
56.2
79.1
43
14.2
34.4
18.8
0
38.2
38.2
137
5.1
7.2
144.2
Mean Max Mean Min
Temp (° C) Tem p(°C)
-2.7
-12.7
9.4
0.4
23.8
13.2
10.2
1.2
Mean
Total Rain Total Snow Total Precip
(mm)
T em p (°Q
(mm)
(cm)
-7.7
7.6
52
55.8
4.9
22.4
27.8
48
18.5
0
127.2
127.2
131.7
5.7
125.3
7
Mean Max Mean Min
Temp (° C) Temp (° Q
-8.2
-20.7
7.7
-5.8
23.7
11.1
7.2
-1.5
Mean
Total Rain Total Snow Total Precip
(mm)
Temp (° C)
(mm)
(cm)
-14.4
39.7
6.6
39.5
0.9
26.8
39.6
66.5
17.4
80.3
0
80.3
2.8
57.1
113.1
45
-4.88
9.33
24.05
8.68
-16.28
-2.53
12.00
-0.25
-10.58
3.40
18.03
4.20
6.75
18.00
104.98
111.85
53.63
30.45
0.00
22.40
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44.68
55.40
104.98
136.75
Table A9
Meteorological Data from Environment Canada climate stations located
in the Boreal Region (1992)
Timmins 1992
Dec
April
Jul
Oct
Sudbury 1992
Dec
April
Jul
Oct
North Bay 1992
Dec
April
Jul
Oct
Kapuskasing 1992
Dec
April
Jul
Oct
M eans Dec
M eans Apr
M eans Jul
Means Oct
Mean Max Mean Min
Temp (° C) Temp C C)
-17
-6.1
6.4
-6
8
20.9
6.9
-13
Mean
Temp (° Q
-11.6
0.2
14.5
2.8
Mean Max Mean Min
Temp (° C) Temp (° C)
-3.6
-11.7
6.3
-2.7
20.6
10.2
8.4
0.5
Total Rain Total Snow Total Precip
Mean
(cm)
(mm)
Temp (° C)
(mm)
2.6
84
74.6
-7.6
45.7
1.8
41.7
4
0
84.5
15.4
84.5
4.5
52
12
65.2
Mean Max Mean Min
Temp (° C) Temp (° C)
-2.4
-11.1
5.9
-3.2
20.1
10.6
-0.1
8.5
Total Rain Total Snow Total Precip
Mean
(mm)
(cm)
(mm)
Tem p(°C)
3.8
54.4
43.2
-6.8
24.6
3.2
17.6
1.3
75.8
0
75.8
15.4
73.6
4.2
60.8
14.6
Mean Max Mean Min
Temp (° C) Temp (° C)
-8.4
-17.6
-6.7
5.2
20.8
8.6
5.6
-1.5
Mean
Total Rain Total Snow Total Precip
Temp (° C)
(mm)
(cm)
(mm)
77.8
100.3
-13
25.2
79
-0.7
67.6
9
14.7
110.2
0
110.2
91
43.8
44.2
2.1
-5.13
5.95
20.60
7.35
-14.35
-4.65
9.35
-0.60
-9.75
0.65
15.00
3.40
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
77
80.1
8.8
69.1
1.8
67.3
95.4
95.4
0
48.5
29.3
21.2
10.10
50.30
91.48
46.48
74.08
4.50
0.00
23.00
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73.78
52.85
91.48
69.58
Table A10
Meteorological Data from the Environment Canada climate stations located
in southern Ontario (1988)
W indsor 1988
Dec
April
Jul
Oct
W aterloo 1988
Dec
April
Jul
Oct
Toronto 1988
Dec
April
Jul
Oct
Sarnia 1988
Dec
April
Jul
Oct
Owen sound 1988
Dec
April
Jul
Oct
Mean
Mean Max Mean Min
Temp (° C) Temp (8 C) Temp (° C)
-5.3
-1.6
2.1
3
8.5
13.9
18.6
24.9
31.3
12.1
3.9
8
Total Rain
(mm)
34
53.3
51.8
89.8
Total Snow
(cm)
22.6
1
0
0.2
Total Precip
(mm)
54.8
54.1
51.8
90
Mean Max Mean Min
Temp (° C) Temp (° C)
-7.7
-0.1
10.6
0.9
14.5
28.8
1.6
10.5
Mean
Temp (° C)
-3.9
5.8
21.7
6.1
Total Rain
(mm)
36.9
58.6
223.2
102.1
Total Snow
(cm)
34
1
0
0.2
Total Precip
(mm)
63.3
58.8
223.2
102.5
Mean Max Mean Min
Temp (° C) Temp (° C)
-7.1
1.5
10.7
1.1
29.7
16.1
2.7
11.6
Mean
Temp (° C)
-2.8
5.9
22.9
7.2
Total Rain
(mm)
18.6
55.2
109.7
67.4
Total Snow
(cm)
15.6
0
0
0
Total Precip
(mm)
31.5
55.2
109.7
67.4
Mean Max Mean Min
Temp (° C) Temp (° C)
1.2
-5.9
11.3
1.9
16.5
29.3
3.4
11.5
Mean
Temp (° C)
-2.4
6.6
22.9
7.5
Total Rain
(mm)
19
59.2
105.7
86
Total Snow
(cm)
21.8
0.8
0
0
Total Precip
(mm)
39.6
60
105.7
86
Mean Max Mean Min
Temp (° C) Temp (° C)
-5.8
0.5
9.5
0.8
26.9
16.2
10.6
4.3
Mean
Temp (° C)
-2.7
5.1
21.6
7.5
Total Rain
(mm)
48.2
74
83.4
180
Total Snow
(cm)
104.6
4
0
4
Total Precip
(mm)
152.8
78
83.4
184
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A10 continued
Meteorological Data from the Environment Canada climate stations located
in southern Ontario (1988)
London 1988
Dec
April
Jul
Oct
Ham ilton 1988
Dec
April
Jul
Oct
Delhi 1988
Dec
April
Jul
Oct
Barrie 1988
Dec
April
Jul
Oct
Blyth 1988
Dec
April
Jul
Oct
M eans Dec
Means A pr
M eans Jul
M eans Oct
Mean Max Mean Min
Temp (° C) Temp (° C)
0.8
-6.3
11.5
1.6
29.3
16
10.8
2.5
Mean
Temp (° C)
-2.7
6.6
22.7
6.7
Total Rain
(mm)
39.9
55.3
120.2
124.1
Total Snow
(cm)
48.7
6.4
0
11.8
Total Precip
(mm)
74.4
62.5
120.2
138.3
Mean Max Mean Min
Temp (° C) Temp (° C)
-6.4
1.2
10.6
1.2
29.3
16.7
11.3
2.8
Mean
Temp (° C)
-2.7
5.9
23
7.1
Total Rain
(mm)
44.3
53.4
111.3
93.1
Total Snow
(cm)
28
I
0
0.2
Total Precip
(mm)
62.7
54.4
111.3
93.3
Mean Max Mean Min
Temp (° C) Temp (° C)
l.l
-6.6
11.9
2.1
29.8
15.5
10.8
2.9
Mean
Temp (° C)
-2.8
7
22.7
6.9
Total Rain
(mm)
35
71.8
181.4
102.6
Total Snow
(cm)
56.6
1.6
0
0
Total Precip
(mm)
91.6
73.4
181.4
102.6
Mean Max Mean Min
Temp (° C) Temp (° C)
-0.4
-8.4
Mean
Temp (° C)
-4.4
Total Rain
(mm)
16
Total Snow
(cm)
60
Total Precip
(mm)
76
22.5
6.5
87.3
104.8
0
13.4
87.3
118.2
Mean
Temp (° C)
-3.9
6.2
22.6
6.9
Total Rain
(mm)
42.5
73
62.7
193.8
Total Snow
(cm)
128.2
4.2
0
23.8
Total Precip
(mm)
170.7
77.2
62.7
217.6
-2.99
5.76
22.75
7.04
33.44
55.38
113.67
114.37
52.01
2.00
0.00
5.36
81.74
57.36
113.67
119.99
28.4
10.1
16.6
2.8
Mean Max Mean Min
Temp (° C) Temp (° C)
-0.6
-7.1
10.9
1.5
29.5
15.6
10.1
3.6
0.73
10.09
29.23
10.94
-6.66
1.41
16.23
3.05
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A ll
Meteorological Data fromthe Environment Canada climate stations located
in southern Ontario (1990)
W indsor 1990
Dec
April
Jul
Oct
Waterloo 1990
Dec
April
Jul
Oct
Toronto 1990
Dec
April
Jul
Oct
Sarnia 1990
Dec
April
Jul
Oct
Owen sound 1990
Dec
April
Jul
Oct
Mean Max Mean Min
Temp (° C) Temp (° C)
-3.5
3.9
4
14.8
17.1
26.6
6.4
16
Mean
Total Sain Total Snow Total Precip
(cm)
(mm)
Temp (° C)
(mm)
36.4
138.6
0.2
97.8
70.8
9.4
12
63.2
0
59.8
21.9
59.8
116
0
11.2
116
Mean Max Mean Min
Temp (° C) Temp (° C)
1.7
-5.8
2.3
13.2
13.4
25.4
2.9
13.6
Mean
Total Rain Total Snow Total Precip
(cm)
(mm)
(mm)
Temp (° C)
31
129.2
96
-2.1
45.4
7.7
41
4.2
0
107.5
19.5
107.5
108.8
0
8.3
108.8
Mean Max Mean Min
Temp (° C) Temp (° C)
3.3
-5.2
13.7
2.8
15.1
26.6
4.3
14.6
Mean
Total Rain Total Snow Total Precip
(mm)
(cm)
Temp (° C)
(mm)
112.8
28.8
-0.9
81.5
53
2.6
8.3
50.8
68.4
68.4
0
20.9
87.8
9.4
0
87.8
Mean Max Mean Min
Temp (° C) Temp (° C)
-4.4
2.9
2.5
13.9
25
14.8
14.7
4.6
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
Temp (° C)
85.7
26.1
-0.8
59.8
16.5
64.3
47.8
8.2
0
80
19.9
80
0
102.8
9.7
102.8
Mean Max Mean Min
Temp (° C) Temp (° C)
-3.8
2.2
3.2
12.1
15
23.7
5
13.1
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
Temp (° C)
101.8
-0.8
52
49.8
83
0
7.6
83
0
98
19.4
98
122.8
122.8
0
9.1
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table All continued
Meteorological Data fromthe Environment Canada climate stations located
in southern Ontario (1990)
London 1990
Dec
April
Jul
Oct
Hamilton 1990
Dec
April
Jul
Oct
Delhi 1988
Dec
April
Jul
Oct
Barrie 1990
Dec
April
Jul
Oct
Blyth 1990
Dec
April
Jul
Oct
Means Dec
Means Apr
Means Jul
Means Oct
Mean Max Mean Min
Temp (° C) Temp (° C)
2.5
-5.2
13.1
3.1
25.3
14.7
14.1
4
Mean
Total Sain Total Snow Total Precip
(cm)
(mm)
Temp(°C)
(mm)
151.4
103.9
47.9
-1.4
56.2
11.6
67.6
8.1
144
0
144
20
1.6
108.9
9.1
107.3
Mean Max Mean Min
Temp (° C) Temp(°C)
3
-4.9
13.6
3.2
25.7
15.2
13.9
4.2
Total Rain Total Snow Total Precip
Mean
(mm)
(cm)
(mm)
Temp (° C)
34
127.4
157.3
-0.9
63.7
68.5
4.2
8.4
77.4
77.4
0
20.5
0
86.6
9
86.6
Mean Max Mean Min
Temp (° C) Temp (° C)
-3.7
3.2
13.5
3
26.2
14.6
14
4.7
Mean
Total Rain Total Snow Total Precip
(cm)
(mm)
Temp (° C)
(mm)
149.6
28
177.6
-0.2
71.5
3
74.5
8.3
83.7
83.7
0
20.4
91.4
0
91.4
9.4
Mean Max Mean Min
Temp (° C) Temp (° C)
1.8
-6.8
13
1.2
25.6
15.2
13.5
3.2
Mean
Total Rain Total Snow Total Precip
(cm)
(mm)
Temp (° C)
(mm)
33.2
53.4
86.6
-2.5
7.1
49.8
51.8
2
66.4
0
66.4
20.4
108.4
0
108.4
8.4
Mean Max Mean Min
Temp (° C) Temp (° C)
1.7
-5.4
13.1
2.9
24.2
14.4
13.6
4.7
Total Rain Total Snow Total Precip
Mean
(mm)
(cm)
(mm)
Temp (° C)
65.5
131.5
66
-1.8
40
18.6
S8.6
8
87.5
0
87.5
19.3
0
151
151
9.2
2.62
13.40
25.43
14.11
-4.87
2.82
14.95
4.40
-1.12
8.11
20.22
9.28
86.45
56.70
87.27
108.29
40.36
7.47
0.00
0.16
127.25
63.75
87.27
108.45
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A12
Meteorological Data from the Environment Canada climate stations located
in southern Ontario (1992)
Windsor 1992
Dec
April
Jul
Oct
Waterloo 1992
Dec
April
Jul
Oct
Toronto 1992
Dec
April
Jul
Oct
Sarnia 1992
Dec
April
Jul
Oct
Owen Sound 1992
Dec
April
Jul
Oct
Mean Max Mean Min
Temp (° Q Temp(°C)
2.8
-2.5
11.5
2.6
24.7
16
14.3
5.1
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
Temp (° C)
58.6
47
11
0.2
107.7
7.1
0.8
106.9
183.4
183.4
0
20.3
63.8
9.7
62.8
1
Mean Max Mean Min
Temp (° C) Temp (° C)
0.7
-5.9
9.3
0.5
22.2
11.7
12
1.2
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
Temp (° C)
73.8
42.6
-2.6
34.4
117.2
116.6
1
4.9
158.2
0
17
158.2
68
5
6.6
63
Mean Max Mean Min
Temp (° Q Temp (° C)
1.4
-4.7
9.9
1.2
22.9
13.1
12.5
2.3
Mean
Total Rain Total Snow Total Precip
(mm)
(cm)
Temp (° C)
(mm)
56.3
35
-1.6
24.3
133.8
133.4
0.6
5.6
134.5
0
18
134.5
66
0.4
7.4
65.2
Mean Max Mean Min
Temp (° C) Temp (° C)
2.6
-3.4
9.2
1.4
23
13.8
13.5
3.2
Mean
Total Rain Total Snow Total Precip
(mm)
Temp (° C)
(mm)
(cm)
52.9
7
-0.4
45.9
110.2
108.6
1.6
5.3
167.2
0
18.4
167.2
42.6
8.4
42.6
0
Mean Max Mean Min
Temp (° C) Temp (° C)
1.9
-3.3
8
0.4
20
12.4
11.8
4.2
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
Temp (° C)
(cm)
44
67.6
-0.7
23.6
84.2
4.3
53.2
31
81
0
16.2
81
46
55
8.1
9
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A12 continued
Meteorological Data from the Environment Canada climate stations located
in southern Ontario (1992)
London 1992
Dec
April
Jul
Oct
Hamilton 1992
Dec
April
Jul
Oct
Delhi 1992
Dec
April
Jul
Oct
Barrie 1992
Dec
April
Jul
Oct
Blyth 1992
Dec
April
Jul
Oct
Means Dec
Means Apr
Means Jul
Means Oct
Mean Max Mean Min
Temp (° C) Temp (° C)
1.4
-4.7
10
0.6
22.9
13.2
2.5
12.5
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
Temp (° C)
101.8
-1.6
55.8
59
86.8
5.3
85.2
1.8
204.6
204.6
0
18.1
86.1
80.4
6
7.5
Mean Max Mean Min
Temp (° C) Temp (° Q
1.2
-5.1
9.1
0.5
22.5
12.9
12.3
2.5
Mean
Temp (° Q
-2
4.8
17.8
7.4
Total Rain Total Snow Total Precip
(mm)
(cm)
(mm)
82.6
43.6
42
103.4
10.7
97.9
189.4
189.4
0
79.1
2.6
76.5
Mean Max Mean Min
Temp (° C) Temp (° Q
2.2
-4.3
10.5
0.6
24.2
13.5
2.4
13.2
Mean
Temp (° Q
-1.1
5.6
18.9
7.8
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
91.1
56.8
34.3
103.3
102.7
0.6
174.5
0
174.5
80.4
0
80.4
Mean Max Mean Min
Temp (° C) Temp (° C)
0.8
-6.3
8.8
-0.6
12.7
21.8
11.6
1.5
Mean
Total Rain Total Snow Total Precip
(mm)
(cm)
Temp (° C)
(mm)
87.8
-2.8
13.8
74
60.1
4.1
50.1
10
86.3
17.3
86.3
0
53.4
6.4
44.9
8.5
Mean Max Mean Min
Temp (° C) Temp (° Q
0.2
-4.5
8.7
0.6
21.3
12.1
2.6
11.1
Mean
Total Rain Total Snow Total Precip
(mm)
(mm)
(cm)
Temp (° C)
93
-2.1
34
59
150.5
4.7
141
9.5
145.5
16.7
145.5
0
97.7
21.5
6.9
76.2
1.52
9.50
22.55
12.48
-1.47
5.17
17.87
7.62
-4.47
0.78
13.14
2.75
37.92
99.56
152.46
63.80
40.79
6.76
0.00
5.40
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76.55
105.72
152.46
69.21
Table AI3
Meteorological data used in Winnipeg Climate Centre soil moisture model
(Sept. 6-12,1993)
Location
Pierson
Melita
Brandon
Virden
Birtle
Sandy Lake
Neepewa Water
Oakner
Binscarth
Roblin
Swan River
The Pas
Pine River
Dauphin
Wasagaming
Wilson Creek
Gladstone S.
Delta FS
Langruth
Portage
Cypress River
Starbuck
Altona
Baldur
Miami
Emerson
Morden CDA
Plum Coulee
Pilot Mound 2
Steinbach
Beausejour 2
Winnipeg
Glenlea
Ostenfield
Zhoda
Pinawa WNRE
Sprague
Indian Bay
Rennie
Bissett
Stony Mountain
Gimli
Arborg
Lundar
Period
PCPN
11.2
14.4
10.4
12.6
16.2
20.2
10.6
5.5
18
13.4
222
0
27.8
21
8.6
24.8
11.6
16.9
24
52
9.2
24.8
123
102
6.8
12
9.4
6.9
10.8
17.2
2.6
8.2
12.6
13.6
21
9
14.6
9.2
182
14.9
15
9
16
15.8
Season to date
PCPN(%N)
366(125)
288(82)
429(136)
347(116)
437(140)
360(123)
423(126)
354(115)
380(127)
438(149)
435(141)
406(142)
489(140)
397(123)
366(103)
446(125)
388(119)
463(142)
385(116)
421(119)
490(137)
592(168)
509(152)
473(135)
468(130)
530(149)
543(157)
539(157)
571(167)
403(115)
477(138)
538(153)
521(152)
587(162)
407(116)
434(120)
524(129)
414(106)
492(138)
412(114)
565(165)
403(117)
425(134)
468(145)
SOIL
(% Q
250(100)
189(95)
225(100)
250(100)
225(100)
225000)
200(100)
223(99)
225000)
225(100)
250(100)
193(96)
200(96)
250(100)
175(100)
200(100)
200(100)
200(100)
225(100)
245(98)
174(99)
225(100)
200(100)
225(100)
199(99)
225(100)
200(100)
247(99)
200(100)
225(100)
245(98)
225(100)
225000)
200(100)
200(100)
225(100)
200(100)
148(99)
150(100)
200(100)
225(100)
223(99)
225(100)
225(100)
Soil Moisture
(%)
16.03
12.12
14.42
16.03
14.42
14.42
12.82
14.29
14.42
14.42
16.03
12.37
12.82
16.03
11.22
12.82
12.82
12.82
14.42
15.71
11.15
14.42
12.82
14.42
12.76
14.42
12.82
15.83
1X82
14.42
15.71
14.42
14.42
1X82
12.82
14.42
12.82
9.49
9.62
12.82
14.42
14X9
14.42
14.42
PCPN - Precipitation (mm)
Soil - Modelled Soil Moisture Reserve (mm)
%N - Percent of Normal
%C - Percent of Available Water Holding Capacity
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IMAGE EVALUATION
TEST TARGET (Q A -3 )
v\
^
J
A
.•> ^
s
b
<
*
150mm
A P P L IE D A IIVHGE. Inc
1653 E ast Main Street
Rochester, NY 14609 USA
— Phone: 716/482-0300
- ^ = ~ - ^ = Fax: 716/288-5969
■
=
'■
0 1993. Applied Image, Inc.. ADRights Reserved
with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
6 922 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа