close

Вход

Забыли?

вход по аккаунту

?

Microwave radiometry and snow water equivalence retrievals on snow-covered sea ice in the marine cryosphere

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 of the
copy subm itted.
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 Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
NOTE TO USERS
The original manuscript received by UMI contains pages with
indistinct print. Pages were microfilmed as received.
This reproduction is the best copy available
UMI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Microwave Radiometry and Snow Water Equivalence
Retrievals on Snow Covered Sea Ice in the Marine
Cryosphere
Sheldon D. Drobot
A T hesis
S ubm itted to the Faculty of G ra d u ate Studies
in P a rtia l Fulfillm ent of th e R equirem ents
for the Degree of
M aster of A rts
C entre for E arth O bservation Science
D epartm ent of G eography
University of M anitoba
Winnipeg, M anitoba
© A ugust, 1997
/
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
Id k fl
■ t ■
National Library
0f Canada
Bibliotheque nationale
du Canada
Acquisitions and
Bibliographic Services
Acquisitions et
services bibliographiques
395 Wellington Street
Ottawa ON K1A0N4
Canada
395. rue Wellington
Ottawa ON K1A0N4
Canada
Your He Votre reference
O ur 619 N otre reference
The author has granted a non­
exclusive licence allowing the
National Library of 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 o f 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 impnmes
ou autrement reproduits sans son
autorisation.
0 61 2 32915-1
-
-
Canada
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
THE UNIVERSITY OF MANITOBA
FACULTY OF GRADUATE STUDIES
MASTER'S THESIS/PRACTTCUM FINAL REPORT
The undersigned certify that they have read the Master's Thesisd^puctixnouentitled:
MICROWAVE RADICHETRY AND SNOW MATER EQUIVALENCE RETRIEVALS ON SNOW
COVERED SEA ICE IN THE MARINE CRYOSPHERE.
subm itted by
SHLEDON D . DR0BOT
in partial fulfillm ent o f the requirements for the degree of
________ MASTER OF ARTS_________________
The Thesis&t&Xfititibl Examining Committee certifies that the thesis/practicum (and oral
exam ination if required) is:
A MAO
(Approved or Not Approved)
©
Thesis
Advisor:
_s~\
,
D.G. BARBER
□
Practicum
S.M .P. BENBOM__
L . SHAFAI
Date:
7
7
(fonns\thereptm - 08/95)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
THE UNIVERSITY OF MANITOBA
FACULTY OF GRADUATE STUDIES
* * * * *
COPYRIGHT PERMISSION PAGE
MICROWAVE RADIOMETRY AND SNOW WATER EQUIVALENCE RETRIEVALS
ON SNOW COVERED SEA ICE IN THE MARINE CRYOSPHERE
BY
SHELDON D. DROBOT
A Thesis/Practicum submitted to the Faculty of Graduate Studies of The University
of Manitoba in partial fulfillment of the requirements of the degree
of
MASTER OF ARTS
Sheldon D. Drobot 1997 (c)
Permission has been granted to the Library of The University of Manitoba to lend or sell
copies of this thesis/practicum, to the National Library of Canada to microfilm this thesis
and to lend or sell copies of the film, and to Dissertations Abstracts International to publish
an abstract of this thesis/practicum.
The author reserves other publication rights, and neither this thesis/practicum nor
extensive extracts from it may be printed or otherwise reproduced without the author's
written permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
Snow w ater equivalence (SWE) derivation over sea ice re q u ire s a b etter
u n d e rs tanding of how variations in th e evolving snow -sea ice m ix tu re affect
m icrow ave emission.
In th is th esis, th e effects of (a) v a ria tio n in liq u id
w a te r content w ithin a seasonally dynam ic snowpack an d (b) h e te ro g en e ity
in underlying ice are exam ined. Geophysical snow d a ta and in s itu passiv e
m icrow ave signatures were collected in the C anad ian A rctic A rc h ip e la g o
d u rin g
the spring
C ryospheric
of 1996 u n d e r
E xperim ent
the
(C-ICE).
C o llab o rativ e-In terd isc ip lin ary
Surface
based
ra d io m e te r
m easu rem en ts were collected a t 19, 37 and 85GHz (both v e rtic a l a n d
h o rizontal
polarizations).
Snow
data
and
Special
S e n so r
M icrow ave/Im ager (SSM/I) d a ta w ere collected from 1993-1995 d u rin g th e
S easonal Sea Ice M onitoring a n d M odelling (SIMMS) program .
R esults indicate SWE can be derived w ith an in situ m icrow ave ra d io m e te r
w h en th e snowpack is dry.
M ultiple regression techniques a re show n to
b etter estim ate in situ SWE over the case site.
W ith in c re a se d w a te r in
liquid phase, em ission from th e snow pack causes c u rre n t a lg o rith m s to
overestim ate SWE. V ariation in ice type and spatial p a tte rn of th e ice lim it
th e applicability of SWE derivation w ith SSM/I. The presence o f m u ltiy e a r
ice (MYI) lowers em issivity values th a t leads to an u n d e re stim a tio n of SWE
w ith c u rre n t algorith m s.
*
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
Special t h a n k s go out to Dr. David G. B arber for his c o n tin u al support a n d
patience.
In p articu la r, I thank him for show ing me th e n atu re of
scientific inquiry, and giving me the opportunity to explore the rem ote
n o rth e rn reaches of our world. I would also like to acknow ledge the help
my fellow colleagues provided both in the field a n d in th e lab. Jo h n Y ackel
an d Jo h n Iacozza w ere in strum ental in h elping m e acquire th e data u sed
in th is w ork. T hanks also to Theresa Nichols, J o h n H an esiak , and David
M osscrop for th e ir editorial comments.
I w ill never forget the love and support of my fam ily a n d friends, who have
m ade my tim e in W innipeg truly enjoyable. I w ould also like to recognize
W alt L ipinski as being an energetic, tho u g h tfu l ju n io r high teacher w ho
first got m e in terested in this place called E a rth .
T h an k s also to the various institutions and agencies th a t have supported
th e final work.
In p articu lar, CRYSYS, C a n a d ia n Ice Services, NSERC,
ONR, an d th e N o rth ern Studies T raining P ro g ra m w ere in stru m e n ta l in
com pleting th is thesis. T hanks to Polar C o n tin en tal S h elf for logistical field
su p p o rt.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
A b stract........................................................................................................................ii
A cknow ledgem ents.................................................................................................. iii
L ist o f F ig u res........................................................................................................... vii
L ist o f T ab les.............................................................................................................. x
CHAPTER 1 - Introduction......................................................................................1
1.1 Introduction.....................................................................................................1
1.2 R esearch Design............................................................................................ 4
1.3 C hapter Reviews............................................................................................ 5
CHAPTER 2 - S cien tific F o cu s............................................................................... 7
2.1 Introduction.................................................................................................... 7
2.2 T he M arine Cryosphere............................................................................... 8
2.2.1 Sea Ice (2,3)..........................................................................................9
2.2.2 Snow (4).................................................................................................15
2.3 T he Role of Sea Ice and Snow in M arine Cryospheric P rocesses
22
2.3.1 Influence of Snow O ptical P ro p e rtie s ..........................................23
2.3.2 Influence of Snow Physical P ro p e rtie s ....................................... 26
2.4 Evidence for Change in the M arine Cryosphere.....................................29
2.5 C onclusions.................................................................................................... 34
CHAPTER3 - B ackground......................................................................................36
3.1 Introduction.................................................................................................... 36
3.2 Microwave Interaction T h e o ry ..................................................................38
3.2.1 E xperim ental R e s u lts .................................................................... 47
3.3 H istorical Use of Microwave R adiom etry w ithin the M a rin e
C ry o sp h e re ............................................................................................................ 51
3.4 R esearch Q u estio n s......................................................................................59
3.5 C onclusions.................................................................................................... 61
/v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4 - I n S itu D erived SWE a n d th e Effect of W a te r in Liquid
P h a s e .............................................................................................................................62
4.1 Introduction..................................................................................................... 62
4.2 M ethods.............................................................................................................64
4.2.1 Snow an d S ea Ice Sam pling............................................................65
4.2.2 Microwave R adiom etry ....................................................................67
4.3 R esults and D isc u ssio n ............................................................................... 68
4.3.1 Geophysical P roperties.....................................................................68
4.3.2 Microwave R adiom etry ....................................................................76
4.4 C onclusions..................................................................................................... 82
CHAPTER 5 - Effects o f Sea Ice H eterogeneity o n SS M /I SWE
D erivation.....................................................................................................................84
5.1 Introduction.....................................................................................................84
5.2 M ethods.............................................................................................................86
5.2.1 In Situ vs. SSM /I SWE E stim atio n ................................................86
5.2.2 Spatial H eterogeneity as a Source of V a ria tio n .........................88
5.2.3 SAR C h aracterization of S patial H e te ro g e n e ity ....................... 91
5.3 R esults and D isc u ssio n ............................................................................... 92
5.3.1 In S itu vs. SSM /I SWE E stim atio n ................................................92
5.3.2 Ice H eterogeneity as a Source of V ariation................................. 94
5.3.3 SAR C haracterization of S patial H e te ro g e n e ity ....................... 98
5.4 C onclusions..................................................................................................... 106
CHAPTER 6 - S u m m a ry a n d C onclusions...........................................................108
6.1 S u m m a ry ......................................................................................................... 108
6.2 C onclusions..................................................................................................... 110
6.3 F u tu re D irec tio n s.......................................................................................... 112
C ited R eferences......................................................................................................... 115
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A p p en d ices................................................................................................................... 125
Appendix A: Acronyms a n d A bbreviations................................................... 126
Appendix B: L ist of S ym bols...............................................................................127
Appendix C: L ist of T e rm s ................................................................................. 128
Appendix D: N orm ality P lo ts from C hapter 3 ............................................... 129
Appednix E: ANOVA R e su lts from C hapter 3 ..............................................132
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Figure 2.1 (a) M icrowave interactions and (b) processes w ith in th e m a rin e
c ry o s p h e re .................................................................................................................9
Figure 2.2 M inim um a n d m axim um sea ice ex ten t............................................10
Figure 2.3 Average salin ity of sea ice as a function of ice th ick n ess for cold
sea ice d in in g th e grow th se a so n ......................................................................13
Figure 2.4 Idealized ice salin ity curves for first-year ice (a-d) an d m u ltiy e a r
ice (e, f)......................................................................................................................14
Figure 2.5 P recipitation recorderd a t Resolute B a y ............................................. 16
Figure 2.6 E quilibrium a n d k in e t ic growth snow grains....................................19
Figure 2.7 Sketch of snow grains, air, and w ater d istrib u tio n s in the
p e n d u la r re g im e .................................................................................................... 21
Figure 2.8 Sketch of snow grains, air, and w ater d istrib u tio n s in the
fu n icu lar re g im e ...................................................................................................22
Figure 2.9 Snow-chlorophyll relationships in th e Arctic. S tation 1 R^= 0.84;
Station 2 R2= 0 .8 6 ...................................................................................................26
Figure 2.10 O bserved tre n d s of Arctic w inter m ean te m p e ra tu re s from 19611990 ........................................................................................................................... 30
Figure 2.11 Observed tre n d s of Arctic s u m m e r m ean te m p e ra tu re s from
1961-1990...................................................................................................................31
Figure 2.12 Com posite greenhouse w arm ing projections as predicted by
seven Global C lim ate Models is presented as the difference betw een
c u rre n t conditions a n d those u nder doubled CO 2 .........................................^
Figure 2.13 Composite of greenhouse precipitation projections as predicted
by seven Global C lim ate Models is presented as th e ratio betw een a
doubled CO 2 scenario an d c u rre n t conditions............................................... 34
Figure 3.1 C alculated b rig h tn ess tem perature as a function of snow a te r
eq u iv a le n ce ..............................................................................................................37
Figure 3.2 A tm ospheric window s facilitating m icrowave rem ote sensin g . 38
Figure 3.3 E m issivity a s a function of ice th ick n ess........................................... 40
Figure 3.4 Volume sc a tte rin g p h en o m en a...........................................................41
Figure 3.5 M easured p erm ittivity of artificially grown sea ice as a function
of brine volume fra c tio n .......................................................................................43
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3.6 Perm ittivity as a function of density in dry snow. E x p e rim e n ta l
data provided by Nyfors (1983), H allikainen (1977), C um m ing (1952) a n d
H allikainen et al. (1986)...................................................................................... 44
Figure 3.7 Modified Debye-like model perm ittivity of snow as a function of
frequency and w ater in liquid p h a se ................................................................46
Figure 3.8 Modified Debye-like model lossof snow as a function of
frequency and w ater in liquid p h a se ................................................................46
Figure 3.9 E xperim ental values obtained by various au th o rs for sea ice at
1GHz: (a) perm ittivity, (b) loss............................................................................48
Figure 3.10 E xperim ental values obtained by various a u th o rs for sea ice
between 9 and 16GHz: (a) perm ittivity, (b) lo s s........................................... 49
Figure 3.11 Changes in perm ittivity as a result of increasing w a te r in liquid
phase. The increm ental perm ittivity equals e’ws - e’d s..............................50
Figure 3.12 W et snow dielectric loss as a function of w ater in liquid p h ase 51
Figure 3.13 Em issivity variations in MYI as a function of frequency
(incidence angle = 50°).........................................................................................55
Figure 3.14 Average brightness tem p eratu re as a function of frequency over
snow covered first-year sea ice for snow thicknesses of (a) <3m m , (b) 3 to
50mm and (c) >50m m .......................................................................................... (77
Figure 4.1 C-ICE '96 Field S ite .................................................................................64
Figure 4.2 Radiom eter sam pling site in C -IC E '96............................................. 65
Figure 4.3a. Absolute a n d 4.3b. Relative calibration curves for th e SBR
sy ste m .........................................................................................................................
Figure 4.4 Snow depth d u ring C-ICE'96................................................................68
Figure 4.5 Profile distribution of (a) liquid w ater volume, (b) m odelled
37GHz loss, (c) density, (d) modelled 37GHz perm ittivity an d (e) salinity70
Figure 4.6 Seasonal evolution of (a) liquid w ater volume, (b) m odelled 37GHz
loss, (c) density, (d) modelled 37GHz perm ittivity and (e) s a lin ity
72
Figure 4.7 D iurnal sep aratio n of the seasonal evolution in (a) liquid w a te r
volume, (b) modelled 37GHz loss, (c) density, (d) m odelled 37GHz
perm ittivity and (e) s a lin ity ................................................................................75
Figure 4.8 In itial regression plots w hich include all 3 d iu rn a l s a m p lin g
periods lum ped over th e full duration o f the field e x p erim en t..................77
Figure 4.9 W ater volume by subset......................................................................... 78
Figure 4.10 Incidence angle effects on SWE estim atio n ................................... 80
Figure 5.1 SIMMS field sites.....................................................................................87
vm
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 5.2 Case study d iag ram of underlying ice types. D ark g rey sh a d e s
represent FYI, while lig h t grey shades rep resen t M Y I.............................89
Figure 5.3 F-13 SSM/I a n te n n a p a tte rn s.................................................................91
Figure 5.4 Bivariate plot of in situ versus SSM/I SWE. Ju lia n days in clu d ed
in the g r a p h .............................................................................................................94
Figure 5.5 Em issivity v ariations overlaid on original ice c o n c e n tra tio n
p a tte rn s .................................................................................................................... 95
Figure 5.6 E m m isivity variatio n as a function of frequency, ice
configuration and SSM/I a n te n n a p a tte r n s ................................................... 96
Figure 5.7 37V T b based on ice type an d tem perature. 19V is sim ila r
97
Figure 5.8 Ice type during 1993 and 1994 as determ ined by the NASA T eam
A lg o rith m ................................................................................................................ 99
Figure 5.9 E nlarged portion of Figure 5.8 o u t lin in g ice type d u rin g 1993 a n d
1994 as determ ined by th e NASA Team A lgorithm ...................................... 99
Figure 5.10 ERS-1 classified im age from M ay 13,1993. Boxed a re a s in d icate
one SSM/I pixel (25km2). N um bers are used to d istin g u ish th e boxed
areas in th e te x t.................................................................................................... 101
Figure 5.11 ERS-1 classified im age from M ay 7, 1994. Boxed a re a s in d ic ate
one SSM/I pixel (25km2). N um bers are used to d istin g u ish th e boxed
areas in th e te x t .................................................................................................... 103
F igure 5.12 ERS-1 classified im age from April 19, 1995. Boxed a re a s
indicate one SSM/I pixel (25km2). N um bers are used to d istin g u ish the
boxed areas in the t e x t ........................................................................................ 105
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
T able 2.1 Typical albedo values in th e m arine cryosphere.................................24
Table 2.2 L inear tre n d analysis of snow depth and ice th ic k n e s s................... 27
Table 2.3 Trends in Arctic sea ice covers, 1978-1994........................................... 32
T able 3.1 Response of M icrowave Em ission W ith Various Snow G eophysical
P ro p e rtie s................................................................................................................ 39
Table 4.1 C orrelation m atrix betw een
snowpack layers and b rig h tn e ss
te m p e ra tu re s........................................................................................................... 76
Table 4.2 Subsets of 50° T b v s . S W E ........................................................................ 78
T able 4.3 Single and m ulti-frequency/polarization R2 ....................................... ®
T able 4.4 M ultiple lin e a r regression r e s u l ts ........................................................ 81
Table 5.1 Derived SWE d a ta acquisition dates....................................................... 88
T able 5.2 Case study table of underlying ice ty p e s ...............................................89
T able 5.3 Microwave em issivities of sea ice used in case stu d ies.................... 90
T able 5.4 B rightness tem p era tu re s from the te s t d a te s...................................... 93
T able 5.5 SWE derived w ith [4.1] and prescribed ice te m p e ra tu re s..................98
Table 5.6 Ice type composition of ERS-1 classified image
from May 13,1993................................................................................................. 102
T able 5.7 Ice type composition of ERS-1 classified image
from May 7 ,1 9 9 4 ...................................................................................................104
T able 5.8 Ice type composition of ERS-1 classified image
from April 19, 1995...............................................................................................106
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 1 - In trodu ction
1.1 Introduction
The m arine cryosphere consists of snow on sea ice, th e u n d e rly in g ocean
and the atm osphere, as well as biological life living w ith in th ese volumes
and energy exchanges operating across them . Scientific in te re s t centres on
th e m arine cryosphere as an early indicator of global change resu ltin g
from its sensitivity to a series of enhanced feedback processes (EPCC, 1990).
Snow, w ith a low th e rm a l conductivity and high albedo, is a n in tegral
component of th is m a rin e cryosphere.
Physically, th e snow volume
governs key energy exchanges between the ocean a n d th e atm osp h ere
(Barber et a l.t 1995), w hich in tu rn controls processes su ch sea ice accretion
an d ablation.
I t is therefore essential to obtain im proved inform ation
regarding snow covered sea ice in the m arine cryosphere.
The
In terg o v ern m en tal
P anel
on Climate
C hange
(IPCC)
projects
w arm ing in th e m a rin e cryosphere will reduce the sp a tia l a n d tem poral
extent of snow cover (IPCC, 1995). The development of rem o te sensing is a n
im portant step in docum enting such change, as it provides excellent spatial
and tem poral coverage in contrast with traditional in s itu observations
(Massom, 1991). In p a rtic u la r, derivation of snow w a te r equivalence w ith
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
microwave radiom etry appears to be one of the more prom ising m o n ito rin g
approaches (Goodison, 1997).
SWE is the am ount of w a ter a m elted
snowpack would yield. SWE accounts for variations in snow density as w ell
as snow depth, which is necessary to fully docum ent m odifications in snow
cover am ount. F urtherm ore, SWE retrievals are successfully acquired over
te rre stria l surfaces (e.g. Goodison et al., 1990). U nfortunately, the com plex
n a tu re of a snow covered sea ice volum e has prevented operational SWE
estim ation in the m arine cryosphere to date (Carsey et al., 1992).
The
conceptual
u n d er p in n in g s
for
deriving
SWE
w ith
m icrow ave
radiom etry are based on the effect a snow layer has on m icrow ave
em issions originating w ithin the underlying sea ice.
In cre asin g snow
thickness or density increases the scattering potential of a snow pack,
resu ltin g in decreases in microwave em ission from the ice. However, th e
effect of w ater in liquid phase on microwave em ission is not com pletely
understood nor is the significance of variatio n in the underlying ice type.
Specific objectives of this thesis are linked to determ ining th e effects sn o w
an d sea ice have on the successful derivation of snow w a ter equivalence
over the m arine cryosphere.
In p articu lar, this th esis exam ines
a
relationship between snow geophysical properties, m icrow ave em issio n
an d in situ derived SWE. It also outlines how ice type heterogeneity affects
Special Sensor M icrowave/Imager derived SWE.
To m ake this a tractable thesis, sp a tia l and tem poral re s tra in ts a re applied.
The investigation is spatially lim ited to specific field sites located w ith in a
h u n d red kilom eter radius of Resolute Bay, NT. Geophysical sam ples a re
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
only d raw n from saline, sm ooth, first-y ear ice considered typical of th is
region. Tem porally, d a ta collection begins in the la tte r stages of th e cold
w in te r period and extends into th e b e g i n n i n g of th e m elt onset season.
T his th esis is a small p a rt of a large, m ultidisciplinary, in te rn atio n al effort
to u n d e rsta n d the complex interw o rk in g s of the cryosphere.
It seeks to
fu rth e r our knowledge in the fields of geophysical inversion and a lg o rith m
developm ent by im proving c h arac teriza tio n and u n d e rstan d in g of both th e
snow covered sea ice volume and its in teractio n w ith microwave energy.
In th e larg e r framework, scientific re se a rc h in th e cryosphere is focused on
b etter characterization, u n d e rsta n d in g an d prediction of E arth p ro cesses.
C h a rac te riz atio n represents an in itia l attem p t to m onitor and describ e
c u rre n t conditions in the m arine cryosphere. U nderstan d in g fu rth e rs th a t
w ork by providing hypotheses w hich explain w hy things h a p p en a n d
provide knowledge as to how they h ap p en . Finally, prediction is c o n ce rn e d
w ith exploiting the knowledge g ained th ro u g h u n d e rstan d in g to e x p la in
how v ariatio n s in one component of th e m arin e cryosphere will affect o th e r
processes, both locally and globally.
To accom plish the specific goals of t h is th esis a scientific fram ew ork w ill be
p resen ted in the next section. T his provides an overall scientific objective
an d a series of research tools th a t provide the underlying stru c tu re for th e
re m a in in g chapters.
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 2 R esearch D esign
To m ake a significant scientific contribution in this thesis th e follow ing
m ission statem en t is provided:
Science Objective: “To provide insight on the developm ent o f
SW E algorithm s for snow covered sea
ice using m icrow ave
radiom etry. ”
The tools used to assess th e scientific objective are divided into four sections:
1) In -situ Observations: Few experim ents have analyzed snow cover in th e
m arin e cryosphere owing to th e hostile clim ate and rem ote location of
th is region. None have specialized in snow w ater equivalence stu d ie s
over sea ice. This w ork provides d etailed in situ snow observations fro m
various seasonal an d d iu rn a l regim es.
The snow d a ta provides a
com prehensive suite of variables needed to assess SWE variations w ith in
a case study or field context.
2)
M odelled (derived) V ariables:
Passive microwave
in teractio n
controlled by th e sea ice-snow volum e dielectric properties.
is
M odelled
values are used in th is th esis to e stim ate snow pack dielectrics.
The
models are based upon th e re su lts in (1) and can be used in c o n ju n c tio n
w ith th e physical d a ta to u n d e rs ta n d the n a tu re of e lectro m ag n etic
in teractio n between th e snow
and
the passive sig n a tu re
in
th e
m icrowave portion of th e electrom agnetic spectrum .
3) R em otely Sensed D ata: Rem otely sensed d a ta can record changes in
m icrowave emission over a n ice surface. W ith th e high cost of o p e ra tin g
m icrowave radiom eters and in h e re n t calibration difficulties, few stu d ie s
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have successfully
coupled
microwave
em ission
w ith
an evolving
snowpack. This study provides the necessary passive m icrow ave data to
explore derivation of both in situ and SSM/I derived SWE over varying
diurnal tim e steps.
The d iu rn al aspect of th is study is key since
operational SSM/I d a ta is available a t various tim es o f the day in the
m arine cryosphere.
4) Statistical A n a lysis: A ssessing th e scientific objective is facilitated by the
use of statistical techniques as a m eans of e n s u rin g th a t the w ork
presented h ere is reproducible. In p articu lar, c o rrelatio n analysis an d
a full suite of g eneral lin ear regression m odels a re used to help
understand th e complex interaction between m icrow ave energy and the
evolving sea ice-snow m ixture.
1.3 C hapter R eview s
This thesis contains six c h ap ters, each contributing to th e assessm ent of
my stated scientific objective.
The first c h ap ter o u tlin es the purpose,
context and constraints of th e project. C hapter two o u tlin es the scientific
focus and provides descriptions of sea ice, snow an d th e ir role w ithin the
m arine cryosphere. It also exam ines how changing snow conditions could
affect this region and explores evidence th a t ch an g es
have already
occurred.
The third ch ap ter provides th e necessary m icrow ave in te rac tio n theory to
link geophysical snow a n d ice properties to m icrow ave radiom etry.
It
sum m arizes the historical precedent for u sing m icrow ave radiom etry in
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the m arine cryosphere and concludes with a series of re se a rc h questions
which stru c tu re th e rem ain in g chapters.
C hapter
four
exam ines
the
relationship
between
snow
properties,
microwave em ission an d in situ derived SWE over a case study site.
Geophysical snow d a ta collected in the C anadian A rctic A rchipelago
d u ring the sp rin g of 1996 is initially exam ined in a vertical profile over
seasonal and d iu rn a l tim e steps.
Results are th en exploited to im prove
derivation of SWE w ith a n in situ microwave radiom eter.
The fifth ch ap ter expands upon the results of ch ap ter four to exam ine the
effects of heterogeneity in underlying ice type on SSM/I derived SWE. In
situ and SSM/I derived SWE are compared to gain an u n d e rs ta n d in g of the
differences p resen t betw een the two. Then a series of sensitivity tria ls are
perform ed to theoretically d e t e r m in e how variations in ice em issivity affect
SWE estim ation.
Specifically, these trials exam ine heterogeneity in both
the sp atial a rra n g e m e n t and type of ice (first-year ice vs. m u ltiy ea r ice).
Finally, real w orld diversity is exam ined w ith SSM/I a n d th e E arth
Resources S atellite (ERS-1).
The concluding c h a p te r is utilized to sum m arize the rele v an t findings of
this work. It is inten d ed to reinforce the conclusions an d provide in sig h t
into future directions of research raised by this thesis.
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2
-
S cien tific F ocus
2.1 Introduction
As previously stated , snow cover is a m ajor com ponent o f the m a rin e
cryosphere owing to its influence on energy a n d m ass exchanges.
The
seasonal snow cover is a prim ary factor in controlling e n erg y fluxes across
th e m arine cryosphere d u e to its high albedo. D ecreasing snow cover would
lower the average se aso n al albedo of th is region,
in c re a s in g
energy
absorption and lead in g to the well-known tem p era tu re -alb e d o feedback
process described by H enderson-Sellers and Robinson (1987).
Sim ilarly,
m ass exchanges across th e m arine cryosphere a re lim ite d by the low
th erm al conductivity of snow.
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This chapter describes the snow covered sea ice volume an d the effects it
h as on the m arine cryosphere.
It begins w ith a brief description of th e
m arin e cryosphere and its com ponents. Two of these com ponents, se a ice
an d snow, are th en exam ined m ore thoroughly.
The form ation a n d
evolution of both surfaces is given to exam ine th eir relevance to SWE
derivation. The seasonal snow pack is also described w ith respect to th e
effects it has on th e shortw ave rad iatio n balance, m ass tra n s fe r an d th e life
cycles of various biologic entities.
Finally, an exam ination of re c e n t
tem p eratu re and sea ice extent d a ta points out evidence th a t th is region is
c u rre n tly undergoing noticeable change.
2.2 The M arine C ryosphere
In th is thesis, th e m arine cryosphere (Figure 2.1a) can be conceptualized as
a physical system containing sea w a te r (1), m ultiyear sea ice (2), first-y ea r
sea ice (3), snow (4) and th e atm osphere (5).
Each of th ese com pon en ts
vigorously interacts w ith th e others th ro u g h specific energy and m a s s
exchanges (Figure 2.1b). Some of th ese exchanges, such as h e a t tr a n s f e r
betw een the ocean and atm osphere, a re in stru m e n tal in reg u la tin g th e
global energy balance (H enderson-Sellers and Robinson,
1987).
T he
physical properties of snow (and to a lesser degree sea ice) are im p o rta n t
because they restric t energy and m ass exchange betw een th e u n d e rly in g
ocean and the overlying atm osphere.
In
w hat follows,
a d etailed
description of the snow-sea ice m ixture, from form ation th ro u g h evolution,
will be presented in order to better understan d
how th e ir p h y sic al
properties affect th e m arine cryosphere.
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
<a)
\
Thermal
Microwave
Emission
Radar
Scattering an
Reflection
See Water
t
Shortwave
(b)
Shortwave
Aerosal
Evaporation
Longwave
Shortwave
Convection
Upwelling
Brine
Deep Water
Formation
F igure 2.1 (a) Microwave interactions and (b) processes within the m arine cryosphere.
(adapted from Carsey, 1992) Numbers referenced in section headings below.
2.2.1 S ea Ice (2,3)
The polar oceans are seasonally covered by a dynamic, uneven layer of
frozen w a te r term ed sea ice (Carsey et al., 1992). The spatial coverage of
th is ice exhibits a strong seasonal bias (Figure 2.2).
D uring th e s p rin g
m axim um , sea ice coverage can be a s high as 15 x 106 km 2 (LeDrew, 1990).
However, in the sum m er sea ice re tre a ts to th e rem ote n o rth ern regions of
th e m arin e cryosphere, w here it covers as little as 8 x 106 km 2 (LeDrew,
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1990).
T he 7 x 106 km 2 differential is larger th an the a re a of all the
C anadian provinces com bined (B arber, 1993).
Q Absolute Minimum
B Average Minimum
flAbsolute Maximum
f l Average Minimum
F ig u re 2.2 Minimum and maximum sea ice extent, (adapted from Henderson-Sellers and
Robinson, 1987)
In this work, two distinct types of sea ice are exam ined: FY I a n d M Y I.
FYI is a m ixture of p u re ice, salt, liquid brine and air inclusions typically
30cm to 200cm thick (World M eteorological O rganization, 1970). MYI is ice
th a t has survived a t least one su m m e r m elt season. Since it is older, M Y I
tends to grow thicker, w ith 3.0 to 4.0m depths being common (M aykut and
U n dersteiner, 1971), a lth o u g h g re a t variability exists.
It also te n d s to be
more topographically diverse an d less saline th a n FYI.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Form ation o f F irst-year Sea Ice (2):
In itial ice form ation occurs a t the w ater surface w h e re h e a t tra n s fe r
betw een th e cool atm o sp h ere and warm ocean is g rea test.
G row th begins
w ith th e form ation of sm a ll platelets and needles term ed frazil (Tucker et
a l., 1992).
Since g row th occurs in a brine-rich e n v iro n m en t, several
considerations a re im p o rtan t.
First, the presence of s a lt depresses the
freezing point of w a te r according to [2.1] ( N e u m a n n a n d P ierson, 1966)
T f = -0.0 0 3 -0 .0 5 2 7 S W -0 .0 0 0 0 4 S W 2
[2.1]
w here S w is the sa lin ity of w a ter in parts per thousand (%o).
U n d er typical
conditions se a w a te r freezes a t -1.8°C.
The second im p o rtan t condition relates to w a ter density.
F or seaw ater
g re a te r th a n 24.7 %o th e m axim um density is less th a n th e freezing point.
Therefore, continued cooling of seaw ater above th is sa lin ity th resh o ld leads
to a vertically u n sta b le density distribution.
C onsequently, the en tire
convective layer, g en erally 10-40m (Doronin and K heisen, 1975), m u st cool
to the freezing point before sea ice formation begins.
C ontinued clum ping
of frazil crystals resu lts in a
conglom erate
of
unconsolidated cry stals a n d seaw ater labeled grease ice. In th e absence of
w ind p rim a ry form ation ends w hen frazil coalesces into a solid, elastic ice
cover term ed n ilas.
Conversely, pancake ice form s i f w indy conditions
prevail. P ancakes, n o rm ally 0.3 to 3.0m in d iam eter (T ucker et al., 1992),
have slightly reused edges because they adhere new frazil c ry stals to th eir
edges and continually bum p into one another.
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Once ice covers the ocean surface any fu rth er ice grow th m u st occur on th e
ocean side of the ice sheet.
T he grow th rate of this
congelation ice is
governed by a tem p eratu re g rad ien t across the ocean-sea ice-atm osp h ere
boundary (Tucker et al., 1992).
Evolution o f First-year Sea Ice (2)
From a microwave radiom etry perspective, evolution of the ice is m ain ly
concerned w ith the incorporation and rejection of brine in the volum e
(Perovich an d Gow, 1991; Arcone et al., 1986).
governed by w ater salinity a n d
the growth
In itial ice salin ity is
rate, w hich
tem p eratu re dependent. Laboratory m easurem ents
1975; Weeks an d Lofgren, 1967)suggest
is
h ig h ly
(Cox and W eeks, 1988,
the initial salinity
of ice (Si)
is
related to sea w ater s a lin it y (S*,) by [2.2]:
Si = KeffSw
[2.2]
w here Keff is th e effective distribution coefficient.
As ice thickness increases over tim e, the average salinity g e n erally
decreases from 20%o (during form ation) to 5%o after 2 to 4 weeks (Eide a n d
M artin, 1975). In cold ice two se p ara te regression equations adequately
describe ice salinity as a function o f ice thickness (Figure 2.3). Below 40cm,
[2.3] explains 61% of the salinity variatio n , while [2.4] explains 88% of th e
salinity variatio n w hen ice is g re a te r th a n 40cm thick.
S! = 1 4 .2 4 -1 9 .3 9 ^
[2.3]
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[2.41
S2 = 7.88-1.59h2
w here S is salinity in %o, h i is for ice less th an or equal to 40cm and h<i is
for ice g reater th a n 40cm.
18
16
14
S= 14.24 - 19.39H
2 12
10
6
4
2
0
0
1.5
2
2.5
Thickness (m)
3
3.5
4
F ig u re 2.3 Average salinity of sea ice as a function of ice thickness for cold sea ice d u rin g
the growth season, (adapted from Cox and Weeks, 1974)
B rine volume is a function of salinity and tem perature.
brine begins to drain from th e ice sheet.
A fter form atio n
Subsequent b rin e content is
determ ined by the physics of p h ase equilibrium (Weeks a n d Ackley, 1986),
w hich essentially correlates a brin e volume with a given te m p e ra tu re . T he
dom inant m echanism s
of brine drainage are expulsion
a n d gravity
drainage (Eide and M artin, 1975). Brine expulsion is governed by p re s s u re
differences arisin g from ice form ation.
As w ater freezes, it physically
expands, forcing brine to be tran sp o rted away th ro u g h crack s in th e ice
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Tucker et al., 1984).
The m ajority of the brine c h a n n e ls dow nw ards,
although some is forced onto the sea ice surface (M artin , 1979) and even
wicked into the b asal snow layers (Drinkw ater and C rocker, 1988). Tucker
et al. (1992) suggest expulsion dom inates brine m ovem ent only in the upper
ice layers, while Cox and W eeks (1975) show th e effect of expu lsio n on brine
tra n sp o rt dim inishes w ith tim e. Gravity drainage d o m in ates b rin e tra n sit
over m uch of th e y e ar (Tucker et al., 1992). Over tim e th e d esalinizatio n
process leads to low er ice salinities throughout th e m iddle of th e ice volume
(Figure 2.4).
S alinity rem ain s high in the u p p er lay ers because it
entrapped more d u rin g form ation. The bottom layers re m a in h ig h because
they have not lost m uch of the brine entrapped d u rin g ice form ation an d
they receive a continuous supply of brine drained from above (T ucker et al.,
1992).
100
S- 150
200
250 —
300
0
2
4
6
8
10
12
Ice Salinity (%o)
Figure 2.4 Idealized ice salinity curves for first-year ice (a-d) and multiyear ice (e,f)- (adaptedfrom Tucker
etaL 1992)
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
D esalinization also leads to local variations in salinity. S ev eral re se a rc h e rs
have noticed significant v a ria tio n s in ice salinity w ith in m etres of each
other, even if the stru c tu re of th e ice is sim ilar (e.g. E icken et al., 1991;
T ucker et al., 1984). These irreg u larities are likely a re s u lt of variations in
th e m aturity of brine d ra in a g e channels (Tucker et a l., 1992).
W here
d rain ag e channels are well established, ice salinity is decreased.
Evolution o f M ultiyear Ice (3)
In term s of this thesis the m ost im portant physical difference in M Y I
rela te s to its in te rn al stru c tu re . The pattern of summer m elt and w inter
freeze in MYI grow th gives rise to a topographically diverse stru c tu re
consisting of rounded hum m ocks and low-lying m eltponds.
D uring the
su m m e r m elt season, brine d rain a g e rapidly increases a s a re su lt of two
processes.
F irst, w arm in g expands the brine pockets, lead in g to the
form ation of drainage c h an n e ls th a t facilitate the b rine d ra in a g e process.
Second, the large influx of snow m elt from above flushes b rin e dow nw ards
into th e ocean. As a resu lt, salin ities in the upper 50- 100cm o f the ice sheet
a re generally 1%o, com pared w ith 4-7%o for FYI (Tucker et a l., 1992)
2JL2 Snow (4)
Snow cover is ubiquitous over th e sea ice surface in the m a rin e cryosphere.
Its
presence or absence
significantly affects a host of en erg y and mass
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fluxes th a t control h e a t tra n s fe r rates and sea ice form ation. The evolving
snow pack also exerts a n influence over biogeochemical cycles w hile th e
spring m elt provides the m arin e cryosphere w ith a significant source of
fresh w ater.
Snowfall over th e m arine cryosphere is dom inated by a bimodal d istrib u tio n
p attern. A large am ount of snow falls during a u tu m n freeze up an d a g a in
d uring th e onset of spring. This is principally a re su lt of increased s to rm
activity du rin g these tim es. The intervening polar w inter is c h a ra c te riz e d
by lower precipitation, p rim a rily because little energy is in the sy ste m
(Figure 2.5). As a result, snow on sea ice often contains a complex m ix tu re
of layers, each physically distinct due to th eir differing evolutionary
histories.
Snowpack conditions at any given tim e are a resu lt of th is
history, w hich begins w ith th e form ation of snow in the atm osphere.
70
□
60
1990
m 1991
~
£
50
E
I 40
5
:!•
(uj 30
£
m
1992
m
1993
■
1994
20
10
0
•2%
tt.
£
2
o.
o
a.
<
CO
>
o
z
Month
Figure 2.5 Precipitation recorded at Resolute Bay.
(courtesy Arvids Silis, p ersonal
com m unication)
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
F orm ation
At approxim ately -5°C, ice nucleation w ithin a cloud in itiates the process of
snow form ation. A snow flake fuses together through interlocking th e arm s
of ice particles a n d more im portantly, by w ater vapour condensing onto the
interlaced particles (M ason, 1992).
In the special case w here a larg e
am ount of frozen w a ter dom inates the snowflake grow th process, a h a rd
snow pellet term ed graupel forms (Schem enauer et al., 1981). W hen the
snowflake reaches th e ground it is term ed a snow grain.
Evolution
For SWE derivation, evolution of the snowpack is p rim a rily concerned w ith
un d erstan d in g how snow cover variations affect m icrow ave em ission.
Barber et al. (1995) provided statistical characterizatio n of a seasonally
evolving snow cover based on crystal geometry.
In the colder regim e (-
20°C), only large basal (bottom of the snow cover) crystals could be separated
from th e re st of th e snowpack. However, in the w arm er p a rt of th e season (5°C), the basal layer w as distinguishable from a n original snow layer w ith
medium sized crystals and a new snow layer consisting of sm aller crystals
which had been deposited over the field cam paign.
As te m p e ra tu re s
approached 0°C, B arber et al. (1995) also noticed g ra in sizes grew larg er in
the 9 to 18 cm region above the sea ice.
F urtherm ore, grow th in the new
snow layers proceeded m ore rapidly, w ith grain sizes doubling in a m a tte r
of days.
Typically, evolution of th e snow results in slight increases in density as well
as grain size. W hen snow is transported by saltation th e crystal's dendrites
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b reak off due to frictional forces, causing grain com paction upon se ttlin g
(Male, 1980). Solidification can also occur through vapour tra n s fe r a n d
especially m elt-freeze cycles.
B arber et al. (1995) showed how a se a so n a lly
evolving snow pack over first-year ice increased from a n in itial v alu e of 250
k g ‘ n r 3 to 375 kg«m*3. The evolutionary process is largely controlled by
m etam orphic action w ith in th e snowpack.
M etam orphism in Dry Snow
Upon deposition, snow grain evolution begins w ith the continual process of
m etam orphism .
T herm odynam ically, the system seeks a s ta tu s of free
energy equilibrium .
surface
area
Physically, this is achieved w hen the s n o w g ra in
to volum e
ratio
is m inim ized
(L angham ,
1981).
A
tem p eratu re g rad ie n t is constructed between the top of one sn o w g rain a n d
the bottom of a neighbour to facilitate the equilibrium process.
An
associated vapour g rad ien t tra n sfe rs w ater from h igh p re ssu re to low
pressure sections of th e snow pack (in term s of tem p era tu re , from w a r m
sections to cool sections).
W hen the snow pack rem ain s dry, w ater vapour m ass is tra n s fe rre d fro m
the top of one g ra in to th e bottom of an o th er grain th ro u g h a "h an d -to hand" process (Yoshida, 1955).
Since th e effective conductivity o f ice is
about one h u n d red tim es th a t of a ir (Colbeck, 1987), conduction th ro u g h th e
snow grains is likely m ore im p o rtan t th a n vapour tra n s p o rt th ro u g h th e
air. This is especially tru e in cold, dense snowpacks (Fukusako, 1990). T he
determ ining factor leading to grow th of highly faceted “kinetic form ” g r a in s
(Figure 2.6), or m ore rounded “equilibrium form” grains (Figure 2.6), is th e
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stre n g th of th e tem p e ra tu re g rad ie n t (B urton et al., 1951). T he s e p a ra tio n
point is som ew here betw een 10 an d 2 0 °C *m 1 (Colbeck, 1986; L aC h ap elle
and
A rm strong,
1977), w ith
equilibrium
grains
evolving below th e
threshold and kinetic grain s grow ing above. This threshold in creases w ith
increasing density because cry stal grow th ra te s are larg er w ith th e g re a te r
separation p resen t in low density snow (Colbeck, 1983).
Y oshida (1955)
experim entally determ ined te m p e ra tu re gradients of 15 to 7 0 °C » n r1 g rew
faceted
crystals,
while
te m p e ra tu re
gradients
of
6°C • n r 1 fo rm ed
In th e equilibrium form of dry snow m etam orphism
relatively s m a ll,
equilibrium crystals.
rounded crystals form.
G enerally, the dendrites begin to decom pose into
sm a ller fragm ents and large crystals grow a t the expense of s m a lle r
crystals (L angham , 1981). E ventually larg er grains form bonds th ro u g h
sin terin g (de Q uervain, 1963), w hich increases snowpack stre n g th .
equilibrium fo rm
V,
"
.1
kinetic grow th fo rm
'!
I
-
»
* ».
-
\
I*.. *
: I
,
Figure 2.6 Equilibrium and kinetic growth snowgrains.
The kinetic form of dry snow m etam orphism leads to large, g eo m etrically
sh ap ed crystals.
These cry stals develop w hen a significant te m p e ra tu re
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
g rad ie n t is established and m a in ta in e d for an extended period of tim e.
U n d er larger gradients m ass is rem oved from the lower portions of a
snow pack, hastening the developm ent of depth hoar crystals.
These
crystals are permeable, low in d e n sity and w eak of strength. A kitaya (1975)
h as determ ined two distinct types of depth hoar: "solid-type", w ith platy or
c o lu m n ar shape, have sh arp edges and flat surfaces.
These cry stals a re
fairly firm and may appear to be com pacted fine-grained snow. Conversely,
"skeleton-type" depth hoar is highly faceted w ith stepped or ribbed su rfa c e s.
T hese crystals are larg er th a n "solid-type", but have very little s h e a r
stre n g th .
Akitaya
also
discovered
the
"solid-type"
form ed
w h en
tem p e ra tu re gradients are less th a n 25°C «nr1 and "skeleton-type" form ed
w hen th e gradient is g reater th a n 25°C »nr1.
M etam orphism in Wet Snow
M etam orphic action in wet snow packs causes rapid a lteratio n in the
snow grains and eventually destroys the layered character of a snow pack
(B arber et al., 1994). W ith low liquid w ater contents g r a in s can n o t exist
individually and join to form c ry sta l agglom erates (Colbeck, 1987).
T his
typically occurs when bulk w a ter contents are less th an approxim ately 1%.
T his is known as the pen d u lar reg im e (Figure 2.7). D ensities in th e g ra in
clusters m ay be upw ards of 600 k g #m*3, but overall snowpack d en sity is fa r
less due to w ater pockets and continuous a ir spaces ru n n in g th ro u g h the
snow pack (Colbeck, 1986). The ex act n a tu re of crystal grow th in low liquid
w a te r content snow is a m atte r of som e uncertainty, since m e a su re m e n t is
difficult.
However, it appears th e rm a l diffusion plays a more im p o rta n t
role, while conduction plays a less im p o rtan t role (Colbeck, 1987). Liquid
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
w ater between the snow grains
provides
channels
w hereby
th e r m a l
diffusion could be accentuated.
Pendular Regime
Liquid Water Content <7% by Volume
(— Ice
A ir
A ir
W ater
Figure 2.7 Sketch of snowgrains, air and water distributions in the pendular regime, (adapted from
Garrity, 1993)
In th e funicular regim e liquid w a te r contents are g rea ter th a n 7% (F ig u re
2.8).
Surface tension is no longer sufficient to keep snow grains bonded. A s
a resu lt, th ey norm ally form larger, m ore spherical shapes (Colbeck, 1987).
A ir spaces become discrete in n a tu re , w hile w a ter channels m ay r u n
continuously through the snow pack.
W hen
th e snowpack
becom es
sa tu ra te d w ith w ater crystal grow th leads to individual spherical p articles.
T he resu ltin g slushy m ixture is w eak in stre n g th (Colbeck, 1987). Since
sa tu ra te d snow conditions were no t encountered in C-ICE 96, they m erit no
fu rth e r attention.
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Funicular Regime
Liquid Water Content >7% by Volume
A ir
W ater
Ice
Aii
Air
Figure 2.8 Sketch of snowgrains, air and water distributions in the funicular regime, (adapted from
Garrity, 1993)
I t can be appreciated th a t the evolving snow-sea ice m ixture is com plex.
F rom in itia l form ation, both snow a n d sea ice undergo a n extensive p a tte rn
of m etam orphism . A t any p a rtic u la r tim e energy and m ass tra n s fe r in th e
m a rin e cryosphere is linked to th e physical state of the snow -sea ice m a s s .
By governing these energy exchanges over th e seasonal evolution, th e sn o w
covered sea ice volum e controls m an y of the key physical an d ecological
aspects of the m arine cryosphere. T he next section will exam ine som e of
th e processes in tim ately related to a n evolving snow covered sea ice volum e.
2.3 The R ole of Sea Ice and Snow in M arine C ryospheric
Processes
T he m arine cryosphere is d om inated by th e seasonal evolution of a snow
covered sea ice volume, w hich acts to (a) increase th e surface albedo a n d (b)
reg u la te h e a t flows across the m a rin e cryosphere.
The snow cover is
p a rticu la rly im p o rtan t in controlling energy and m ass tra n s fe r ow ing to its
extrem ely high albedo an d low th e rm a l conductivity.
As a re s u lt, th e
22
Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission.
presence or absence of snow dictates light tran sm issio n and in flu e n ce s
ecological life in the m arine cryosphere.
F urtherm ore, the low th e r m a l
conductivity of snow governs the se a ice accretion and decay process, w h ic h
in tu rn affects brine fluxes an d the therm ohaline
circulation.
T he
following section will outline som e of the relevant cryospheric resp o n se s
governed by a snow covered sea ice volume and how changes in snow cover
could affect them .
Since this thesis is concerned w ith snow w ater equivalence m o n ito rin g ,
snowpack effects will be em phasized, w ith a sm aller focus on sea ice.
2.3.1 In flu en ce o f Snow O ptical P rop erties
The optical snow properties drive m etam orphism and energy fluxes w ith in
the snow pack (Mellor, 1965). For instance, the shortw ave surface en erg y
balance is largely controlled by th e albedo of snow.
Shortw ave ra d ia tio n
refers to energy em itted from th e Sim (Ahrens, 1991) and it exhibits a
m arked seasonality in th e Arctic. Irrad ian c e at the top of the a tm o sp h e re
can be over 800 W « m 2 a t solar noon d u rin g the sum m er, but drops to zero
d u ring the periods of polar darkness (Barber, 1993; Oke, 1987).
W hen shortw ave energy is incident upon th e m arine cryosphere, it will be
reflected, tran sm itte d , or absorbed. The high albedo of snow is th e p rim a ry
factor restrictin g shortw ave energy exchanges between th e ocean an d th e
atm osphere because a large percentage of insolation is reflected before it
reaches the w ater.
Albedo, a, is defined as the portion of d o w nw ellin g
rad iation which is reflected from a surface [2.5].
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
KT
row
[2' 51
“ = kT
w here K t is upw elling shortw ave radiation an d
Ki
is dow nw elling
shortw ave radiation.
Seasonal albedos in the Arctic are largely dep en d en t on the n a tu re of the
snowpack, w ith larg er grains and increasing den sity both reducing the
albedo (Table 2.1). Evolution of the snowpack ten d s to low er albedo owing to
increased w a ter in liquid phase, increased g ra in size a n d the transitio n of
snow grains to the "equilibrium form".
The n e t effect is absorbed energy
over a snow covered surface m ay be an order of m ag n itu d e lower th an th a t
of an open w ater surface (Barber, 1993; Grenfell an d Perovich, 1984; M ellor,
1977). T hus, sea ice albedo is a critical com ponent in determ ining the
energy balance of th e m arine cryosphere.
T ab le 2.1 Typical albedo values in the marine cryosphere.
Ocean State
Snow Cover State
Albedo
FYI
D ry
0.90
FYI
W ater saturated
0.70
FYI
Meltponds
0.60
Thin Ice (<30cm)
None
0.40
FYI
None
0.20
W ater
N /A
0.03-0.10
adapted, from Langleben (1969); Grenfell (1979); Barber (1993)
Shortwave energy not reflected m ust be t r a n s m i t t e d or absorbed, in accord
w ith the conservation of energy law.
A lthough tra n sm issio n can be
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approxim ated by Beers law [2.6], B arber (1993) has show n tra n s m is s io n
through snow is alm ost double th e predicted value a t the low er (0.5 pm )
portion of shortw ave spectrum .
Mellor (1965) conducted
a series of
experim ents on the absorption properties of snow in the visible portion of
the EM spectrum . He found absorption increased w ith in cre ased g ra in size
and density and was also frequency dependent, w ith in c re a se d absorp tio n
a t higher wavelengths. Ledley (1991) modelled the effect of c h a n g in g snow
conditions on the shortw ave energy balance.
W ith no snow cover, m e a n
absorbed shortw ave energy a t 75°N was 49.5 W*m*2. W ith th e full snow
model (28cm), this value dropped to 35.3 W «nr2.
K l z = K l s e~az
[2.6]
w here K i z is the shortw ave radiation a t depth 'z' in th e snow , K l s is
shortw ave radiation a t th e snow surface and a is the e x tin ctio n coefficient.
The extinction coefficient is a function of snow depth, density a n d g ra in size
(Barber, 1993; Oke, 1987; Mellor, 1965).
Several lifeforms, such a s th e epontic (sub-ice) com m unities, a re governed
by the optical properties of snow (Welch and B ergm ann,
1989).
T hese
groups utilize photosynthetically active radiation (PAR; 0.4 to 0.7 pm )
tran sm itte d through th e snow pack to grow. Welch et al. (1991) illu s tra te d a
negative exponential rela tio n sh ip between snow depth a n d u n d e rly in g
chlorophyll production (F igure 2.9). As a result, slig h t c h a n g e s in snow
cover will be am plified su b sta n tia lly in the algae life cycle.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.25
2.00
„
°oo
1.75
00
& 1.25
=
1.00
-1
.50
.25
0
5
10
15
20
25
30
35
40
45
50
Snow Depth (cm)
Figure 2.9 Snow-chlorophyll relationships in the Arctic. Station 1 R^= 0.84; Station 2 R^= 0.86.
(adaptedfrom Welch et al., 1991)
H igher levels of the trophic sy stem are also governed by snow cover
am ount. Ringed seals (Phoca h isp id a ) utilize the snow cover on sea ice for
protection from predation an d th e h a rs h Arctic w inter (Sm ith a n d S tirlin g ,
1977). Jew ett and F eder (1980) hypothesize the disap p earan ce of bottom
feeding fish populations along co astal shelves is attributable to in c re a s e d
w a te r tem peratures re su ltin g from lack of snow.
2J3J2 Influence o f S now P h y sic a l P ro p e rtie s
Sea ice grow th and decay is largely a function of the overlying snow cover,
owing to the low th e rm a l conductivity of snow, Xse. Snow conductivity is
highly dependent upon
tem p eratu re as well.
snow
density w ith
some
dependence
upon
As d en sity increases, the n u m b er a n d size of
in te rg ra n u la r contacts increase, re su ltin g in higher values of th e r m a l
conductivity (Fukusako, 1990). Conversely, as tem p eratu re in creases th e
th e rm a l conductivity decreases (S akazum e and Seki, 1978).
L angham
(1981) attributes the negative rela tio n sh ip to vapour diffusion along a n
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
associated tem p eratu re g rad ien t. The accom panying energy loss th ro u g h
la ten t h e at of evaporation w ould affect
The net effect is th a t
is
approxim ately 8 to 10 tim es sm aller th a n the therm al conductivity of sea ice
(Kotlyakov and Grosswald, 1990). Since ice growth is a function of h e a t loss
to the atm osphere (Maykut, 1986), lower h e at tran sfer rates w ill reduce sea
ice thickness. Thicker and less dense snow covers increase the in s u la tin g
properties of snow, fu rth er re stric tin g energy tran sfer across the m a rin e
cryosphere.
B oth modelling (Ledley, 1991) a n d em pirical studies (Brown an d Cote, 1992;
Nakawo and Sinha, 1981) have dem onstrated increased snow cover leads to
decreased ice thickness.
Ledley (1991) used a coupled energy balance
clim ate-therm odynam ics sea ice (CCSI) model to study the effect of snow
a n d sea ice on clim ate. A fter ru n n in g th e model for 20 years, the no snow
scenario resulted in a sea ice thickness of 2.85m, while the full snow m odel,
w ith an 80 cm depth, concluded w ith a 2.76m ice depth. Brown and Cote
(1992) m onitored sea ice thickness in relation to snow cover a t 4 sites acro ss
th e C anadian Arctic A rchipelago.
In all cases an increased snow cover
lead to decreased ice thickness a n d vice versa (Table 2.2).
Table 2.2 Linear trend analysis of snow depth and ice thickness.
Site
Period
Ice (cm » y rI)Snow (cm^yi-1)
A lert Inlet
1956-87
-0.71
+0.43*
A lert (Dumbell Lake)
1957-87
-0.90*
+0.30
E u rek a
1952-89
-0.00
+0.04
Mould Bay
1954-89
-0.29
+0.20
Resolute
1952-89
+0.99*
-0.33*
* indicate statistical significance a t a=0.05 (adapted from Brown and Cote, 1992)
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
D uring sp rin g th e low therm al conductivity of snow lim its h e at tra n s fe r
from the relatively w arm atm osphere to the relatively cool ocean (opposite to
the fall scenario).
F urtherm ore, the high albedo m entioned previously
reflects m uch of th e incom ing solar radiation, fu rth e r slowing m elt.
In
addition, the snow pack acts as a significant h e a t sink owing to its larg e
latent h e at of fusion (B arry et al., 1995). Com bined, these physical snow
properties reduce th e ra te of ice m elt in the spring.
By regulating sea ice grow th and decay snow contributes to determ ining the
brine flux of the A rctic Ocean.
Cold, dense w a te r expelled d u rin g ice
form ation sinks from the ocean surface and is replaced by w arm er, less
dense w a ter from th e surrounding seas (A agaard a n d C arm ack, 1989).
This perpetuates a global oceanic conveyor belt, often term ed th erm o h alin e
circulation, th a t m oves cold w ater to the equator a n d w arm w ater to the
poles. A agaard a n d C arm ack (1989) suggest the "G reat Salinity A nom aly"
of the late 1960s-1970s (Dickson et al., 1988) m ay be a n exam ple of past
clim atic eras in w h ich the oceanic conveyor belt w as suppressed by large
influxes of fresh w a te r.
Increased snow m elt could dilute the brine rich
w aters such th a t th e y lack the density to sink an d m a in ta in the conveyor
belt.
Paleoclim atic analysis by Broeker and P en g (1989) an d modelled
resu lts from B ryan (1986) and Manabe and Stouffer (1988) im ply th e no n ­
circulating conveyor belt system is also a stable ocean-atm osphere regim e.
Therefore, in creased snow cover could affect the c u rre n t dom inance of the
deep w ater conveyor b e lt circulation p attern in th e future.
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This section h a s show n the physical snow properties ex ert su b sta n tia l
influence over conditions in the m arine cryosphere.
The n a tu re of the
snowpack contributes to defining everything from sea ice grow th an d
ablation through to th e developm ent of life in the m arine cryosphere. W ith
an indirect influence on th erm o h alin e circulation, the snow also affects
clim ate on a global basis. It sta n d s to reason th a t changes in th e physical
m ake-up of th e snow w ould a lte r th e n a tu re of its in te rac tio n w ith the
m arine cryosphere. As a resu lt, conditions in this region w ould likely be
altered. The following section provides some evidence th a t conditions in the
m arine cryosphere a re a lre a d y changing.
2.4 E vidence for C hange in th e M arine C ryosphere
N ansen (1902) w as one of the first research ers to theorize ch an g es in the
physical Arctic system could affect clim ate. A lthough th ese claim s have
yet to be proven conclusively, a grow ing body of lite ra tu re points to th e fact
th at physical p a ra m e te rs an d tem p eratu re in m arine cryosphere are
changing.
This section exam ines th a t theory, specifically focusing on a ir
tem perature a n d sea ice extent.
One of the m ajor reaso n s th e m arin e cryosphere is considered susceptible
to clim ate change is due to the sea ice-albedo feedback m ech an ism (Rind et
al., 1995). As te m p e ra tu re s increase, the snow an d sea ice volum e w ill
likely m elt, reducing the albedo. W ith lower albedos, more en erg y w ill be
absorbed, 'feeding back' in to th e original w arm ing cycle.
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C hapm an an d W alsh (1993) have collected satellite based te m p e ra tu re
m easurem ents confirm ing a rise in tem perature over the Arctic (Figure 2.10
and 2.11). A lthough w in te r tem p eratu re anomalies are higher, correlation
analysis found a s ta tis tic a lly significant decrease in ice ex ten t only in
sum m er.
O th e rs have noted a sim ila r w arm ing tre n d in ocean w ater
(Carmack et al., 1995; M ikhalevsky et al., 1995; Aagaard an d Carmack, 1994)
but have not correlated it to ice extent.
Temperature
Anomaly (°C)
15
Figure 2.10 Observed treads of Arctic winter mean temperatures from 1961-1990 (adapted from
Chapman and Walsh, 1993)
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Temperature
Anomaly (°C)
0.75
m
1
0.75
F ig u re 2.11 Observed trends of Arctic sum m er mean temperatures from 1961-1990 The
warming trend is not as apparent, (adapted from Chapman and Walsh, 1993)
Gloerson and Campbell (1991) reported a statistically significant decrease in
sea ice ex te n t was observed in th e S canning M ultichannel M icrowave
R adiom eter (SMMR) data (Table 2.3). Johannessen et al. (1995) extended
the tim e series to 1994 w ith the SSM /I sensor.
Their -0.54 x 106 k m ^ y r'1
decrease in ice extent, statistically valid a t a p-value < 0.05, is significantly
g reater th a n the -0.0315 x 106 km2yr*1 decrease observed by Gloerson and
Campbell.
Unfortunately, neither stu d y associated this w ith air or ocean
tem peratures.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2 3 Trends in Arctic sea ice covers, 1978-L994.
V ariable
Slope
S tandard Deviation
(lO Skn^yi-1)
( lO f ik m V 1)
%Change per
decade
p-value
Sea Ice Extent
1978-1987
-0.032
0.008
-2.5
<0.05
1987-1994
-0.054
0.017
-4.3
<0.05
1978-1987
-0.030
0.009
-2.8
<0.05
1987-1994
-0.049
0.016
-4.5
<0.05
Sea Ice Area
Data from 1978-1987 is from the SMMR sensor and those for 1987-1994 are from the SSM/I
sensor, (adapted from Johannessen et al., 1995; Gloerson and Campbell, 1991)
A com pilation
of m odelling
studies
predicts
the
observed
rise
in
tem perature will continue to increase with a CO2 doubling (F igure 2.12).
These models, term ed g e n eral circulation models (GCMs), can be used to
predict future changes
in the climate over the m a rin e
cryosphere.
Typically, they segm ent th e E a rth into a series of grid cells.
In each cell
they use num erical eq uations to solve for conservation of m a ss, energy an d
m om entum both w ith in a grid cell and between grid cells (H endersonSellers and Robinson, 1987). They also consider physical processes such as
the sea ice-albedo feedback m echanism . However, caution m u st be tak e n in
in terpreting resu lts from th ese models, as even the m ost advanced m odels
are relatively coarse a n d often tre a t the physical processes in a highly
sim plified m anner.
F or instan ce, a comparison of th e snow -clim ate
feedback m echanism produced by 17 GCM models show ed very different
results.
Some m odels classified th is as a weak negative feedback, w hile
others classified it as a stro n g positive feedback (Cess et al., 1991).
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 2.12 Composite greenhouse wanning projections as predicted by seven Global Climate Models is
presented as the difference between current conditions and those under doubled CCH {adapted from
http://www.circles.org/cdroms/CIESIN/HTMLAOTG/5ArcRole/E3Scen.htm)
The m odelling stu d ies associate su b sta n tia l a lte ra tio n s in precipitation
p attern s (Figure 2.13) w ith increasing tem perature. W ith increased snowfall
m any of the processes controlled by the snow cover would change.
For
example, a n increase in snow cover would increase th e albedo of the m arine
cryosphere, which may delay ice m elt in the s u m m e r
Conversely, the thicker
snowpack would also p rev e n t the ice from growing th ick er in the fall. The
thinner sea ice may then m elt sooner, counterbalancing the higher albedo.
In the context of this thesis, the predicted precipitation change is one of the
most critical results presented in this chapter. It provides clear impetus th a t
improved methods of m onitoring the snow cover are needed.
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Annual Precipitation
Precipitai
Increase (2xCQj / CC^
II
Figure 2.13 Composite of greenhouse precipitation projections as predicted by seven Global Climate
Models is presented as the ratio between a doubled CO2 scenario and current conditions. (adapted from
http://www.circles. org/cdroms/CIESIN/HTMLAOTG/5ArcRole/E3Scen. him)
2.5 C o n clu sio n s
The m arine cryosphere is characterized by the dynam ic properties of a snow
covered se a ice volum e.
From initial form ation, th e s e p roperties are
continually evolving a n d altering the environment around them . In particular,
energy fluxes a re controlled by the high albedo of snow , while physical
processes are intim ately linked to the snow's low therm al conductivity. Small
perturbations in the physical properties of a snow cover m ay bring about large
changes in local clim ate and ecology. If these snowpack changes altered the
therm ohaline com ponent of ocean circulation, the global clim ate could be
affected. Although few studies exist to document in te ra n n u a l variability in
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
th e snow pack, se v era l w orks p o in t out th e a ir te m p e ra tu re in th e A rctic h a s
w arm ed in th e p a s t th irty y ears. F urtherm ore, significant decreases in sea ice
e x te n t h av e also b e e n n o ted .
O n th e b a sis of c u rre n t g e n e ra l c irc u la tio n
m odels, it se em s lik e ly th is w a rm in g w ill co n tin u e, w ith a n a sso c ia te d
in cre ase in p re c ip ita tio n .
T h e se m odels p o in t o u t th e n eed to find b e tte r
m ethods to c h a ra c te riz e snow conditions across the e n tire m arin e cryosphere.
T he use of m icrow ave ra d io m e try for SW E retriev als is one su ch m ethod a n d
th e n ext c h a p te r o u tlin e s th e p rinciples b eh in d using m icrow ave ra d io m e try
for SWE derivation.
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 3 - B ackgrou n d
3.1 Introduction
Im proved ch aracterization of the snow pack is necessary to u n d e rs ta n d
clim atic conditions and tren d s (B arry et al., 1995). The large sp a tia l scales,
hostile environm ental conditions and relatively sparse in situ observational
netw orks suggest rem ote sen sin g is a more
technique to pursue.
effective m e a s u re m e n t
M icrowave sensors, with all-w eather,
all-lig h t
capabilities, offer g rea ter flexibility in m onitoring the cryospheric snow
cover, since cloud and darkness often prevail in this region.
A lthough severed indices of snow cover am ount could be m onitored (for
instance snow depth). SWE is the best estim ator because it accounts for
changes in snow density as well as depth. On the basis of m e a su rin g snow
depth alone, a one m etre thick, 400 k g * n r3 snow pack would a p p e a r
significantly different from a two m etre thick, 200 kg* n r 3. H ow ever, th is
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
difference could be chiefly a product of different evolutionary h isto ries.
th e 200 kg«m*3 snow pack is dom inated by new snow fall,
If
it will soon
increase in density (Barber, 1993) and begin to settle. C onversely, SWE can
estim ate snow am ount w ith o u t reg ard to the evolutionary p a tte r n of the
snow, as the two situ atio n s described above would yield th e sam e SWE
value.
Furtherm ore, th e passiv e microwave signature is in tim a te ly linked
w ith snow w ater equivalence (F igure 3.1), a t least over te r r e s tr ia l su rfa ce s
(C hang et al., 1987). Several rese a rc h e rs have successfully e x tra c te d SWE
over te rre stria l surfaces, ra n g in g from the C anadian p ra irie s (Goodison et
al., 1990), to m ountain b asin s (C hang et al., 1991) and forested la n d (H all et
al., 1982).
240
18GHz
g220
2IGHz
I 200
H 160
5 140 -
5 120
•
100
37GHz
0
10
20
30
40
50
60
SW E (cm)
70
80
90
100
F igu re 3.1 Calculated brightness tem perature as a function of snow w ater equ iv alence.
Calculations are based on horizontal polarization and a 50° incidence angle, (adapted
from Chang et al., 1987)
In w h at follows, the in te rac tio n of microwave energy a n d th e p h y sical
properties (covered in section 2.2) will be discussed.
Both se a ice, w h ere
m icrowave em ission o rig in ates from and the snow, w h ere a tte n u a tio n
occurs,
will be covered h ere.
Finally, historical use o f m icrow av e
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
radiom etry in the m arin e cryosphere is detailed as a p recedent for u s in g
passive microwaves in th is work.
3.2 M icrow ave In teraction Theory
Microwave energy occupies frequencies from 1GHz to 300GHz in th e
electrom agnetic
spectrum .
W ithin
this
range,
passive
m icrow ave
m easurem ents record em ission from a target in selected "a tm o sp h e ric
windows" (Figure 3.2) w here attenuation is m inim ized.
C om m on passive
microwave frequencies a re 10GHz, 19GHz, 21GHz, 37GHz, 85GHz a n d
94GHz (Carsey, 1992).
Wavelength (cm)
0.3
100
0.2
0.15
water vapour absorbtion bands
0.12
0.10
oxygen absorbtion bands
£
70
35 GHz
window
60
85GHz
window
20
40
60
80
100 120 140 160 180 200 220 240 260 280 300
Frequency (GHz)
F ig u re 3.2 Atmospheric windows facilitating microwave remote sensing, (adapted fro m
Ulaby et al., 1986)
O ver the m arine cryosphere microwave em ission is a function of ice a n d
snow physical properties.
Sea ice is known to be the m ajo r source of
microwave em ission in dry snow covers (Eppler et al., 1992). How ever, F Y I
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
an d MYI have s t r ik in g ly different emission, properties ow ing to th e ir
in te rn a l structures. Brine d rain ag e d u rin g the w arm su m m er creates a
porous stru ctu re in MYI w hich is a n excellent volume sc atterin g so u rce
(G renfell, 1992). This accounts for the decreased em ission noted in MY I.
Furtherm ore, Eppler et al. (1992) state variability in the physical p roperties
of m ultiyear ice leads to m uch larg e r varian ce in em ission as com p ared
w ith FYT. Interest in the snow relates to its role as the dom inant m e d i u m
of m icrowave attenuation in m icrowave em ission from sea ice (E ppler et
al., 1992). Both Lohanick (1993) and L ohanick and G renfell (1986) have
observed increasing grain sizes lowered brightness te m p era tu re s (Table
3.1). Table 3.1 also points out increases in SWE lowers em ission, w hile
w a ter in liquid phase increases em ission.
By increasing th e v o lu m e's
em ission, liquid w a ter complicates th e SWE retrieval process.
T ab le 3.1 Response of Microwave Emission W ith Various Snow Geophysical Properties
Geophysical Property
Microwave Emission Response
Snow W ater Equivalence
Decreases as SWE increases
Grain Size
Decreases w ith increasing grain size
Liquid W ater Content
Increases rapidly with increasing
liquid
phase
w ater i n
(adapted from Hallikainen and Jolma, 1986)
A lthough em ission is a function of the snow covered sea ice v o lu m e's
physical tem perature, the volum e em its only a fraction of th e energy a
perfect em itter would radiate. T his fraction is described by th e em issivity.
The sig n atu re recorded by a m icrowave
radiom eter, expressed as a
b rig h tn ess tem perature (T b), is therefore a function of te m p e ra tu re a n d
em issivity [3.1]:
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T B(f,0i) = eTsi
[3.1]
w h ere f is frequency, 0j is th e incidence angle, e is em issivity (0<e<l), Tsj is th e
p h y sical te m p e ra tu re of th e snow-ice com bination in Kelvin.
In th e context of m icrow ave radiom etry, th e average em issivity is a fu n ctio n of
freq uency, p o larizatio n , incidence angle, p h y sical com position of th e se a ice
(i.e. ice type) an d s p a tia l p a tte rn in th e field of view (U laby et a l., 1986). W ith
re s p e c t to a p a rtic u la r freq u e n cy a n d p o la riz a tio n , v a ria tio n s in T b a re
p rim a rily due to v a ria tio n s in em issiv ity a n d secondarily to v a ria tio n s in th e
p h y sical te m p e ra tu re (Steffen et al., 1992). In FYI, G renfell et al. (1988) h a v e
show n FY I h a s a fa irly sta b le e m issiv ity once it th ick en s to 10m m (F ig u re
3.3).
1 00 ”
1 * 1 * I
* 1 1 * * 1 * IQ^lI
11
' ‘ * ‘ 1 1 ‘ ‘ * I
0.95 0.90 0.85 >.
i
;f 0.80 %
18GHz
37GHz
§0.75 -
90GHz
LU
0.70 0.65 0.60 4
0.55
4
0
Figure 3.3 Emissivity as a function of ice thickness, (adapted from Grenfell et al., 1988)
In d efining a rela tio n sh ip b etw een SW E a n d T b, we tak e a d v a n ta g e of th e fa c t
t h a t e m issiv ity d e c re a se s w ith in c re a s e s in th e s c a tte rin g p o te n tia l o f a n
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
overlying snow cover (H ay k in et a l, 1994). The n a tu re of th e sc a tte rin g can be
described by th e principles of volum etric scattering.
Scattering in Snow Covered First-year Sea Ice
T h e p a rtia l frac tio n s of ice, a ir, b rine an d liq u id w a te r c re a te a complex
m edium in w hich e m itte d en erg y is n o t sim ply re fle c te d o r red irected , b u t
r a th e r s c a tte re d in a v a rie ty of directions. W ith re s p e c t to frequency an d
p olarization, sc a tte rin g is d ep en d en t on the physical s tr u c tu r e a n d electrical
p ro p e rtie s of th e sn o w covered-sea ice volum e (C a rs e y , 1992).
In dry
snow packs volum e s c a tte rin g dom inates a t frequencies above 20G H z (Figure
3.4) (H allikainen, 1986). S c a tte rin g stre n g th is p ro p o rtio n a l to th e dielectric
discontinuities w ith in th e volum e and density of th e em b e d d ed particles. The
a n g u la r s c a tte rin g p a t t e r n is governed by ro u g h n e ss o f th e d isco n tin u ity
su rface, d ielec tric p ro p e r tie s of th e volum e a n d g e o m e tric size of th e
sc atterin g p articles re la tiv e to th e im pinging m icrow ave e n e rg y (Ulaby et al.,
1986). A good e stim a to r for volum e scatterin g w ith in a sn o w p ac k is provided
b y M arsh all a n d G unn (1952). I t is applicable w ith fre q u e n c ie s over 30GHz
a n d snow grain p a rticle s less th a n 1.5mm in diam eter [3.2].
Emitted
Energy
F ig u r e 3.4 Volume scatterin g phenom ena, (adapted from Ulaby et al., 1986)
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CTbs = 1 6 7 t5 ^ -
[3 - 2 ]
w here a ts is th e m ag n itu d e of b a c k sc a tte r in dB, r, is th e size of sn o w g rain s
a n d Xq is th e w avelength of free space.
Sea Ice a n d Snow Dielectrics
T he com plex d ielectric c o n s ta n t [3.3] is u se d to d e scrib e th e s c a tte r in g
p ro p erties of a snow covered sea ice volum e. Perm ittivity, e', c h arac terize s th e
re la tiv e p e rm ittiv ity w ith re sp e c t to free space (£air= l)» w h ile th e dielectric
loss, e", defin es th e e le c tro m a g n e tic loss of th e m a te ria l.
P e rm ittiv ity
describes w h a t h ap p en s to EM e n e rg y w hen i t im pinges u p o n a b o u n d a ry .
Loss describes th e electro m ag n etic lo ss once energy h a s p e n e tr a te d into th e
m a te ria l.
T he to ta l e le ctro m ag n e tic loss (H a llik ain e n a n d W in e b re n n e r,
1992) is a com bination of a b so rp tio n loss (tra n sfo rm a tio n of e n erg y in to
a n o th e r form ) a n d s c a tte rin g loss (en erg y deflected to tra v e l in d irec tio n s
o th er th a n incident).
e* = e' + je"
[3.3]
W here j = -J- 1
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T he com plex dielectric c o n s ta n t of sea ice, es i, is m a in ly a fu n ctio n of th e
p a r tia l fractions of ice, a ir a n d b rin e w ithin th e s c a tte rin g volum e, relative to
th e incidence angle (Ulaby et al., 1986). M odelling th e p e rm ittiv ity of sea ice,
s si, h a s focused on coupling b rin e volum e w ith th e p e rm ittiv ity of sea ice,
since b rin e perm ittiv ity is m u ch higher th a n ice or a ir p e rm ittiv ity (M atzler
a n d W egm uller, 1987; C um m ing 1952). S everal re s e a rc h e rs h a v e provided
m odels (e.g. Arcone et al., 1986), w ith re su lts from H o e k s tra a n d Cappillino
(1971) (Figure 3.5) shown below [3.4].
e’ •=.— —
~ 1- 3 V b
[3.4]
6
W
5
(Z 4
>»
3
2
frequency = 9JSGHz
1
1.0
1.1
1.2
l/(l-3Vb )
13
Figure 3.5 Measured permittivity of artificially grown sea ice as a function of brine volume fraction.
(adaptedfrom Hoekstra and Cappillino. 1971)
U sin g th e sam e theory em ployed in calculating [3.4], H o e k s tra a n d Cappillino
(1971) provide [3.5] as a n approxim ation for th e dielectric loss of se a ice.
£ si =
Vbeb
[3.5]
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T he complex dielectric co n stan t of snow, £s, is m ainly a function of th e
p a rtia l fractions of ice, w ater, b rin e an d a ir w ithin the sc atterin g v o lu m e ,
relativ e to th e incidence angle (U laby et al., 1986). In the up p er p ortion of a
d ry snowpack, perm ittivity can be described solely by the relative po rtio n of
ice a n d air.
Since both p a ra m e te rs a re consistent over the m icro w av e
portion of the EM spectrum , a sim ple equation [2.6] of the type rep o rte d by
H allikainen et al. (1986) accurately describes
(Figure 3.6).
D ielectric
loss in a dry snow pack can be com puted based on the models sup p lied by
T inga et al. (1973) [3.7].
eds = °-51 + z 88Pds,Pds > 500kg • m -3
Eds = 1 + l^pdspds < 500kg • m -3
w here pds is th e density of d ry snow in g* cn r3.
3.25
3.00
2.75
2.50
> 2.25
"ds ~ 0 51 + 2 88Pdi
1.75
1.50
1.25
1.00
0
200
400
600
800
1000
Density (kg-m"3)
F ig u r e 3.6 Permittivity as a function of density in dry snow. Experim ental data provided
by Nyfors (1983), H allikainen (1977), C um m ing (1952) and H allikainen
(adapted from. Hallikainen et al., 1986)
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
et al. (1986).
n
w here Sj is th e dielectric loss o f ice a n d Vi equals [2.12a]
[3.7a]
Pi
a n d pi = 0.916, th e density of ice in g * c n r3.
Since th e p e rm ittiv ity of w a te r is roughly 40 tim es t h a t of d ry snow (T iuri et
al., 1984), even sm all a m o u n ts o f w a te r in liquid p h a se w ill stro n g ly influence
w e t snow dielectrics. T he m o d ified Debye-like m odel ta k e s a d v a n ta g e of th is
fact in e stim atin g th e p e rm ittiv ity , e’Ws [3-8] an d loss, e"Ws [3-9] of w et snow.
Bw*
[3-8]
2
£w s
[3.9]
wv is th e p ercent w a te r v olum e in liquid phase, f is th e freq u en cy a n d f0 is th e
re la x a tio n frequency of w a te r, 9 .0 7 GHz in th e m icrow ave reg io n (H allik ain en
a n d W in eb ren n er, 1992). A, B, C a n d x a re defined b y fittin g th e m odel to
m e a su re d data. F u ll d e ta ils a r e p re se n te d in H allik ain e n et al. (1986).
B ased on Debye m odels, T iu ri e t al. (1984) provide d escrip tio n s of e’Ws (Figure
3.7) a n d e"ws (F igure 3.8) from 0 to 40GHz.
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.2
Wv = 12%
3.0
18
16
i?
> 14
I 12
| 10
1.4
1.2
0
5
10
15
20
25
30
35
40
Frequency (GHz)
Figure 3.7 Modified Debye-like model permittivity of snow as a function of frequency and water in
liquid phase, (adaptedfrom Tiuri et al., 1984)
1.0
0.9
Wv = 12'
0.8
0.7
0.6
0.5
0.4
03
02
0.1
0
0
5
10
15
20
25
Frequency (GHz)
30
35
40
Figure 3.8 Modified Debye-like model loss of snow as a function of frequency and water in liquid phase.
(adaptedfrom Tiuri et al., 1984)
In the lower portion of a dry snow pack brine content dom inates th e m icrow ave
re sp o n se .
B a rb e r (1993) u s e d m ix tu re m odels b a s e d on th e w o rk of
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
D rin k w a te r a n d C rocker (1988) to describe th e d ielectric p ro p e rtie s u n d e r
th e se conditions. T he com plex m ixture models p red ic t a n o v e ra ll dielectric
p ro p erty b a sed on th e d ielectrics of th e constituent p a rts , n a m e ly ice, air an d
brin e. W hen w a te r in liquid p h a se appears, th e m odels p re s e n te d in [3.8] an d
[3.9] a re appropriate.
3.2.1 E x p e r im e n ta l R e s u lts
D ielectric v alu es u se d in th is th e s is are m odelled r e s u lts b a s e d on physical
p ro p erties. T h is is m ain ly d u e to th e difficulty in a c q u irin g e x p e rim e n ta l
dielectric values. H ow ever, i t is im p o rtan t to illu stra te th e fin d in g s of several
a u th o rs who h av e em pirically determ in ed dielectric p ro p e rtie s , a s em pirical
m ea su re m e n ts a re im p o rta n t to develop the models.
Sea Ice Dielectrics
T he v a st m ajority of em pirical stu d ies on sea ice dielectrics h a v e occurred over
first-y e a r ice su rfa c e s or a rtific ia l ice, w ith very little d a ta a v a ila b le for
m u ltiy e a r ice (H a llik a in e n a n d W in eb ren n er, 1992).
V a n t (1976) a n d
H a llik a in e n (1983) e x am in e d p e rm ittiv ity changes a t 1GH z w ith resp ec t to
te m p e ra tu re a n d sa lin ity (F ig u re 3.9). They noted in c re a s e s in p e rm ittiv ity
a n d loss w ith in c re a s e s in s a lin ity an d ra p id in c re a s e s a s te m p e ra tu re s
ap p ro ach ed 0°C.
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1GHz. V int (1976)
5.C - J
0.9G H z . H i l K k a in c n ( 19* 3)
1.0%.
2. 4 J
4.C -
1.4% .
10.5%.
I
«-!*• / /
S.l%^>
0.4 -
3 .0 -
oa
■50
-20
-10
-5
.2
■I
Figure 3.9 Experimental values obtained by various authors for sea ice at 1GHz: (a) permittivity, (b)
loss, (adaptedfrom Hallikainen and Winebrenner. 1992)
From 9 to 16GHz, s e v e ra l re s e a rc h e rs have m e a su re d b o th p e rm ittiv ity an d
loss of r e a l a n d a rtif ic ia lly c re a te d sea ice (F ig u re 3 .1 0 ).
I n a ll cases,
re s e a rc h e rs n o ticed r a p i d in c re a s e s in p e rm ittiv ity a s t h e te m p e r a tu re
approached 0°C a n d slig h t in c re a se s with in creasin g s a lin ity c o n te n t.
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.0
5.5 5.0
|
4.5
I
a.
4.0
■ 1 i----- \
it i
1-------\
i \ i i ■ i----
10GHz. Vant( 1974)
Salinity by Weight
10GHz, Hallikainen (1988)
9.5GHz. Arcone (1985)
10GHz. Hallikainen (1977)
9.5GHz. Hoekstra and Cappil!ano( 1971)
FYI03'
3.5
3.0
2.5
1.0
0.8
F r a z il F Y I
40.
e/:
s/j
0.6
-JO
0.4
A30%.
Columnar FYI /
(ifc .
0.2
0
-50
-20
-10
-5
-2
-1
Temperature (°C)
-0.5
-0.2
-0-1
F igure 3.10 Experimental values obtained by various authors for sea ice between 10 and 16GHz: (a)
permittivity, (b) loss. (adapted from Hallikainen and Winebrenner, 1992)
Snow Dielectrics
P erla (1991) outlined the perm ittivity of snow at 0°C for 1MHz.
He found
average perm ittivity increased from 1.98 w hen liquid w a ter volum e w as 0%
to 7.10 w hen wv w as ju s t u n d e r 1%.
Tiuri et al. (1984) e x a m in e d
perm ittivity of new and old snow a t 5.6 and 12.6GHz. T hey found
is
independent of snow geom etry, as old, new, fine grained an d coarse snow
all yielded sim ilar resu lts. W ith th e inclusion of liquid w a te r they noticed
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increases in perm ittivity an d loss, w ith loss values risin g m ore rapidly.
The modified Debye-like m odels presented in [3.8] an d [3.9] w ere found to
agree well w ith experim ental observations. From 3 to 37GHz H allik ain en et
al. (1986) reported
concurred w ith the resu lts of T iuri et al. (1984).
Perm ittivity increases due to wet snow (Figure 3.11) closely followed
predicted values from th e modified Debye-like models. Dielectric loss in w et
snow also resem bled m odelled values (Figure 3.12)
0.6
ModifiedDebye-likemodel
frequency =37GHz
>» 0.5
0.4
■3 °-3
S 0.2
0.1
0
0
2
4
6
8
10
Water in Liquid Phase (% 100)
12
14
F ig u re 3.11 Changes in perm ittivity as a result of increasing w ater in liquid phase. The
incremental perm ittivity equals e’Ws - E’ds- (adapted, from Hallikainen et al., 1986)
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.7
ModifiedDebye-likemodel
frequency =37GHz
0.6
«■*r 0.5
O
a 0.4
■S 0 3
Q
0.2
0.1
0
4
6
8
Liquid Water Content (% 100)
10
12
F ig u re 3.12 Wet snow dielectric loss as a function of water in liquid phase, (adapted fro m
Hallikainen et al., 1986)
The dielectric theory show n here provides the basis for u n d e rs ta n d in g th e
interactions of a snow covered sea ice volume w ith m icrow ave en erg y
em ission.
The next section will describe some of the historical r e s e a r c h
conducted w ith microwave radiom eters in the m arine cryosphere. It is not
intended to be a n exhaustive account of all microwave re se a rc h in th e
Arctic,
but ra th e r to (a) provide a precedent for using
m icro w av e
radiom etry in th is w ork and (b) outline the previous research from w h ic h
th is thesis builds upon.
3.3 H istorical U se o f M icrow ave Radiom etry w ithin th e M arin e
C ryosphere
Figure 2.2 outlined the large seaso n al variability presen t in ice coverage
over the m arine cryosphere. It also shows why satellite rem ote se n sin g is
a n im p o rtan t research tool in th is region.
The vast spatial scales m e a n
m any in situ observations are necessary to characterize the area. H ow ever,
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
th e hostile clim ate m akes in situ operations difficult, costly a n d d a n g e ro u s.
T hus, the use of rem ote sensing, w ith its large sp atial coverage, is a n
obvious rese a rc h tool for th e m arin e cryosphere.
Since th is region is
enshrouded in d ark n ess over the w in ter and covered in cloud m uch of th e
spring, m icrow ave rem ote sensing is a more effective m e a su re m e n t tool
th a n visible or in frared sensors. The following section is in ten d ed to outline
some of th e histo rical re se a rc h conducted in the m a rin e
cryosp h ere.
Although it is not p ractical to provide a n exhaustive account of m icrow ave
radiom etric re se a rc h in th e m arine cryosphere, the w orks w hich have
bearing on th is thesis are noted.
Satellite based passive m icrow ave m easurem ents began in 1968 w ith the
R ussian Cosmos 243.
T he in stru m e n t supported four n ad ir-lo o k in g
radiom eters o p eratin g from 3.5GHz to 37GHz (B asharinov et a l., 1971).
NASA’s (N ational A ero n au tics Space Agency) first spacebom e im a g in g
radiom eter, laun ch ed in 1972, was the 19.35GHz E lectronically S c a n n in g
Microwave R adiom eter (ESMR). Six years later a m ore ro b u st S c a n n in g
M ultichannel M icrowave R adiom eter system was launched on th e N im bus7. SMMR operated from October 1978 through A ugust 1987 on a lte rn a tin g
days to conserve power.
37.0GHz.
D ata w as collected a t 6.6, 10.69, 18.0, 21.0 a n d
The latest addition to orbital im aging system s w as la u n c h e d
onboard th e U nited S ta te s Defense Meteorological S atellite P rogram (DMSP)
satellite.
T erm ed th e Special Sensor M icrow ave/Im ager
(SSM/I), the
in stru m e n t is a 7-channel, 4-frequency, passive m icrowave system .
D u al­
polarization inform ation is available a t 19.35GHz, 37.0GHz a n d 85.5GHz,
w hile vertical polarization d a ta is available a t 22.235GHz.
The SSM /I
sensor provides daily coverage of th e Arctic w ith a sw a th w id th of
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approxim ately 1400km, excluding a 2.4° radius around the n o rth pole
(NSIDC, 1996). Sensor precision is m easured with respect to a calm ocean
surface and varies from 1.0K to 2.7K for the vertical channels a n d 1.8K to
2.0K for the horizontal cha n n els (Hollinger, 1989). In stru m e n t accuracy is
compared in relation to m odelled T b for calm w ater and varies from -4.3K to
1.1K for V polarization a n d -1.1K to 0.9K for H polarization.
O nboard
radiom etric sensitivity ran g e s from 0.4K for 19GHz an d 37GHz, th ro u g h
0.7K for 22GHz, to IK for 85GHz (Hollinger, 1989).
Some of the earliest science m issions w ith microwave radiom etry w ere r u n
under the Arctic Ice D ynam ic Jo in t E xperim ent (AIDJEX).
W ilheit et al.
(1972) used airborne m icrowave radiom eters in June 1970 to d istin g u ish ice
from open w ater.
However, they noted a diversity in m ea su re d
ice
emissivities which could n o t be explained. Gloersen et al. (1973) followed by
using surface based radiom eters to classify m ultiyear ice flows a n d
refrozen leads th a t were observed on the airborne im agery.
M eeks et al.
(1974) conducted sim ilar experim ents w ith ground based rad io m eters a n d
successfully delineated FY I from MYI in a 110m2 study a re a based on
em issivity differences betw een the ice types.
Cam pbell et al. (1976)
reanalyzed AIDJEX airc raft d a ta to distinguish five “zones” of ice th a t they
felt w ere typical of the B eaufort, Kara, Laptev and E ast S iberian Seas. T he
first zone consisted of sh o refast FYI, the second was ch arac terize d as a
sh ear zone containing FYI an d MYI, while the th ird contains FYI a n d
sm all MYI floes. The fo u rth zone was described as being sim ila r to zone 2,
while the 5th zone contained alm ost exclusively MYI.
The first A ID JE X
experim ents concluded by sta tin g changes in Tb were due to v ariatio n s in
ice salinity as well as the d istribution of a ir bubbles and brine pockets.
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Results from th e se experim ents led to the form ation of ice differentiation
algorithm s
th a t w orked adequately under d ry snow
C avalieri et a l., 1984).
compute
sea
conditions
(e.g.
Using the SMMR in s tru m e n t research ers could
ice concentration,
m ultiyear
ice
fraction
and
the
ice
tem p eratu re. E rro rs in m ultiyear fraction w ere encountered w hen heavy
cloud cover dom inated, or if a strong surface w ind persisted. For exam ple,
C avalieri et al. (1984) noted a 6 m ^ s-1 surface w ind could cause M Y I
fraction m iscalculations by as much as 16%. In all situations the presence
of a w et snow pack negated the operational abilities of th e algorithm s.
Radiom etric observations made in these ex p erim en ts have discovered
em issivities over cold, dry FYI are very consistent (Figure 3.3) Steffen a n d
M aslanik (1988) have docum ented ice salinities re m a in above 6%o over th e
w inter and do not introduce large em issivity variations.
The la rg e st
change in Tb over FYI is associated w ith the overlying snow cover (Eppler
et al., 1992). Conversely, large variation in em issivity h a s been noted over
MYI surfaces, especially with increasing frequency (Figure 3.13).
T his
large fluctuation introduces significant erro r into SWE derivation, as
variations in Tb a re due to the ice in addition to th e snow.
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.00
.95
.90
.85
~> .80
8 .75
^
H max
V max
-70
.65
.60
.55
.50
1
10
100
Frequency
F igu re 3.13 Emissivity variations in MYI as a function of frequency (incidence angle =
50°). (adapted from Eppler et al., 1992)
R esearch on the link betw een snow cover an d microwave ra d io m etry lagged
y ears behind sea ice studies. A lthough correlation analysis betw een a la n d
based seasonally dynam ic snow pack an d T b had occurred in th e 1970s,
(e.g. Kong et al., 1979, C hang et al., 1976), w ork by M atzler et al. (1982) w a s
th e first to formalize th e influence of snow cover on sea ice b rig h tn e s s
tem p eratu res. They discovered a snow pack w ith g rea ter th a n 1% liq u id
w a te r volume affected the em issivity of the combined sea ice-snow volu m e.
They also noted strong a tte n u a tio n in microwave response from 5 to 35GHz,
w hich w as presum ed to be an effect of volum etric scattering. As a re s u lt, it
w as stressed the successful
application of algorithm s
w ould
re q u ire
u n d erstan d in g of th e evolving snow cover.
Begin n in g in the early
u n d erstan d in g the
1980s, significant
interaction
effort began
of passive microwave
to focus
en erg y
w et
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
on
w ith a
snow pack. In 1982, Grenfell and L ohanick (1985) operated a ground based
rad io m eter u n it a t Mould Bay, NWT, to in vestigate the changes in T b a s
th ey related to melt-freeze cycles and m eltpond form ation.
In the early
sp rin g , it w as found FYI Tbs w ere in d ep en d en t of frequency and closely
associated to the physical tem p eratu re o f th e ice surface (i.e. snow cover
h a d negligible effects). MYI signatures d ecreased w ith frequency. D u rin g
th e su m m er, both FYI and MYI sig n a tu re s increased w ith in c re a s in g
frequency, owing to the effects of a w et snow pack. However, if a slush layer
w as p resen t, a significant decrease in T b w as noted. No explanations w ere
given to account for this. By late sum m er, w ith th e form ation of m eltponds,
MYI and FYI signatures w ere equal.
Comiso (1986) noted the use
of d u al
polarization,
single
frequency
observations w ere beneficial in d is c r im in a t i n g inform ation u n d e r w et snow
conditions.
He reported em issivity in c re a se d in both MYI and F Y I
s ig n a tu re s w hen snowpack liquid w ater volum es were above 0%, but less
th a n 5%.
U n d er melt-freeze cycles d e creases in Tb w ere ascribed to
in creased volum e scattering w ithin the snow volume.
G renfell (1986) provided strong evidence on th e scatterin g effect of a th ic k
snow pack w ith the assistance of in s itu m icrowave radiom eters.
He
m e a su re d T bs a t 10, 18, 37 and 90GHz over snow w ith thickness of
(a)<3m m ,
(b) 3 to 50mm and (c) >50m m .
In shallow snow, b rig h tn e ss
te m p e ra tu re s w ere approxim ately equal (F ig u re 3.14), while T bs decreased
w ith frequency in the thickest snow packs, ow ing to volum etric sc a tterin g .
U n d e r laboratory conditions, Grenfell and Comiso (1986) noticed d e crea sin g
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tb a t high frequencies a fter a snowfall event. Comiso et al. (1989) co n tin u ed
th is w ork by system atically rem oving snow from a first-year ice su rfa c e
w hile sim ultaneously o b tain in g microwave m easurem ents a t 90G H z. T hey
found Tb rose ten s of Kelvin w ith the removal of snow.
270
250
£
©
(0
w
230
<D
CL
E
J-©
S 210
©
c
2O i
CD
190
O
H-poi
•
V-pol
170
5
25
45
65
85 5
25
45
65
85 5
25
45
65
85
Frequency (GHz)
F ig u re 3.14 Average brightness tem perature as a function of frequency over snow covered
first-year sea ice for snow thicknesses of (a) <3mm, (b) 3 to 50mm an d (c) >50m m .
(adapted from Grenfell, 1986)
L ohanick and G renfell (1986) continued the work from M ould Bay by
co rrelatin g T b w ith snow a n d ice properties near T uktoyaktuk, NW T. T hey
m ea su re d brightness te m p e ra tu re s a t 10, 19, 34 an d 37GHz, as well a s
snow/ice interface tem p e ra tu re , ice and snow salinities, snow d ep th a n d
snow g rain geometry. R esults indicated a strong linkage e x isted betw een
T b and th e brine volume in th e ice. Snow depth also app eared to c o rre late
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
negatively with microwave em ission.
As snow thickness increased, th e
acquired brightness tem perature decreased.
They also noted co m b in in g
snow depth w ith snow/ice interface tem p era tu re s lead to better c o rrelatio n s
w ith T b .
The m onitoring w ork eventually led to several attem pts a t m od ellin g
changes in Tb based on a n evolving snow cover. Onstott et al. (1987) w e re
the first to provide qualitative models from the spring to su m m er season to
describe the interaction of snow and T b - From 5 to 94GHz, they p o stu lated
decreases in em issivity w ere re su lta n t from enhanced volume sc atterin g in
th e upper snowpack. As the tem p era tu re w arm ed, large scatterin g g r a in s
formed through m etam orphism .
D uring m eltpond form ation they noticed
decreased em issivities, but large inter-day variations.
They concluded
fluctuations in the snowpack caused m ajor difficulties in developing
algorithm s for d e t e r m in in g ice and snow inform ation rem otely.
Lohanick (1990) exam ined the im portance
microwave emission.
of snow
basal layers
on
Over a first-year ice surface, he noted the bottom
50mm of a 250mm dry snowpack supplied a n additional 20K to the acq u ired
brightness tem perature a t 33GHz. As a result, T b did not correlate w ell
sim ply w ith underlying ice conditions or the overlying snow d ep th .
Lohanick (1993) continued w ith an exam ination of m icrowave em ission on
an evolving snow cover over artificial sea ice. He noted a drop in Tb by a s
m uch as 100K a t 10GHz w hen a slush layer developed after snowfall. M ore
im portantly to this thesis, he noted slig h t drops in T b as the snow g ra in s
grew over time.
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Garrity (1993) fu rth ered previous research by looking a t d iu rn a l v a ria tio n s
in a snowpack. It w as found th a t changes in snow liquid w a te r conten t a n d
snow stru c tu re affected the daily T b if cold nights followed w arm days
(G arrity, 1991). G a rrity exploited this inform ation to successfully classify
FYI from MYI even d u rin g the m elt onset.
H er classification is b ased
solely on the snow pack and how te m p era tu re gradients in th e snow la y e rs
over FYI and MYI a re different a t varying p a rts of the day. T he t h in n e r
FYI snow cover is generally affected by increasing a ir te m p e ra tu re s m o re
rapidly th a n MYI snow packs. U sing a 37GHz microwave rad io m eter, s h e
concluded by saying th e m orning is the best tim e for ice d ifferen tiatio n in
the m id-spring, w hile the afternoon is superior d u rin g the la tte r stag es of
sp ring.
Results from th e p a st 30 years have show n the T b of a p a rtic u la r fe a tu re a t
any tim e is a product of scatterin g and em ission from the ice a n d sn o w
cover. We now u n d e rsta n d the im portance of quantifying snow in passiv e
microwave studies. F urtherm ore, we have begun to exploit th e in fo rm a tio n
in the T b caused by variations in the snow. In fact, th e volum e s c a tte rin g
principles of a snow pack provide th e theoretical foundation of th is thesis.
3.4 R esearch Q uestions
The scientific objective provided in section 1.2 gave a gen eral scope for th e
research. Now th a t th e variables of in te re st an d the ratio n ale for stu d y in g
the work have been discussed, more detailed questions can be raised .
A
series of research questions a re listed below, followed by specific ra tio n a le
for analyzing each. T hey provide th e fram ew ork for chapters 3 a n d 4.
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 - How does variation in the snowpack affect in situ derivation o f SW E over
F Y I?
U ncertainties in the sp atial an d tem poral extent of snow cover u n d e r
projected w an n in g scenarios can be m onitored with derived SWE only if
definitive links can be draw n betw een SWE and microwave em ission. As a
re su lt, this work begins w ith a n analysis of the snow cover a n d th e n
exploitation of th a t analysis to obtain a statistical connection betw een SWE
a n d T b - Previous research by Goodison (1990) pointed out w a te r in liquid
p h ase is a key problem w ith c u rre n t SWE algorithm s, so it is a p rim a r y
focus here.
2 - How does variation in the u n d erlyin g ice type affect S S M /I derivation o f
SWE?
V ariatio n s in the u nderlying ice leads to a m ixture of em issivities w ith in
a n rem otely sensed pixel.
It is n o t clear how th a t variatio n w ill affect
SSM /I brightness tem p era tu re s, or even how m uch
v ariatio n ex ists.
T heoretical case studies are constructed to u n d erstan d the effects of s p a tia l
v ariab ility in the underlying ice type on derived SWE. SSM/I a n d sy n th e tic
a p e rtu re rad a r im ages a re classified to docum ent the real w orld v ariab ility
in ice type.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5 C onclusions
This chapter has introduced the m arin e cryosphere as a in teg rated sy stem
w hich is key to u n d e rstan d in g global change.
The sea ice a n d snow
volumes were described in detail, from initial form ation to c o n tin u ed
evolution. This was followed w ith a broader description of how snow is
relevant to the larg er scientific com m unity. As well, this ch ap ter o u tlin e d
the microwave interaction theory necessary to link SWE w ith m icrow av e
radiom etry.
A series of exam ples were given show ing som e of th e
historical work done w ith passive m icrowaves in the m arine c ry o sp h ere.
Finally, detailed research questions w ere presented to fram e th e r e m a in in g
c h ap ters.
Inform ation from the previous two chapters will now be exploited to better
u n derstand SWE derivation over th e m arine cryosphere. Specifically, I w ill
com m ent on the abilities of a m icrowave radiom eter to m easure in s itu
SWE and on the affects w ater in liquid phase has on the SWE derivation.
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 4 • In S itu D e riv e d
SWE and th e Effect o f W ater in
L iqu id P hase
4.1 Introduction
Investigating SWE derivation with an in situ m icrow ave
provides the possibility to concurrently m onitor ch an g es
rad io m eter
in b rig h tn ess
tem p eratu re w ith changes in the snow cover, specifically th e SWE.
The
resu lta n t d a ta set provides a suite of radiances th a t can be rela te d to know n
changes in SWE. In addition, the sm all sp atial reso lu tio n of th e in situ
microwave radiom eter m ean s heterogeneity in the u n d e rly in g ice type w ill
be sm all, fu rth e r e n h an c in g the interpretability of th e m e a su re m e n ts
(Eppler et al., 1992). T his chapter will exam ine in situ SWE derivation w ith
respect to the research question labeled in section 3.4:
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1 - How does variation in the snowpack affect in situ derivation o f SW E over
F Y I?
A series of sub-objectives are helpful in assessing th is question:
1A - How do the geophysical properties o f snow over first-ye a r sea ice
vary over seasonal a n d d iu rn a l time steps?
IB - Can the inform ation fro m 1A, specifically w ater in liq u id p h a se ,
be exploited in SW E derivation over first-year sea ice w ith a n in
situ microwave ra d io m eter?
The m ain objective in th is ch ap ter is to investigate the rela tio n sh ip betw een
SWE a n d microwave em ission th ro u g h a n evaluation of th e in s itu p h y sic al
an d electrical properties w hich give rise to th is connection (1).
I t w ill
specifically focus on the co-occurrence of brightness te m p e ra tu re s a t 50°
(n ea r SSM /I incidence) and th e d iu rn a l and seasonal evolution of th e
scatterin g .
The
intention is to investigate th e sta tistica l re la tio n s h ip s
betw een T b and SWE and to ex p lain th ese em pirical resu lts by a s s e s s in g
th e physical an d electrical ch arac teristics which drive th is re la tio n sh ip .
The in itia l step will be to exam ine how th e snow properties v ary o ver a daily
tim e sp a n (1A). It is im portant to com pletely ch aracterize and u n d e r s ta n d
how th e snow is developing before it can be exploited for SWE d eriv a tio n .
Q uestion IB will focus on sta tistic a l linkages betw een the m icro w av e
b rig h tn ess tem peratures and snow properties. In p a rtic u la r, th e effects of
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
w ater in liquid phase on the SWE retrievals w ith a n in situ m icrow ave
radiom eter w ill be analyzed.
4.2 M ethods
D ata
w ere
collected
as
p a rt
of the
Collaborative
In te rd isc ip lin ary
C rysosphere E xperim ent (C-ICE ‘96) from M ay 11 to J u n e 15, 1996. T he
field site w as located approxim ately 70km n o rth e a st of Resolute Bay,
C ornw allis Island, NWT, C anada (Figure 4.1).
Canadian
North
- <
C o r n w a llis
Isla n d
F ig u re 4.1 C-ICE '96 Field Site.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2.1 Snow and S ea Ice S am p lin g
Snow profiles w ere collected from a n area adjacent to th e ra d io m e te r site
(Figure 4.2). Sam pling w as perform ed w ithout replacem ent in a 0.375 m 2
'crystal pit'.
All m easu rem en ts w ere taken on the diffusely illu m in a te d
snow wall (perpendicular to th e so la r disk) in order to m inim ize th e effect of
solar insolation. D ata w ere g a th e red three tim es daily, a t a p p ro x im ate ly
7:30 am (am), 1:30 pm (noon) an d 6:30 pm (pm) CST.
F ig u re 4.2 Radiometer sampling site during C-ICE'96.
D uring each m easurem ent se t snow depth, density, volum etric liq u id w a ter
content, ice surface an d snow sa lin ity and ice and snow te m p e ra tu re s w ere
recorded. Depth was recorded to th e nearest h alf cm w ith a m e tre stick.
Density sam ples were rem oved a t 2 cm vertical in terv als w ith a 66.36 cc
density cutter and placed into a sealed plastic bag. E ach sam ple w as th e n
weighed to the n e arest 10th of a g ra m on a digital scale and su b seq u en tly
converted to k g*m '3 u sin g th e gravem etric technique. T his is p recise to ±
40 kg*m '3 based on replicate sam p lin g and previously p u b lish ed re s u lts
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(G arrity and B urns,
1988).
The percent w ater in liquid phase w as
estim ated w ith a cap acitan ce
plate (dielectric device).
V alues w ere
recorded from th e p late a t in terv als of 1 m above the snow surface (relative
calibration to air), on th e snow surface and at the m idpoint of each 2 c m
snow sam ple on the p it w all.
The 0 to 6cm layers above th e ice su rface
contain brine w hich n e g ates the capacitance approach to e stim a tin g liquid
w a ter content. In th e se b asal layers liquid w ater con ten t is estim ated by
extrapolation based on th e values observed in the brine free layers of th e
u p p er snow pack. Snow sa lin ity was m easured by m eltin g th e snow density
sam ples to room te m p e ra tu re an d m easuring bulk sa lin ity u sin g a n optical
refractom eter. Ice surface sa lin ity was m easured by scrap in g th e top 2 m m
of the ice surface for m e a su re m e n t using an optical refracto m eter.
Snow
tem p eratu re d a ta w ere recorded continuously w ith 24 AWG,
Cu-Co
therm ocouple ju n ctio n s a n d recorded as 15 m in u t e averages by C am pbell
Scientific In stru m e n ts d a ta loggers (model 21X).
T he ju n ctio n s w ere
embedded in brass tu b in g (9 x 0.5 cm) which in tu rn w ere fastened at 3cm
levels into a wooden dowel.
The sensor array s (including leads) w ere
painted w hite to m inim ize th e rm a l contam ination. T he snow w as packed
evenly during backfilling a n d the sensor leads, w hich extended to the d a ta
logger, were buried to fu rth e r m inim ize th erm al co n tam in atio n .
F u rth e r
details of these m ethods a re presented elsewhere (B arber et al., 1994, 1995).
Physical properties of th e snow pack were subsequently divided into 6 layers
for analysis purposes. L ay ers containing brine a t 2cm, 4cm and 6cm, w ere
u n altered due to the sig n ifican t effect salinity h as on dielectrics a t these
depths. The rem ain in g snow pack was stratified into th re e equal seg m en ts.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Layering
the
snow pack
in
this
m aim er
allows
for
m e a n in g fu l
intercom parisons betw een days w ith different snow depths.
Electrical properties derived in this chapter are based on th e m odels of
B arb er (1993), w hich w ere discussed in chapter three.
4.2.2 M icrow ave R adiom etry
B rightness
te m p e ra tu re s
w ere
m easured
with th ree d u a l
p o larize d
R ussian ATTEX radiom eters a t 19, 37 and 85GHz. All m e a su re m e n ts w e re
ta k e n on an u n d istu rb ed snow surface flanked by the snow pit s a m p lin g
g rid (Figure 4.2). The S urface Based Radiometer (SBR) system w as o p e ra ted
coincident in both tim e a n d space w ith the snow sam pling p rogram .
D u rin g each d a ta acquisition set sam ples were recorded a t 5 d e g ree
in crem en ts from 20° th ro u g h 70°. A custom designed softw are p ro g ra m
autom atically controlled rad io m ete r operations. The SBR sy stem rec o rd ed
a voltage which w as th e n converted to a T b . Samples w ere in te g ra te d over a
50 second look period, providing a n error estim ate of less th a n IK .
C alibration of th e SBR sy stem w as based upon linear regression b etw een a n
Ecosorb® hot load and cosm ic background radiation.
Due to th e need fo r
c le a r skies, only 10 calib ratio n s were performed over the field se a so n .
Consequently, all calibrations (Figure 4.3a) were averaged into one g e n e r a l
calibration equation for e ach frequency. All voltages were th e n re -a n a ly z e d
u sin g this g eneral calib ratio n .
It is felt th a t this is the best option to
m inim ize sensor effects w h ich could obscure the results. R elative stab ility
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Figure 4.3b) appears adequate.
A nonparam etric K ruskal-W allis test
confirms th e 5 sam ple ru n s a re statistically equivalent (p-value = 0.96).
37V
270
450
400
37H
260
350
300
250
R 250
“ 200
150
100
220
6 210
Voltage
200
13:47
Time
► 15:14
F ig u re 4.3a. Absolute and 4.3b. Relative Calibration curves for the SBR system.
4J3 R esults and D iscussion
4.3.1 G eophysical P rop erties
Snow depth began a t 15cm and quickly rose to near 30cm, w here it
fluctuated for th e rest of th e field project (Figure 4.4).
£18
130
140
150
Julian Day
160
170
F ig u re 4.4 Snow depth
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Profile D istributions
Over the field season th e vertical profile of liquid w a te r (Figure 4.5a)
exhibits noticeable in creases towards the ice surface.
T his is a resu lt of
increased tem p eratu res a n d drainage as the sp rin g season advances.
Modelled loss values a t 37GHz (Figure 4.5b) closely correspond to liquid
w ater volumes. This re s u lt is expected due to th e influence w a ter in liquid
phase has on th e complex perm ittivity of snow (Ulaby et a l. , 1986). A th ird
variable of interest, snow pack density, shows some diversity in th e profile
(Figure 4.5c).
D ensities a re highest in the m iddle layers and decrease
tow ards th e snow-ice b o u n d ary as well as the snow -air boundary.
New
snowfall w ith low d en sities accounts for the lower values in th e upper pack,
while the form ation of la rg e kinetic growth grain s ex p lain s the low er
densities in the basal layers. Interestingly, results from previous re s e a rc h
(e.g. B arber et al., 1995) in dicate density in the basal lay ers (1-3 in th is
study) should be significantly less than the m iddle a n d u p p e r snow pack
layers.
However, d u rin g C-ICE '97 an abnorm al a m o u n t of ice m a ss
agglom erates w ere evident in the snowpack, especially in th e basal layers.
F u rth er, ice lenses w ith d en sities approaching 900 kg»m*3 w ere a p p a re n t
in th e lower 3 snow layers. Modelled perm ittivity a t 37GHz e1 (Figure 4.5d)
closely resem bles density m easu rem en ts due to th e influence density h a s
on perm ittivity. Finally, significant variation in salinity is p resen t (Figure
4.5e). Snow wicks brine from the ice surface and salin ity decreases w ith
distance from th e ice.
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Layer 6
Layer 5
Layer 4
Layer 3
Layer 2
Layer I
-c
CE
■ ran —
r-C
■ HI
-I
0
1
2
3
4
5
0
.
.05
I
I— i__ I
.1
.15
Liquid Water Volume (% 100)
Layer 6 \
Layer 5
Layer 4
Layer 3
Layer 2
Layer 1
r
100
1
’
f
1
l'
1
j
---------------------n
—
■—
i
i
i - i
i
'
.
t
i—
i
'
■
i
'
.
H
L
,------------- 1
■—
i
250
300
350
400
450
1-5
Density (kg*m-3)
Layer 3 —l_ L i— ■'
iH
1
Layer 2
Layer 1 I________i___
0
5
1
i
L.
.35
.4
i
.45
j
1---------------- -
d
H
■
'H
- ( - ■
1
.
...........................................................
■ ^
150 200
1
1
i .1 _ i —
|--------------------------------------------- ,
i
,---------------1
■
I
.3
c
i-i
i—
- i
1
•—
.—
.
'
i—
-
I
.25
Modelled 37GHz Loss
,
’
i
.2
1-------------------------------------------- 1
T
.
i
1.75
1------------------------------------------.
.
i
2
.
i
2.25
2.5
Modelled 37GHz Permittivity
1
e
1 1----- t - J - I -1
10
Salinity (ppt)
15
20
Figure 4.5 Profile distribution of (a) liquid water volume, (b) modelled 37GHz loss, (c)
density, (d) modelled 37GHz perm ittivity and (e) salinity.
Seasonal Evolutions
Seasonally, the w a ter c o n ten t rem ain s relatively constant d u rin g th e firs t
h a lf of th e field season a n d th e n shows an increased variability in th e la tte r
h a lf (Figure 4.6a). As th e season progressed towards spring, th e in cid en ce
of w arm storm fronts increased, explaining th e increased variance of liq u id
w ater. The general in c re ase in the late season is a resu lt of elevated a ir
tem peratures corresponding to th e spring tem perature evolution. M odelled
37GHz loss values (Figure 4.6b) do not increase as su b stan tially a s liq u id
w ater because of the corresponding loss in snow salinity as w e tn ess
increases.
Both density (F igure 4.6c) an d derived 37GHz p erm ittiv ity
(Figure 4.6d) show a slig h t increase, especially in the up p er layers, a s th e
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
spring progresses. N a tu ra l compaction, wind effects a n d m elt-freeze cycles
will all increase snow pack density, especially in the u p p e rm o st layers.
Salinity reduces slightly as sp rin g advances (Figure 4.6e).
W ith w a rm e r
tem p eratu res brine d ra in a g e begins and correspondingly low er salinity
concentrations a re observed.
0.5
0.4
0.3
Layer 6
Layer 6
0.2
0.1
0.0
Layer 5
Layer
0.4
0.3
0.2
o
o
0.0
Layer 4
5
Layer 4
- 04
o 0.3
n 02
g 0.1
Layer 3
£ 0.0
Layer 3
! 0.4
-C 0.3
|
Layer 2
0.2
0.1
0.0
0.4
0.3
0.2
0.1
0.0
Layer 1
Layer 1
0.40.3-
0 .2 0. 1 -
130
140
150
Julian Day
160
170
0.CF
130
140
150
Julian Day
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
170
2.4
500
400
300
200
Layer 6
Layer 6
100
400
Layer 5
300
200
Layer 5
100
400
Layer 4
>
S' 300
|
200
t
100
Layer 4
a.
n
s 400
a
300
200
Layer 3
100
400
300
200
100
Layer 2
■
400 ■
300 •
200-
Layer 1
100-
130
140
150
Julian Day
160
170
130
140
150
Julian Day
160
170
Layer 3
Layer 2
c.
c.
>%
c
Layer 1
130
140
150
Julian Day
160
170
Figure 4.6 Seasonal evolution of (a) liquid water volume, (b) modelled 37GHz loss, (c)
density, (d) modelled 37GHz perm ittivity and (e) salinity.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
D iurnal Variations
Dium aUy, a noticeable difference betw een the 3 sample sets for liquid w a te r
co ntent (Figure 4.7a) and m odelled 37GHz loss (Figure 4.7b) becomes c le a r
tow ards the end of the field project.
A m w ater contents rem a in fairly
stable, while noon and especially p m sets showed a m arked rise in liq u id
w a te r content. Little difference in e ith e r density (Figure 4.7c) or m odelled
37GHz perm ittivity (Figure 4.7d) is noted, as expected. Density w ill c h a n g e
m ainly through m etam orphic
action, which is more noticeable over
seasonal scales ra th e r th a n d iu rn a l scales - during the course of a few
h o urs, the density is unlikely to su b stan tially change.
Finally, the s m a ll
observed decrease in salinity (Figure 4.7e) implies th a t gravity d rain a g e of
w a te r in liquid phase is sm all over th e sam pling period.
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.5
0.4
0.3
0.2
Layer 6
Layer 6.
0.1
0.0
Liquid Water Volume (%I00)
Layer 5
Layer 5
0.4
0.3
0.2
Layer 4
O
r~
Layer 3;
0.0
0.4
0.3
0.2
0.1
in
0.0
jjj
5
0.4
0.3
I
Layer 4
Layer 3
0.2
0.1
0.0
0.4
0.3
Layer 2
Layer 2.
0.2
0.1
0.0
Layer 1
Layer 1.
0.4
0.3
0.2
0.1
0 .0 L
130
140
150
Julian Day
160
170
130
140
150
Julian Day
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
160
170
500
400
300
2.4
200
100
1.5
Layer 6
400
300
Layer 5
E
u
Layer 4
100
g 400
Q 300
200
100
Layer 3
400
300
CL
N
X
or ^
2.1
o
1.8
u
■g
2
Layer 2
2.1
200
Layer 2
100
400
300
130
1.5
>>
>
400
<7 300
I 200
200
100
Layer 5
2.1
200
100
“
Layer 6.
2.1
Layer 1
2.1
Layer 1
140
150
160
170
Julian Day
130
140
150
Julian Day
160
170
20
Layer 3
Layer 2
— AM
Noon
“ PM
e
a
C/5
Layer 1
130
140
150
Julian Day
160
170
F igu re 4.7 D iurnal separation of the seasonal evolution in (a) liquid water volume, (b)
modelled 37GHz loss, (c) density, (d) modelled 37GHz permittivity and (e) salinity.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4 3 .2 M icrow ave R ad iom etzy
Geophysical effects on SW E monitoring
C om parison
betw een
snow
density
and
m icrow ave
tem peratures (Table 4.1) indicates a consistent
brightness
negative correlation.
Increasing densities ind uce more volume scattering, thereby reducing th e
Tb -
T ab le 4.1 Correlation M atrix between snowpack layers and brightness temperatures.
Layer
19H
19V
37H
37V
85H
85V
6
-0.23
-0.23
-0.28
-0.32
-0.30
-0.17
5
-0.08
-0.10
-0.13
-0.09
-0.08
-0.05
4
-0.23
-0.19
-0.17
-0.13
-0.08
-0.07
3
-0.25
-0.18
-0.24
-0.21
-0.20
-0.40
2
-0.25
-0.22
-0.31
-0.30
-0.44
-0.46
1
-0.13
-0.17
-0.31
-0.36
-0.26
-0.29
D isregarding n a tu ra l v a ria b ility in snow properties w ill adversely affect th e
SWE predictive capabilities of microwave radiom etry.
V ariability w h ich
can be explained by com binations of frequency and polarization is low w ith
combined d a ta from a m , noon and p m sets
(Figure 4.8).
A t horizontal
polarization, 85GHz re s u lts are the weakest; 37GHz sire strongest.
F or
vertical polarization, 19GHz is the poorest estim ator, w hile 85GHz is best.
19GHz is the least sensitive frequency to changes in th e snow pack and th is
explains the lower R2 values.
Inversely, 85GHz is know n to be very
responsive to slig h t v a ria tio n s in liquid w ater content a n d this likely
w eakens the relationship betw een SWE and Tb a t H polarization.
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Y = 22.996 - .056fTBl9V)
Y = 23.103- .061 (TB19H)
210 220 230 240 250 260 270 280 290
TB 19H
240
250
260
270 280
TB I9V
290
300
310
Y = 19.428 - .043rrB37V)
Y = 18.334 - ,042(TB37H)
14
12
12
10
a 10 -
-
8
6
4
180
200
220
240 260
TB37H
280
300
320
200
Y = 13.13 - 024(T b85H)
12
240
260
280
TB 37V
300
320
Y = 12.906 - 022(T b85V)
12
-
-
a: 10 -
w 10 -
120
220
140
160
180 200
TB85H
220
240
260
120
160
200
240
TB 85V
280
320
Figure 4iJ Initial regression plots which include all 3 diurnal sam pling periods lum ped
over the full duration of the field experiment.
Im proving
microwave
u n d erstan d in g
sources
radiom etric
of erro r
SWE m easu rem en ts
such
as liquid w ater,
n e c e ssita te s
d en sity a n d
dielectrics. Figure 4.7 show ed th a t w ater content and m odelled 37GHz loss
were higher in the noon a n d p m sets. To further illu stra te , th e seaso n ally
averaged liquid w ater volum e (Figure 4.9) clearly displays th e p m set h a s
th e most liquid w ater, followed by noon and then am .
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Layer 6
Layer 5
Layer 4
Layer 3
Layer 2
Layer 1
.5
1.5
2.5
Liquid Water Content
3.5
F ig u re 4.9 W ater Volume by Subset.
Subsetting the d a ta into a m , noon an d p m sets shows th a t am d a ta e x p la in s
a m uch h igher percen tag e of the observed variability th an eith er noon or
pm
results (Table 4.2).
N orm al probability plots and D urbin-W atson
statistics validate p a ra m e tric analysis in all but 85V.
Note the low est
value (pm) coincides w ith th e h ig h est liquid w a ter content and dielectric
loss values. D ielectrically, th e p m snow pack is em ittin g noticeable en erg y
w hich
elevates the b rig h tn e ss
tem p eratu re;
low ering the
predictive
capabilities. The noon set, w ith a slightly lower w a te r content, exhibits a
sim ilar effect on a sm a lle r scale.
Once the effects of liquid w a te r a n d
dielectric loss a re analyzed, SWE prediction from microwave ra d io m etry
im proves dram atically.
T able 4.2 Subsets of 50° Tg vs. SWE
Frequency
R2
am
noon
pm
19H
0.70
0.27
0.04
19V
0.73
0.26
0.01
37H
0.91
0.34
0.03
37V
0.89
0.27
0.01
85H
0.85
0.38
0.01
85V
0.71
0.32
0.05
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SW E A lgorithm D evelopm ent
O perational SWE prediction includes various combinations of p o lariza tio n
an d frequency.
A lthough th e in itial and subset analyses u se only sin g le
frequencies, th e m ajority of SWE retrieval algorithm s use a com bination of
19 and 37GHz. Two of th e m ore popular algorithm s, a fter C h a n g et al.
(1987) [4.2] and Goodison et al. (1990) [4.3] are analyzed w ith a m d a ta in th is
study.
SWE = K 1(T 19H - T 37 h )
[4.2]
s w e = K l(T 3 7 y -T i9 v ) + K i
[4 3]
w here Kx is a n orbital correction based on geographic location a n d T x is th e
brightness tem p eratu re a t a p a rtic u la r frequency and polarization.
R esults (Table 4.3) based on [4.2] a n d [4.3] are poorer SWE e stim ato rs t h a n
single 37H or 37V for the conditions observed in this case study. C o m b in in g
th e V and H polarizations from [4.2] and [4.3] show no im provem ent over
37H (refinem ent 1), or a m u ch w orse re su lt (refinem ent 2). A dding 85GHz
to 37GHz provides is also less precise. Sim ilarly, polarization in fo rm a tio n
is of little value. It seem s even th e sm allest am ount of liquid w a te r n e g ates
th e potential of using polarization inform ation or 85GHz to p red ict SWE.
Sim ple combinations of frequency a n d polarization do not im prove th e SWE
predictive ability of m icrowave radiom etry.
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T a b le 4.3 Single an d M ulti-Frequency/Polarization R 2 .
Variables
R2
C om m ents
19H-37H
0.72
Based on Chang et al., 1987
37V-19V
0.89
Based on Goodison et al., 1990
19V-37H
0.90
Refinement 1
19H-37V
0.68
Refinement 2
85V-37V
0.12
85GHz
85H-37H
0.08
85GHz
19V-19H
0.10
Polarization Effect
37V-37H
0.18
Polarization Effect
85V-85H
0.04
Polarization Effect
E xam ination of th e incidence angles show 20° through 60° are fairly
uniform as SWE estim ators (Figure
4.10). 60° is the best estim ato r for 3
frequencies, while 20°, 30° a n d 40° each estim ate one frequency th e best.
However, the differences in R2 values a re not considered statistica lly
distinguishable. Overall, 60° is m arginally th e best, while 70° is clearly th e
worst. A lthough varying th e incidence angle appears to be of lim ited v a lu e
in SWE prediction, the d a ta is too lim ited to provide a definitive answ er.
1.0
0.9
I9GHz
0.8
37GHz
<N
04 0.7
0.6
0.5
20°
30°
40°
50°
60°
70°
Angle
F ig u re 4.10 Incidence angle effects on SWE estimation.
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
M ultiple regression techniques ap p ear to improve the SWE d e riv a tio n
abilities of the microwave radiom eter. U sing all 6 channels, 96.8 p e rc e n t of
th e observed v ariatio n can be explained (Table 4.4).
S e p a ra tin g into H
polarizations (92.9%) and V polarizations (91.5%) across all freq u en cies
yields equally good estim ato rs, while 37GHz (90.7%) is slightly b etter th a n
19GHz (89.1%) an d 85GHz (85.7%).
C aution m ust be noted because th e
im provem ent is, a t lea st in p a rt, due to the statistical artifac t of in c re a s in g
th e num ber of independent variables. This given, continued re s e a rc h in to
m ultiv ariate approaches to SWE estim ation is still recom m ended.
T a b le 4.4 Multiple Linear Regression Results.
Variables
R2
Com m ents
All Freq/Pol
0.97
M ultiple Linear Regression
V Pols
0.92
M ultiple Linear Regression
H Pols
0.93
M ultiple Linear Regression
19GHz
0.89
M ultiple Linear Regression
37GHz
0.91
M ultiple Linear Regression
85GHz
0.86
M ultiple Linear Regression
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.4 C onclusions
In the introduction to th is ch ap ter, the following questions w ere noted:
1 - How does variation in the snowpack affect in situ derivation o f SW E over
F Y I?
LA - How do the geophysical properties o f snow over first-ye a r sea ice
vary over seasonal a n d d iu rn a l time steps?
IB - Can the inform ation from LA, specifically w ater in liq u id p h a se ,
be exploited in SW E derivation over first-year sea ice w ith an in
situ microwave radiom eter?
In this ch ap ter it w as found snow pack density and perm ittiv ity w ere larg est
in the m iddle of th e snow pack, while w ater in liquid phase a n d d ielectric
loss were g rea test in the b asal layers. Salinity was g reatest a t th e snow -ice
interface and rapidly d ecreased to Oppt above 6cm.
Over th e se a so n a l
evolution snow pack density, w a ter in liquid phase, p erm ittiv ity a n d loss
slightly increased, w hile sa lin ity showed a slight drop. O n a d iu rn a l tim e
step snowpack density, p erm ittivity and salinity rem ained fairly c o n sta n t,
w hile w ater in liquid phase a n d loss showed larg er v ariatio n .
Typically,
w a ter contents were h ig h est in the evening and lowest in th e m o rn in g .
Exploiting the d iu rn al sep ara tio n of th e seasonal evolution w as show n to be
a n effective approach in rem otely m onitoring SWE.
By u sin g sa m p le s
w hen liquid w ater content w as lowest, statistically significant re la tio n s h ip s
w ere determ ined betw een SW E and the microwave brightness te m p e ra tu re .
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This ch ap ter has d em o n stra te d th a t snow w ater equivalence retriev a ls c a n
be m ade w ith a m icrow ave radiom eter, so long as the liquid w a te r co n ten t
is accounted for. However, even under dry snow conditions, heterogeneity
in the ice type could negatively affect SWE derivation.
T his problem is
compounded w hen th e larg e spatial scales of the m arin e cryosphere a re
considered.
For in stan c e, considerable diversity in ice type h a s been
observed over the c e n tra l A rctic (Cavalieri et al., 1991; Com iso,
Carsey, 1982).
snow
1990,
Since m onitoring the spatial and tem poral v a riatio n s in
cover over th e
m arin e
cryosphere will req u ire
a m icrow ave
radiom eter w ith a la rg e r sp a tia l coverage, the next c h a p te r ex am in es th e
effect of ice diversity on SSM /I SWE derivation.
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 5 - Effects o f S e a Ice
H eterogeneity on SSM/I SW E
D erivation
5.1 Introduction
A ccurate SWE retrievals over large spatial regions
are
n e ce ssa ry
to
characterize the spatial a n d tem poral variation of snow in th e m a rin e
cryosphere. The m ain disad v an tag e with using in situ m e a su re m e n ts is
th e finite spatial scale.
W hile they m ay be representative w ith respect to
local conditions, the observational network is lim ited in th e m a rin e
cryosphere. Conversely, th e advantage of satellite m icrowave rad io m etry is
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the ability to m onitor th e e n tire m arin e cryosphere on a daily tim e step
irreg ard less of d a rk n e ss or cloudiness (Eppler et al., 1992). F u rth e rm o r e ,
since SSM/I observations have been recorded since 1987, a n o p e ra tio n a l
SSM/I SWE algorithm
could yield a long term record of snow cover
variation over the m a rin e cryosphere.
This would provide in sig h t into
w hether the projected clim atic w arm ing outlined in c h a p te r two h a s
already affected th e snow cover. The disadvantage in going from in situ to
satellite based m e a su re m e n ts is increased diversity in the field of view .
W ith spatial resolutions on th e order of 25km2, a SSM/I pixel c a n co n tain a
m ixture of ice types a n d snow conditions.
This c h ap ter w ill e x a m in e
heterogeneity in the u n d e rly in g ice type with respect to th e re s e a r c h
question noted in section 3.4:
2 - How does variation in the underlying ice type effect S S M /I derivation o f
SWE?
In order to assess th is a series of sub-objectives are provided:
2A - Are the SW E derivations from an in situ m icrow ave ra d io m e te r
a n d a satellite based microwave radiometer associated?
2B - To w hat extent does variation in the underlying ice type (F Y I vs.
M YI) statistically affect brightness tem peratures
on the F-13
S S M / 1 sensor?
2C - W hat does real w orld underlying ice type variability look like?
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The m ain objective is to ad d ress the operational use of SSM/I for sea ice
SWE estim ation w ith resp ect to problems of heterogeneity in th e u n d e rly in g
ice type (2). Since th e preceding chapter reported fairly good success w ith
an in situ radiom eter, th e first step will be to investigate the link betw een in
situ SWE m e a su re m e n ts and SSM/I derived SWE from 1993 a n d 1994 (2A).
Then a series of sen sitiv ity tria ls will be performed to better u n d e rs ta n d
w hat effect sp a tia l heterogeneity has on emissivity (2B). Specifically, th e
effects of (a) ice type heterogeneity and (b) spatial p a ttern on em ission w ill
be studied. Finally, th is c h ap ter will examine real world v ariab ility w ith
the both SSM/I and E a rth Resources Satellite (ERS-1) images (2C).
5J2 M ethods
The m ethods w ill be outlined in three sections which
correspond
to
Objectives 2A, 2B a n d 2C.
5.2.1 In S itu v s. SSM /I SWE E stim ation
Data were collected over two different field cam paigns, 1993 a n d 1994, fro m
the Seasonal Sea Ice M onitoring and Modelling Site (SIMMS).
sites (Figure 5.1) a re located around Resolute Bay, NT, C anada.
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Both field
D evon
bland
Lancaster
Sound
• SIMMS93
• SIMMS94
b lin d
F ig u re 5.1 SIMMS Field Sites.
In S itu Data P hysical Data
D ata collection m ethods w ere identical to those presen ted in section 4.2.1,
w ith the following exceptions (M isurak, 1994, 1993):
• Profile samples w ere collected every 3cm in a lOOcc d en sity cutter.
• Sam ples were ta k e n a t 1:30pm only.
S S M /I
SSM/I im agery used in th is project were computed from gridded sw ath
d ata, as opposed to th e comm only available daily SSM /I EASE-Grid
N o rthern H em isphere b rig h tn ess tem peratures.
U sing a single pass
m inim izes averaging of daily T b , which could have deleterious effects on
SWE estim ation.
G rids re p re se n t spatially interpolated d a ta and w ere
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
processed to facilitate the operational and scientific use of th e SSM/I sensor.
The interpolation technique maximizes radiom etric integrity of th e o rig in a l
brightness tem p eratu res, while m ain tain in g high sp atial an d te m p o ra l
precision (NSIDC, 1996). Resolution in the grid series is 12.5km2 for 85GHz
d ata and 25km 2 for all other channels.
Derived SWE w as com puted w ith
[5.1], based on the te rre stria l algorithm supplied by W alker and G oodison
(1993). Five separate dates over the field seasons were used (Table 5.1). I n
1993, Ju lia n days 138 and 162 appeared to have su b stan tial am o u n ts of
liquid w ater. Based on the effect of w ater in liquid phase p resented in th e
previous chapter, th ese days should not correlate well w ith in situ SWE.
sw e=
-20.7-49.27(37V-19V)
lo
^
w here 37V is 37 GHz V polarization and 19V is 19 GHz V polarization.
T ab le 5.1 Derived SWE data acquisition dates.
Y ear
Julian Day In-situ Julian Day SSM/I
W ater Volume (%100)
1993
126
125
0.00
1993
138
139
2.22
1993
162
163
8.42
1994
127
126
0.89
1994
130
128
0.30
5.2J2 Spatial H eterogeneity as a Source o f V ariation
For the sensitivity tria ls used here, frequency, polarization an d in cid en c e
angle w ere held constant. Nine separate case studies were created b ased
on the configuration and composition of ice types (Figure 5.2; Table 5.2).
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Each case is composed of a 25x25 square grid, which re p re se n ts th e s u rfa c e
im aged in one SSM /I pixel.
F ig u re 5.2 Case study diagram of underlying ice types. Dark grey shades rep resent FY I,
while light grey shades represent MYI.
T ab le 5.2 Case study table of underlying ice types.
Case
%FYI
%MYI
Spatial Distribution of FYI
75P
75
25
Perim eter of FOV
75C
75
25
Centre of FOV
75R
75
25
Randomly D istributed
50P
50
50
Perim eter of FOV
50C
50
50
Centre of FOV
50R
50
50
Randomly D istributed
25P
25
75
Perim eter of FOV
25C
25
75
Centre of FOV
25R
25
75
Randomly Distributed
FOV = field of view.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Em issivities (Table 5.3) collected from E ppler et al. (1992) w ere u se d in a
random num ber g en erato r to calculate the 25s grids w ith re sp e c t to
polarization and frequency.
A verage emissivity w as first c a lc u la te d
w ithout the SSM/I an te n n a p a tte rn s to exam ine the effect of ice type. T h e n
th e average em issivity w as re-calculated w ith the F-13 SSM /I a n te n n a
p a ttern s (Figure 5.3) to ex am ine th e effect of spatial differences in ice. F o r
each frequency and polarization, th e a n te n n a patterns w ere converted into
a weighted function w ith su m equal to 625 (25x25 grid). E m m isivity v a lu e s
in the centre of th e field of view received more weight.
T a b le 5.3 Microwave emissivities of sea ice used in case studies.
Ice Type
F irst year
M ultiyear
Polarization 19GHz
H
0.888
(0.019)
37GHz
0.913
(0.020)
85GHz
0.886
(0.031)
V
0.941
(0.019)
0.955
(0.015)
0.926
(0.045)
H
0.780
(0.080)
0.706
(0.115)
0.650
(0.011)
V
0.850
(0.068)
0.764
(0.079)
0.680
(0.105)
( ) = standard deviation of pooled emissivity values, (adapted from Eppler et al., 1992)
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
dB
■M---------------------------------------------- 25km --------------------------------------------------►
F ig u re 5.3 F-13 SSM/I antenna patterns, (courtesy Gene Poe, personal communication)
5& 3 SAR C haracterization o f S p atial H eterogeneity
Spatial heterogeneity in real world situations was exam ined w ith the
NASA T eam Algorithm (Cavalieri et al., 1984; Gloerson a n d C a v alie ri,
1986) for SSM/I im agery and B ayesian classification w ith ERS-1 im a g e ry .
The NASA algorithm [5.2] presum es FYI, MYI and open w a te r (OW)
dom inate the cryospheric icescape.
I t uses a spectral g rad ie n t ra tio [5.1]
and a polarization ratio [5.3] to determ ine the combination of MYI, FY I a n d
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
OW in a p articu lar cell. All SSM/I pixels w ithin 96 to 99°N a n d 74 to 75°W
w ere included for th is study.
GR(19V,37V) =
T B(37 V) - T B(19 V)
T B(37V )+T b (19V)
[5.2]
PR(19GHz) = TB(1 9 V )-T B(19H)
T b (19V) + T b (19H)
[5.3]
Since th e sp atial resolution of a low-res ERS-1 im age is 100m2, it follows th e
synthetic a p e rtu re ra d a r (SAR) im age will be able to ex am in e ice type
v ariatio n a t a m uch sm a ller scale in com parison w ith th e NASA T e a m
A lgorithm .
S egm entation of a SAR scene from 1993, 1994 a n d 1995 w a s
perform ed w ith a supervised classification schem e. To be co n sisten t w ith
th e NASA Team Algorithm , ice types were classified as e ith e r FYI or M Y I.
Rubble zones w ere grouped w ith MYI.
M ultiple tra in in g sites w e re
selected for each ice type (land areas were visually m asked o u t beforehand)
an d a Bayesian m axim um likelihood processor w as used to calculate th e
e n tire scene.
5.3 R esults and D iscussion
The results will be presen ted in th ree parts, identical to th e m eth o d s
section.
5.3.1 In S itu vs. SSM/I SW E E stim a tio n
B ased on the success of c h ap ter th ree in-situ SWE values from SIM M S 1993
and 1994 were com pared (Table 5.4). A relationship appears to be p re s e n t
(Figure 5.4), but the lim ited d a ta cannot be statistically analyzed.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T h is
sm all d a ta volum e is a resu lt of lim ited snow m easu rem en ts in 1993 a n d
1994 an d a n absence of SSM/I d a ta in 1995 and 1996. Except for M ay 14,
1994, the SSM /I algorithm u n d e re stim a ted SWE. This m ay be a re s u lt of
w a ter in liquid phase.
W ith th e appearance of w ater, the m icrow ave
rad io m eter in creasin g ly begins to m onitor a re tu rn only from the w a te r in
the top portion of th e snowpack. R adiances increase sharply an d begin to
m erge a t 19 a n d 37GHz (Eppler et al., 1992). As a resu lt, 37V-19V in [5.1]
approaches zero an d SWE is u n d erestim ated .
The W alker a n d G oodison
(1993) algorithm can sep arate w et snow areas, while Mote a n d A n d e rso n
(1995) delineated w et snow a re as in the Arctic.
Therefore, it should be
possible to determ in e if th e snow pack on th e m arine cryosphere is wet.
T ab le 5.4 Brightness tem peratures from th e test dates.
Y ear
Ju lia n Day
37GHz V
19GHz V
1993
126
242.900
251.200
21.570
30.298
1993
138
260.800
264.500
8.990
30.766
1993
162
262.000
266.200
10.350
21.162
1994
127
240.500
253.900
35.530
43.584
1994
130
241.200
254.600
35.960
35.497
Derived SWE In-situ SWE
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
□
1993
A
1994
0
0
5
10
15
20
25
30
SSM/I SWE (cm)
35
40
45
F ig u re 5.4 Bivariate plot of in situ versus SSM/I SWE. Julian days included in the graph.
5.3.2 Ice H eterogeneity a s a Sou rce o f V ariation
W ith the lack of em pirical m easu rem en ts, theoretical sensitivity tria ls a re
employed here to assess the effect of heterogeneity in the u n d erly in g ice.
The spatial p a tte rn of FYI and MYI from Figure 5.2 is still a p p a re n t a fte r
applying the random ly g enerated em issivities for FYI and MYI
(F ig u re
5.5). All 6 frequency/polarization a re sim ilar in nature, w ith only 19GHz H
pol being shown. The sm aller sta n d a rd deviation of FYI leads to a m o re
uniform em issivity, as opposed to th e highly speckled MYT areas.
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
F ig u re 5.5 Em issivity v a ria tio n s o f MYI overlaid on original ice concentration p a tte rn s.
Segm enting into the nine case situations, em issivity is show n to v a ry w ith
frequency, polarization a n d percent composition of FYI (F igure
Em issivity decreases w ith increasing frequency and is la rg e r
5.6).
for V
polarization. As expected, em issivity is highest for 25% FYI an d low est for
75% FYI. W ith the addition of SSM/I antenna p attern s, the sp a tia l location
of the ice becomes im portant. Relative to the initial ru n , cases w ith FY I in
the centre of the field of view display increases in em issivity, w hile cases
with FYI on the periphery display decreases.
W hen th e ice is ran d o m ly
distributed em issivity does n o t change substantially.
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t
*
i
*
i
.
I
75P
75C
50R
50P
50C
»
.
|
25P
25C
.
i
O
.
i
•
i
.
i
* - -1
Without SSMI
Antenna Patterns
•
m
m
Om
25R
With SSMI
Antenna Patterns
•
qq
qq
am
9D
9
9
93
09 19V
I9H
am
75R
•
mo
om
om
mo
75P
C 75C
o
3 50R
J 1 50P
e
U
o
■
m
90
cm
om
9D
m>
75R
50C
• O
09
O9
9
«D
•
- 25R
•
9)
25P
•o
90
25C
om
37H
75P
75C
50R
m
50P
m
9 o
09
O 9
9 O
9
o
•
50C
9D
90
0 9
25R
25P
25C
■j ---1--
.65
.70
.75
37V
0 9
am
mo
om
o m
75R
O
9D
9
1--.80
85H
1---1---1--.85
.90
.95
«---1---
0 9
.70
85V
1---1---1---«---1---»---r
.75
.80
.85
.90
.95
Emissivity
F ig u r e 5.6 Emmisivity variation as a function of frequency, ice configuration and SSM /I
an ten n a patterns.
Since th e algorithm used for SWE estim atio n in th is study [5.1] uses 19 a n d
37GHz V brightness te m p e ra tu re s, em issivity values w ere converted to
b rig h tn ess tem peratures w ith prescribed ice tem p eratu res (F igure 5.7).
For a n ice tem perature of 270K, T bs ran g e from less th a n 220K to m ore th a n
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
250K.
O ther ice te m p e ra tu re s yield sim ilar ranges
in the a c q u ire d
brightness tem p eratu re.
260
250
•
240
□
230
-
□
220
□
□
O
A
A
♦
O
O
♦
♦
25P
25R
210
200
190
180
25C
A
O
♦
A
□
A
O
♦
O
♦
50C
50P
50R
Ice Configuration
O
♦
75C
•
□
'
•
T=270K
□
A
-
□
T=260K
A
O
-
A
T=250K
°
♦
♦
1
75P
75R
O
T=240K
+
T=230K
F ig u re 5.7 37V T b based on ice type and tem perature. 19V is similar.
Computation of SWE w ith [5.1] w as used to exam ine effects of the ran g e in
brightness tem peratures. Since th e sensitivity trials do not place any sn o w
on the ice, a value n ear zero w ould be considered accurate.
Even th o u g h
[5.1] has not been developed specifically for the Arctic, it appears it m ay be
useful in deriving SWE over predom inately FYI surfaces (Table 5.5), a s
SWE values a re low, especially w ith colder ice. This table also show s th e
im portance of ice location w ith in a pixel. W ith FYI on the periphery, s u c h
as case 75P, resu lts degrade sub stan tially . In fact, a scene w ith 50% F Y I
located in the centre includes less erro r th a n one w ith 75% located on th e
periphery. W hen MYI begins to dom inate the scene, SWE e stim ates w ith
th e cu rren t p aram eters of [5.1] could overestim ate snow w ater equivalence
by over 45cm. This is a re s u lt of th e difference in em issivity betw een M Y I
and FYI. W hereas average em issivity for FYI is 0.941 a t 19GHz a n d 0.955
a t 37GHz, it is 0.850 a t 19GHz a n d 0.764 a t 37GHz (Table 5.3). The in c re a s e d
difference in MYI will equate to a larger difference in T bs . This tra n s la te s
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to a n increased gap in 37v-19v in [5.1], which leads to th e ov erestim atio n .
In these cases, the sp a tia l location of FYI is not im p o rtan t, a s su b s ta n tia l
error is noticeable even w h en FYI is centered w ithin the FOV (e.g. 25C).
Since the variance in MYI em issivities is high, it would not be possible to
simply apply a n offset to th e SWE algorithm based on MYI percentage. F o r
the precise use of SWE estim ates, these results show it is im p o rta n t to
m onitor scenes th a t are principally FYI.
T able 5.5 SWE derived with [5.1] and prescribed ice temperatures.
Ice Config.
T=270K
T=260K
T=250K
T=240K
T=230K
25P
45.41
43.69
41.96
40.24
38.51
25C
39.50
37.99
36.49
34.98
33.48
25R
47.63
45.82
44.01
42.21
40.40
50P
42.45
40.84
39.22
37.61
35.99
50C
15.11
14.51
13.90
13.30
12.70
50R
29.89
28.74
27.59
26.44
25.29
75P
15.85
15.22
14.59
13.96
13.33
75C
0.33
0.27
0.22
0.16
0.11
75R
2.55
2.41
2.27
2.13
2.00
5.3.3 SAR C haracterization o f Spatial H eterogeneity
Based on the above resu lts, it becomes clear successful SWE d eriv atio n
m ust take into account the underlying ice type (as well as sp a tia l p a tte rn ).
U sing the NASA Team A lgorithm [5.2, 5.3], both 1993 and 1994 a p p e a r to
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have substantial am ounts of MYI w ithin each pixel (Figure 5.8). W hen th e
region of in terest is expanded it is a p p aren t 1994 contains m ore M Y I
(Figure 5.9).
I_1—1_L_l__I_■ ■ ■ ■ I
-1_______
o.ioo
0.075
OW
0.050
FY
0.025
o 0.000
o
1994
□
1993
-0.025
-0.050
-0.075
MY
-0.100
"I— r ~ '—
0
— I— 1—1— 1— '-----1— r - i —r f - r -
.05
.10
20
.15
.25
PR
F ig u re 5.8 Ice type during 1993 and 1994 as determined by the NASA Team A lg o rit h m .
0.000
*
*
*
*
*
*
*
-0.005
-
0.010
-0.015
oj
o
-0.020
O
1994
-0.025
□
1993
-0.030
-0.035
-0.040
-0.045
0.02
0.03
0.04
PR
0.05
0.06
Figure 5.9 Enlarged portion of Figure 5.8 outlining ice type during 1993 and 1994 a s
determined by the NASA Team Algorithm.
W ith the NASA Team A lgorithm , only the general c h arac ter of a n SS M /I
pixel can be determ ined.
However, the use of SAR can classify th e pix el
w ith greater resolution. In 1993 (Figure 5.10), the im age appears to h av e a
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
m ixture of MYI and FYI.
Significant am ounts of MYI a re noticeable
so utheast of L ow ther Isla n d and e a st of Young Island. As a re s u lt, secto rs
6, 8 and 9 would not yield accurate SWE estim ates. Sectors 2-5 a n d 7 a p p e a r
to be predom inately FYI (Table 5.6), so variations in em issivity w ould m o re
likely be a resu lt of snow pack properties.
m ass from B ath u rst Islan d .
Sector 1 co n tain s a larg e la n d
G an (1996) has pointed out th e difficulty in
obtaining SWE over th is type of surface.
1994 (Figure 5.11) is noticeably MYI deficient in com parison w ith 1993.
T his is not in accordance w ith th e SSM/I results, but m ay be due to th e size
of th e ERS-1 scene - the SAR im age does not cover the e n tire region of th e
SSM/I image. From a SWE derivation perspective, sectors 4,5,7 a n d 9 a re
composed m ainly of FYI (Table 5.7). The ice types in sectors 6 a n d 8 a re
m ostly FYI, but Low ther a n d G riffith Islands could obscure th e ability of
SWE estim ation. B a th u rs t a n d Cornwallis Islands dom inate sectors 1-3.
1995 (Figure 5.12) dem onstrates MYI dom inates this region in som e y e a rs.
This would overestim ate SWE in sectors 1-4 and 6-7. Sector 5 could yield
acceptable resu lts since FY I dom inates the centre of th e pixel.
Sectors 8
an d 9 do contain m ore FY I (Table 5.8), w ith Low ther Isla n d posing a
possible problem in sector 8.
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1993
[Bathurst Islam
Garrett Islam
Lowther Islani
Young Islain
■ Land
■ FYI
: MYI
□ Unclassified
F ig u re 5.10 EKS-1 classified image from May 13, 1993. Boxed areas indicate one SSM/I
pixel (25km2). Numbers are used to distinguish the boxed areas in the text.
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T a b le 5.6 Ice type composition of ERS-1 classified image from May 13, 1993.
Sector
%FYI
%MYI
%Land
% U nclassified
1
32.1
3.2
63.6
0.1
2
85.9
0.1
14.0
0.0
3
98.7
2.3
0.0
0.0
4
97.2
2.6
0.0
0.2
5
89.4
8.2
2.4
0.0
6
55.5
23.6
20.0
0.9
7
94.1
4.7
0.0
1.2
8
67.1
27.4
2.9
2.5
9
25.8
70.0
1.5
2.7
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1994
(Bathurst Islam
Cornwallis Islam
Griffith Islam
Lowther Islam
■
■
3
□
Land
FYI
MYI
Unclassified
F igu re 5.11 ERS-1 classified image from May 7, 1994. Boxed areas indicate one SSM/I
pixel (25km2). Numbers are used to distinguish the boxed areas in the text.
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T a b le 5.7 Ice type composition of ERS-1 classified im age from M ay 7, 1994.
Sector
%FYI
%MYI
% L and
% U nclassified
1
57.7
0.5
41.2
0.6
2
37.7
0.5
60.9
0.9
3
2.1
0.0
97.8
0.1
4
94.8
1.3
3.8
0.1
5
96.2
2.0
1.0
0.8
6
82.1
2.0
16.7
0.2
7
97.3
0.6
2.1
0.0
8
81.4
1.0
17.3
1.2
9
96.6
3.4
0.0
0.0
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1995
€>
/
N
arrett rg'ETTTTi
O
O
irnrniTrrrinrrrr.
i-s;
■ Land
■ FYI
a MYI
□ Unclassified
F igu re 5.12 ERS-1 classified image from April 19, 1995. Boxed areas indicate one SSM/I
pixel (25km2). N um bers are used to distinguish the boxed areas in th e text.
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
T a b le 5.8 Ice type composition of ERS-1 classified im age from April 19, 1995.
Sector
%FYI
%MYI
% Land
% U nclassified
1
0.0
100.0
0.0
0.0
2
0.5
98.9
0.2
0.4
3
1.2
97.4
1.4
0.0
4
9.7
88.3
1.1
0.9
5
56.4
39.3
2.3
2.0
6
21.7
61.2
16.1
1.0
7
41.6
45.1
7.5
5.8
8
69.7
27.6
0.8
1.9
9
88.2
9.5
0.0
2.3
5.4 Conclusions
Several objectives were presented in the beginning of th is chapter:
2 - How does variation in the underlying ice type effect S S M /I derivation o f
SWE?
2A - Are the SW E derivations from a n in situ microwave radiom eter
and a satellite based microwave radiom eter associated?
2B - To w hat extent does valuation in the underlying ice type (F Y I vs.
MYI) affect em issivity on the F-13 S S M / I sensor?
2C - What does real world underlying ice type variability look like?
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although the 5 points listed h e re a re positively related, it is not a p p ro p ria te
to do a statistical te st to d e t e r m i n e w h eth er th is relationship h as s ta tis tic a l
significance.
However, it a p p e a rs the presence of w a te r in liquid p h a s e
forces an underestim ation of SWE w ith a cu rren t te rre stria l based SWE
alg o rith m .
The
sensitivity
tria ls
em ployed
here
showed
em issivity
d e c re a se d
significantly w ith (a) decreasin g FY I ice type and (b) w ith a n in c re a se in
MYI in th e centre of th e pixel. T he increase in derived SWE w as a re s u lt of
an increase in th e em issivity difference for MYI. As a resu lt, derived SWE
could be in error by over 45cm.
In all years a m ixture of FYI a n d MYI was noted. The ERS-1 c la ssific a tio n
stated 1994 contained th e m ost FYI and would be th e best c an d id a te to
explore SSM/I derived SWE. Conversely, 1995 was dom inated by MYI, w ith
1993 being m ainly FYI, but also containing some MYI in c e rta in re g io n s.
The SSM/I classification noted la rg e r MYI contents in 1994, w ith less in
1993. However, since th e ERS-1 im age has greater sp a tia l reso lu tio n , it
should be more precise in ice classification.
These results show
MYI
significantly
overestim ates
SWE w ith
th e
algorithm used on the C a n a d ia n prairies. Some m ethod of a n aly zin g th e
ice concentration w ithin a n SSM /I pixel is needed to adequately m o n ito r
SWE.
Although th e NASA T eam A lgorithm can describe th e g e n e ra l
m ake-up of a pixel, this w ork h a s also shown the spatial p a tte rn of ice is
im portant. To assess th a t, SAR im agery should be used.
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sum m ary
CHAPTER 6
C onclusions
and
6.1 Sum m ary
T he introductory c h ap ter signaled the m arine cryosphere as a reg io n
playing a principal role in global change and in our u n d e rs ta n d in g of s u c h
change. It suggested a v aria tio n in th e sp atial and tem poral ex ten t of snow
m ay be one repercussion of w a rm in g in the cryosphere. F u rth e rm o re , th e
in itia l chapter stated m e a su rin g SWE is an effective w ay to m onitor an y
v ariations in th is snow cover.
C h a p te r two delineated th e physical aspects of the m arin e cryosphere.
It
described the form ation and evolution of a seasonally dynam ic snow
covered sea ice volume. T his c h a p te r also portrayed the snow cover as a n
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in te g ral component of m ass a n d energy fluxes, as well a s a controlling
force in sea ice grow th and decay. It was also show n how chan g es in the
snow cover could affect local a n d global climate, as well a s ecology in the
m arin e cryosphere.
C h a p te r three illu strated how th e physical properties of snow and ice
in te ra c t w ith microwave energy.
It suggested SWE is a n effective way to
m onitor any changes in the snow pack and portrayed sea ice as the m a in
source of microwave em ission, w ith snow cover being the m ajo r source of
m icrow ave attenuation. This c h ap ter also described th eo retical models to
predict the dielectric properties of the snow
and
ice
and
provided
experim ental findings from th e recent past.
C h a p te r four incorporated th eo ry from chapters two an d th re e to assess the
ability of an in-situ microwave radiom eter in SWE estim ation. T he prim ary
steps included m onitoring th e vertical profiles, seasonal evolutions an d
d iu rn a l ranges in th e physical a n d electrical properties of th e snow and ice.
S ta tistica l links betw een Tg a n d SWE were not ap p aren t w h en th e full su ite
of observations were used. However, investigation of the am , noon and p m
sets noted a significant difference in liquid w ater content a n d dielectric loss
- th e m orning set w as significantly drier th an either th e noon or pm. U pon
segm enting the d ata into th e th re e tim e periods, re-an aly sis showed a
sta tistica l link betw een th e am Tb and SWE. Noon an d pm se ts showed no
correlation. The fourth c h a p te r concluded by sta tin g SWE c a n be derived
w hen liquid w ater contents a re accounted for.
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C hapter five scaled the re su lts from chapter four up to satellite based
microwave radiom etry, since the m arine cryosphere is c h aracterized by a
large geographic area.
T he sensitivity tria ls noted varying e ith er th e ice
type or the spatial p a tte rn of ice resulted in em issivity differences.
T h ese
differences w ere shown to cause potential error in SWE derivation.
In
p articular, th e addition of MYI forced an u nderestim ation of SWE w ith th e
c u rre n t SWE algorithm used in the C anadian prairies.
Segm entation of
SSM/I im agery w ith th e NASA Team Algorithm noted 1993 a n d
1994
contained m ixtures of FYI a n d MYI, while Bayesian classification o f ERS-1
im agery also noted the sp a tia l arran g em en t of the ice in th ese y e ars.
C hapter five ended by noting variation in th e underlying ice is a sig n ific a n t
source of variation th a t SWE algorithm s m u st account for.
6.2 C onclusions
In the in itial chapter, th e following scientific objective w as proposed:
Science Objective: “To provide insight on the developm ent o f
SW E algorithm s for snow covered sea
ice using m icrow ave
radiom etry. “
Based on th e resu lts presented in this thesis, the following conclusions c a n
be draw n:
1) The d iurnal separation of w a te r in liquid phase can be exploited in SWE
derivation over first-year se a ice. In p articular, it w as shown th a t th e
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
scattering properties of a d ry snowpack can be rela te d to th e snow w a te r
equivalence.
H ow ever,
th e presence
of liquid
w a te r
forces
an
u nderestim ation of SWE w ith cu rren t algorithm s u sin g th e ratio of 19
and 37GHz. It w as show n th a t m ultiple regression tech n iq u es m ay be a
better avenue to pursue.
2) The presence of m u ltiy e a r ice confuses SWE e stim atio n d u e to th e la rg e
em issivity v ariations in MYI.
Even w ith a dry snow pack, th e p resen ce
of MYI causes a n overestim ation of SWE due to its lower em issivity. The
spatial location of MYI w ith in a pixel was also noted to be im p o rtan t. I f
MYI is located tow ards th e periphery of the sen so r's field of view, th e
overestim ation is reduced.
3) Although this w ork h a s pointed out th a t SWE derivation is possible over
first-year ice surfaces, a fully operational SWE a lg o rith m rem a in s to be
developed. Such a n alg o rith m will have to first identify th e percentage of
w ater in liquid phase w ith in th e snowpack. If it is found to be negligible,
th en the algorithm
m u s t exam ine the spatial heterogeneity of th e
underlying ice type.
B ased on the results of th is th esis, only in th e
situation w here FYI d o m in ates the scene could a SWE re trie v a l th e n be
m ade.
Several factors lim it th e re s u lts presented in this w ork. F o r c h a p te r th re e ,
th e m ajor provisions re g a rd th e sm all sam ple sizes p re se n t in both th e in
situ and SSM/I analyses. D espite the seem ingly good re la tio n sh ip p re se n t,
C-ICE 96 was the p rem iere of the SBR system in A rctic SWE analyses.
It
covers only a two m onth w indow in the spring of 1996 a n d all sam ples w ere
777
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
collected a t a fixed site.
It is therefore not possible to com m ent on the
applicability of th ese re s u lts
over different ice types.
I t cann o t be
guaranteed these resu lts w ould be m anifest in oth er a reas of FY I, or even
in other years.
For exam ple,
H allikainen
(1989) noted
sig n ific a n t
differences in m icrowave em ission over Finland from 1979 to 1982 even
though the SWE w as th e sam e. H all et al. (1991) observed a sim ila r effect
over Alaska. These discrepancies have not been reported by o th e r a u th o rs ,
nor is it clear why th e v a ria tio n s exist.
F u rth e r resea rch is n e ce ssa ry to
understand th is potential problem .
In chapter 4, th e m ajor lim ita tio n s a re the lack of real d a ta to su p p o rt th e
findings. E arly SIMMS y e a rs m onitored snow only in 3 day in te rv a ls an d a t
few sites. In th e la tte r y e a rs (after 1994), where snow in fo rm a tio n w as
collected for m any m onths on daily scales, SSM/I d a ta is sp a rse , or not
available. F u rth e r, even th o u g h SIMMS/C-ICE snow re s e a rc h e rs sa m p le
from m any sites, all are w ith in a sh o rt distance of one a n o th e r w h en th e
scale of SSM/I is considered.
T ruly adequate in situ sa m p lin g w ould
require collection over th e e n tire A rctic archipelago, so m eth in g w hich is
not feasible.
6.3 Future D irections
It can be appreciated th a t th e process of deriving SWE over th e c ry o sp h ere
is a n inherently difficult ta s k . However, several advances in th e n ex t few
years should alleviate some of th e se problem s. For instance, th e A dv an ced
Microwave Scan n in g R adiom eter (AMSR), to be launched in 2000, h a s tw ice
the spatial resolution of SSM /I a t 19 a n d 37GHz (12.5km), w hile 85GHz is
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
even finer, w ith a resolution of 5km. The sm aller field of view sho u ld re su lt
in more homogenous pixels, enhancing the possibility of SWE retriev a ls.
The AMSR is a twelve channel,
radiom eter system. It m easu res
six frequency
passive
m icrow ave
vertically and h o rizontally polarized
brightness tem p eratu res a t 6.925, 10.65, 18.7, 23.8, 36.5 a n d 89.0GHz. The
AMSR is a p a rt of NASA’s PM-1 series of E arth O bserving Science (EOS)
satellites, developed in conjunction w ith the M ission to P la n e t E a rth
(MTPE).
Future research should focus more on the effects of w a te r in liquid p h ase.
For instance, if we can system atically characterize the ex act con trib u tio n of
wet snow to em ission, th e n it stan d s to reason we could account for wet
snow in the SWE algorithm . The use of lower frequencies m ay be usefu l in
such an endeavour.
Questions rem ain su rro u n d in g the problem of MYI.
The th eo re tic al basis
for SWE is th a t changes in SWE cause the majority of a tte n u a tio n in th e T b FYI over 10mm has a fairly consistent emissivity value, b u t M YI is h ig h ly
variable. This variability cu rren tly precludes the accu rate re trie v a l of SWE
over MYI surfaces. As well, the distribution of snow over MYI is m u c h
more variable th a n snow over FYI. In rough topography, snow d ep th s c a n
range from zero to well over a m etre w ithin a few m etres. C om bined, these
two points emphasize th e extrem e difficulty involved in try in g to derive SWE
over MYI surfaces.
Continued avenues of resea rch should also attem pt to d raw sam p les from a
large spatial region of in situ SWE. This should be linked to a d etailed stu d y
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of the underlying ice. C om bining th ese two with satellite based m icrow ave
radiom etry could yield a successful
SWE algorithm
for the m a rin e
cryosphere.
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cited References
A agaard, K., and E.C. C arm ack . 1994. The Arctic Ocean a n d clim ate: A
perspective. In The P o lar O ceans and Their Role in S h a p in g the G lobal
E nv iro n m en t AGU G eophysical M onograph 85. eds. O. M .
Jo h an n essen ,
R.
D.
M uench,
and
J.
E.
O v erlan d .
A m erican Geophysical U nion, W ashington, D.C. pp. 5-20.
A agaard, K. and E.C. C arm ack. 1989. The role of sea ice a n d other fr e s h
w a ter in the Arctic circulation. Journal o f Geophysical R e se a rc h .
94:14485-14498.
Akitaya, E. (1975) IAHS-ASIH Publication. 114:42-48.
Arcone, S.A., A.G. Gow, a n d S. McGrew. 1986. S tru c tu re a n d d ielectric
properties a t 4.8 and 9.5GHz of saline ice. Jo u rn a l o f G eophysical
Research (C12). 91:14 281-14 303.
Barber, D.G. 1993. A ssessm en t of the Interaction of Solar R adiation (0.3 to
3.0 pm) w ith a Seasonally D ynam ic Snow Covered Sea Ice V o lu m e,
from M icrowave (2.0 to 5.0 cm) Scattering. EOL T echnical R eport
Series ISTS-EOL-TR93-002. U niversity of Waterloo. W aterloo, Ont.
Barber, D.G., T.N. P ap ak y riak o u , E.F. LeDrew, an d M.E. Shokr. 1995. A n
exam ination of the re la tio n betw een the spring period evolution of th e
scatterin g coefficent (o°) a n d radiative fluxes over lan d fast sea ice.
International Journal o f Rem ote Sensing. 16:3343-3363.
Barber, D.G., T.N. P ap ak y riak o u , and E.F. LeDrew. 1994. O n th e
relationship betw een energy fluxes, dielectric properties, a n d
m icrowave scatterin g on snow covered first-year sea ice d u rin g th e
spring tran sitio n a l period. Journal o f G eophysical R esea rc h .
99(C11):22 401-22 411.
Barry, R.G., J.M . Fallot, a n d R.L. A rm strong. 1995. T w en tieth -ce n tu ry
variability in snow-cover a n d approaches to detecting a n d m o n ito rin g
changes: sta tu s and prospects. Progress in P hysical G eography.
19:520-532.
B asharinov, A.E., A.S. G urvich, S.T. Egorov, V.I. Zhukov, A.A. K u rsk a y a ,
L.I. Malafeev, D.T. M ateev, A.S. Mikhailov, an d A.M. Shutko. 1971.
R esults of observations o f th e th erm al radio em ission of e a rth ’s surface
in a n experim ent on th e Cosmos-243 satellite. K osm onaut Iss le d .
9:268-273.
Broecker, W., and T. H-Peng. 1989. The cause of the glacial to in te rg la c ia l
atm ospheric CO2 change: A polar alkalinity hypothesis. G lobal
Biogeochemistry Cycles. 3:15-239.
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Brown, R.D. and P. Cote. 1992. In te ra n n u a l V ariability of L an d fa st Ice
Thickness in th e C anadian H igh Arctic, 1950-89. A rctic. 45:273-284.
B ryan, F. 1986. H igh-latitude salin ity effects an d
therm ohaline circulations. N a tu re. 323:301-304.
in te rh e m is p h e ric
B urton, W.K., N. C abrera, and F.C. F ran k . 1951. The grow th of c ry s ta ls
an d the equilibrium stru c tu re of th eir surfaces.
P hiloso p h ica l
Transactions o f the Royal Society o f London. 243:299-358.
Cam pbell, W .J., P. G loersen, W .J. W ebster, T.T. W ilheit, a n d R.O.
Ram seier. 1976. B eaufort sea ice zones as delineated by m icro w av e
imagery. Journal o f Geophysical Research. 81:1103-1110.
C arm ack, E.C., R.W. M acdonald, R.G. Perkin, F.A. M cL aughlin, a n d R .J.
Pearson. 1995. Evidence for w a r m in g of A tlantic w a te r in the s o u th e r n
C anadian B asin of th e A rctic Ocean: R esults from th e L arsen-93
expedition. Geophysical Research Letters. 22:1061-1064.
Carsey, F.D. 1992. M icrowave Remote Sensing of Sea Ice. AGU G eophysical
Monograph 68. W ashington, DC. 462 pp.
C arsey, F.D. 1982. Review a n d s ta tu s of rem ote sensing of sea ice.
Journal o f Oceanic Engineering. 14:127-138.
IE E E
C arsey, F.D., R.G. B arry, D.A. Rothrock, and W.F. Weeks. 1992. S ta tu s a n d
fu tu re directions for sea ice rem ote sensing. In M icrowave R em ote
Sensing of Sea Ice. AGU G eophysical M onograph 68. ed. F. C arsey. pp.
443-446.
C avalieri, D.J., P. Gloersen, and W .J. Campbell. 1984. D eterm in atio n o f sea
ice param eters w ith th e N im bus 7 SMMR. Jou rn a l o f G eophysical
Research. 89(D4):5355-5369.
C avalieri, D., J. Crawford, M.R. D rinkw ater, D.T. E ppler, L.D. F a r m e r ,
R.R. Jentz, and C.C. W ackerm an. 199L A ircraft active an d p a ssiv e
microwave validation of sea ice concentration from the D efense
Meteorological Satellite P rogram special sensor m icrow ave Im a g e r .
Journal o f Geophysical Research. 96(C12):21,989-22,008.
Cess et al., 1991. In terp retatio n of snow -clim ate feedback a s produced by 17
general circulation m odels. Science. 888-891.
C hang, A.T.C., J.L. Foster, a n d A. Rango. 1991. utilization of surface cover
composition to improve the m icrow ave d eterm ination of snow w a te r
equivalent in a m o u n tain basin. International Jo u rn a l o f R e m o te
Sensing. 12:2311-2319.
C hang, A.T.C., J.L Foster, an d D.K. H all. 1987. N im bus-7 derived global
snow cover param eters. A n n a ls o f Glaciology. 9:39-44.
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
C hang, A.T.C., P. G loersen, P. Schmugge, T.T. W ilheit, an d H .J. Zwally.
1976. M icrowave em ission from snow and glacier ice. Journ a l o f
Glaciology. 16:23-29.
C hapm an, W.L., a n d J.E . W alsh. 1993. Recent variations of se a ice and a ir
tem p era tu re
in high latitudes. B ulletin o f the A m e r ic a n
Meteorological Society. 74:33-47.
Colbeck, S.C. 1987. Snow M etam orphism and C lassification. In Seasonal
Snow covers: Phvsics. Chem istry. Hydrology, eds. H.G. Jones an d
W .J. O rville-T hom as.
Colbeck, S.C. 1986. C lassification of seasonal snow cover cry stals. W ater
Resources Research. 22:59S-70S.
Colbeck, S.C. 1983. T heory of m etam orphism of dry snow. Jou rn a l o f
Geophysical R esearch.. 88:5475-5482.
Comiso, J.C . 1990. A rctic m ultiyear ice classification and su m m e r ice cover
u sing passive m icrow ave satellite data. Journal o f G eophysical
Research. 95(C8):13 411-13 422.
Comiso, J.C. 1986. C h aracteristics of Arctic w inter sea ice from satellite
m u ltisp ectral m icrow ave observations. Journal o f G eophysical
Research. 91(C1):975-991.
Comiso, J.C ., G renfell, T.C., Bell, D.L., Lange, M.A., a n d S.F. Ackley.
1989. Passive m icrow ave in-situ observations of w in ter W eddell sea ice.
Journal o f Geophysical Research. 95(C8):10891-10905.
Cox, G.F.N., an d W .F. W eeks. 1988. N um erical sim ulations of th e profile
properties of u ndeform ed first-year sea ice d u rin g th e g ro w th season.
Journal o f Geophysical Research. 93:12,449-12,460.
Cox, G.F.N., and W .F. Weeks. 1975. Brine drainage an d in itial salt
e n tra p m e n t in sodium chloride ice. CRREL R esearch rep o rt 345.
Hanover, NH, USA. 85 pp.
C um m ing, W. 1952. T he dielectric properties of ice and snow a t 3.2cm.
Journal o f A p p lied Physics. 23:768-792.
de Q uervain, M.R. 1963. O n the metam orphism of snow. In Toe and Snow:
properties, processes, and applications - proceedings of a conference
held a t MIT, Feb 12-16,1962. The MIT Press, C am bridge, MA, pp. 377390.
Dickson, R.R., J . M eincke, S.A. Malmberg, and A .J. Lee. 1988. T he "G reat
S alinity A nom aly" in the northern N orth Atlantic 1968-1982. P rogress
in Oceanography. 20:103-151.
Doronin, Y.P., a n d D.E. Kheisin. 1975. Sea Ice.
P ublishers, St. P etersb u rg . 323pp.
G irdrom eteoizdat
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
D rinkw ater, M.R. a n d G.B. Crocker. 1988. M odelling C hanges in th e
Dielectric and S c a tte rin g Properties of Young Snow-Covered Sea Ice a t
GHz Frequencies. J o u rn a l o f Glaciology. 34:274-282.
Eicken, H., M.A. L ange, a n d G.S. Dieckmann. 1991. S p atial v ariab ility of
sea-ice properties in th e N orthw estern W eddell Sea. J o u rn a l o f
Geophysical Research. 96:10 603-10 615.
Eide, L.I., and S. M artin. 1975. The formation of brine d rain a g e fe a tu re s in
young sea ice. J o u rn a l o f Glaciology. 14:137-154.
Eppler, D.T., L.D. F a rm e r, A. L. Lohanick, M.R. A nderson, D .J. C a v alie ri,
J. Comiso, P. G loersen, C. G arrity, T.C. Grenfell, M. H a llik a in e n ,
J.A. M aslanik, C. M atzler, R.A. Melloh, I. R ubinstein, a n d C.T. Sw ift.
1992. Passive m icrow ave sig n atu res of sea ice. In M icrowave R em ote
Sensing of Sea Ice. A G U Geophysical Monograph 68. ed. F. C arsey. pp.
41-71.
Eppler, D.T., L.D. F a rm e r, A.W. Lohanick, and M. Hoover. 1986.
Classification of se a ice types with single-band (33.6GHz) a irb o rn e
passive m icrowave im agery.
Journal o f Geophysical R e se a rc h .
91(C9):10 661-10 695.
Fukusako, S. 1990. T herm ophysical properties of ice, snow, and se a ice.
International Jo u rn a l o f Thermophysics. 11:353-372.
G arrity, C. 1993. P assive m icrowave remote sensing of snow covered
floating ice d u rin g sp rin g conditions in the Arctic a n d A n tarctic. AES
MWG-OR 93-1. 348pp.
G arrity, C. 1991. P assive m icrow ave remote sensing of snow covered
floating ice d u rin g sp rin g conditions in the A rctic an d A n ta rc tic .
Ph.D. d issertation, Y ork U niversity, CRESS d ep artm en t, N orth Y o rk ,
O ntario.
G arrity, C. and B. B u m s. 1988. E lectrical and Physical P roperties of Snow
in Support of BEPERS-88, Technical Report MWG 88-11. Y o rk
University, N orth York, O ntario. 65 pp.
Gloersen, P., and W .J. C am pbell. 1991. Recent variations in A rctic a n d
A ntarctic sea-ice covers. N a tu re. 352:33-36.
Gloerson, P., and D.J. C avalieri. 1986. Reduction of w e a th e r effects in th e
calculation of sea ice concentration from m icrow ave ra d ia n c e s .
Journal o f Geophysical Research. 91(C3):3913-3919.
Gloersen, P., W. N ordberg, T .J. Schmugge, and T.T. W ilheit. 1973.
Microwave sig n a tu re s of first-year and m ultiyear ice. J o u rn a l o f
Geophysical Research. 78:3564-3572.
Goodison, B.E. 1997. T he cryosphere: An indicator of clim ate ch an g e ?
Geomatics in th e E ra of R ad ars a t Conference. O ttaw a, ON. M ay 26-30,
1997.13 pp.
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Goodison, B.E. 1990. Rem ote S e n sin g of Snowcover. In The C a n a d ia n
Remote Sensing C ontrib u tio n to U nderstan d in g Global C h a n g e , ed s.
E.F. LeDrew, M. Stone, a n d F. Hegyi. pp. 93-110.
Goodison, B.E., A.E. W alker, a n d F.W. T hirkettle. 1990. D e term in atio n of
snow w ater equivalent on th e C an ad ian prairies u sin g n e a r re a l-tim e
passive microwave d a ta . Proceedings of th e W orkshop on A p p licatio n s
of Remote Sensing in H ydrology, Saskatoon, C anada, eds. G.W. Kite
and A. Wankiewicz. pp. 297-309.
G renfell, T.G. 1992. S urface-based passive microwave stu d ies of m u ltiy e a r
sea ice. Journal o f Geophysical Research. 97(C3):5063-5074.
G renfell, T.C. 1986. S urface-based passive microwave observations of se a
ice in the B ering an d G re en lan d seas. IEEE T ransaction s on
Geoscience a n d Remote S e n sin g . GE-24:378-382.
G renfell, T.G. 1979. The effects o f ice thickness on the exchange of so la r
radiation over th e P olar O ceans. Journal o f Glaciology. 22:305-320.
Grenfell, T.G., D. Bell, A. L ohanick, C. Swift, and K St. G erm ain . 1988.
M ultifrequency passive m icrow ave observations of salin e ice grow n in
a tank. Proceedings of th e IGARSS'88 Symposium, pp. 1687-1690.
G renfell, T.C and J.C . Comiso. 1986. M ultifrequency passive m icro w av e
observations of first-year se a ice grown in a tank. IE E E T ra n sa c tio n s
on Geoscience a n d Rem ote S en sin g . GE-24:826-831.
Grenfell, T.G., and A.W. L ohanick. 1985. V ariations of the m icro w av e
signatures of sea ice d u rin g th e late sp rin g and early su m m e r n e a r
Mould Bay, NWT. J o u rn a l o f Geophysical Research. 90(C3):5063-5073.
G renfell, T.G, and D .K Perovich. 1984. S pectral albedo of sea ice a n d
incident solar irrad ian c e in th e S outhern Beaufort Sea. J o u rn a l o f
Geophysical Research. 89(C3):3573-3580.
H all, D .K , M. Sturm , C.S. B enson, A.T.C. C hang, J.L. Foster, H. G arb eil,
and E. Chacho. 1991. P a ssiv e m icrowave rem ote an d in situ
m easurem ents of A rctic a n d su b arctic snow covers in A laska. R em o te
Sensing o f E nvironm ent. 38:161-172.
H all, D .K , J.L . Foster, a n d A .T.C. Chang. 1982. M e a su re m e n t a n d
m odelling of m icrow ave em ission from forested snow fields in
Michigan. Nordic H ydrology. 13:129-138.
H allikainen, M.T. 1989. M icrow ave radiom etry of snow. A d va n ced Space
Research. 9:267-275.
H allikainen, M.T. 1983. A new low -salinity sea ice m odel for U H F
radiom etry. International J o u rn a l o f Rem ote Sensing. 4:655-661.
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
H allikainen, M.T. 1977. Dielectric properties of N aCl a t 16GHz. Report S107, H elsinki U niversity of Technology, Radio Laboratory. 37 pp.,
Espoo, F inland.
Hallikainen, M., a n d D.P. W inebrenner. 1992. The physical basis for se a
ice rem ote sensing. In Microwave Remote S ensing of Sea Ice. A G U
Geophysical M onograph 68. ed. F. Carsey. pp. 29-44.
Hallikainen, M.T. an d P.A. Jolm a. 1986. Retrieval of th e W ater E quivalent
of Snow Cover in F inland by Satellite M icrowave Radiom etry. IE E E
Transactions on Geoscience a n d Remote Sensing. 24:855-862.
Haykin, S., E. Lewis, K. Raney an d J.R. Rossiter. 1994. Remote S en sin g o f
Sea Ice a n d Icebergs. John Wiley & Sons, Toronto.
H enderson-Sellers, A.,
and P.J. Robinson.
1987. C ontem porary
Clim atology. Longm an Scientific & Technical. New York. 439 pp.
H oekstra, P. and P . Cappillino 1971. Dielectric properties of sea and so d iu m
chloride ice a t U H F and microwave frequencies. Journal o f
Geophysical Research. 76:4922-4931.
Hollinger, J.P ., B.E. troy, J r., R.O. R am seier, K.W. Asm us, M .F.
H artm an , a n d C.A. L uther. 1984. M icrowave em ission from h ig h
Arctic sea ice d u rin g freeze-up. Journal o f Geophysical R esea rch .
89(C5):8104-8122.
IPCC. 1995. C lim ate change 1995. Im pacts, a d ap tatio n s and m itigation of
clim ate change. C ontribution of W orking Group II to the Second
A ssessm ent R eport of the Intergovernm ental P anel on C lim ate
Change. C am bridge U niversity Press. 877 pp.
IPCC. 1990. C lim ate change: The IPCC scientific assessm ent, eds. J .T .
Houghton, G .J. Je n k in s, and J.J . E p h rau m s. Cam bridge U niversity
Press. 365 pp.
Jew ett, S.C., and H.M . Feder. 1980. A utum n food of a d u lt starry flo u n d er
Platichthys ste lla tu s from th e NE Bering Sea an d th e SE Chukchi Sea.
Journal o f International Exploration. 39:7-14.
Johannessen, O.M., M. Miles, and E. Bjorgo. 1995. The A rctic’s s h rin k in g
sea ice. N ature. 376:126-127.
Kong, J.A ., R. S hin, J.C . Shiue, and L. T sang. 1979. Theory a n d
experim ent for passive microwave rem ote sen sin g of snow packs.
Journal o f Geophysical Research. 84(B10):5669-5673.
Kotlyakov, V.M., a n d M.G. Grosswald. 1990. In te ra c tio n of Sea Ice, Snow
and G laciers w ith the A tm osphere and Ocean (P a rt III). Polar
Geography a n d Geology. 14: 155-163
LaChapelle, E.R., a n d R.L. A rm strong. 1977. T e m p e ra tu re patterns in a n
alpine snow cover and th e ir influence on snow m etam o rp h ism .
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Technical report, In stitu te for Arctic and Alpine Research. U niversity
of Colorado. Boulder, CO.
Langham , E .J. 1981. Physics an d properties of snow cover. In Handbook of
snow , eds. D.M .Gray, and D.H. Male. Pergam on Press C a n ad a Ltd,
Willowdale, Ont. pp.275-337.
Langleben, M.P. (1969) Albedo an d degree of puddling of a m elting cover of
sea ice. Journal o f Glaciology. 8:407-413.
Ledley, T.S. 1991. Snow on sea ice: Com peting effects in sh ap in g clim ate.
Journal o f Geophysical Research (D9). 96:17 195-17 208.
LeDrew, E.F. 1990. Influence of Polar regions on C lim ate V ariability a n d
Change., In: Encyclopedia of E a rth System Science, ed. by W .A.
N ierenberg. Academ ic P ress.
Lohanick, A.W. 1993. M icrowave brightness tem p eratu res of laboratorygrown undeform ed first-y ear ice w ith a n evolving snow cover. J o u rn a l
o f Geophysical Research. 98(C3):4667-4674.
Lohanick, A.W. 1990. Some observations of established snow cover on
saline ice an d th e ir relevance to microwave rem ote sensing. In Sea Ice
Properties and P rocesses. Proceedings of the W.W. W eeks Sea Ice
Sym posium , eds. S.A. Ackley and W.W. Weeks, CRREL M o n o g rap h
90-1:61-67, US A rm y Cold Regions Research and E n g in e e rin g
Laboratory, NH, USA.
Lohanick, A.W. and T.C. G renfell. 1986. V ariations in b rig h tn e ss
tem perature over cold first-year sea ice near T uktoyaktuk, NT. J o u rn a l
o f Geophysical Research. 91(C4):5133-5144.
Male, D.H. 1980. The seasonal snowcover. In Dynam ics of snow an d ice
m a sse s, ed. Colbeck, S.C. Academic Press, New York, NY. pp.305-396.
M anabe, S., and R .J. Stouffer. 1988. Two stable equilibria of a coupled
ocean-atm osphere model to g radual changes of atm ospheric CO 2 . P a rt
II: Seasonal response. Jo u rn a l o f Climate. 5:105-126.
M arshall, J.S . and K.L.S. G unn. 1952. M easurem ent of snow p a ra m e te rs
by radar. Journal o f Meteorology. 9:322-938.
M artin, S. 1979. A field study of brine drainage and oil e n tra in m e n t in firstyear sea ice. Journal o f Glaciology. 22:473-502.
M ason, B.J. 1992. Snow crystals, n a tu ra l and m an-m ade. C ontem porary
physics. 33:227-243.
M assom, R. (1991) Satellite Remote Sensing of P olar Rpcnnns
Publishers. Boca R aton, FL.
Lew is
M atzler, C., and U. W egmiiller. 1987. Dielectric properties of fresh w ater ice
a t microwave frequencies. Journal o f Physics D: A pplied P hysics
10:1623-1630.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
M atzler, C., E. S chanda, and W. Good. 1982. T ow ards definition of
optim um sensor specifications for microwave rem ote se n sin g of snow .
IE E E Transactions on Geoscience and Remote Sensing. GE20:57-66.
M aykut, G.A. 1986. C h a p te r 5: T he surface h e at a n d m ass balance. In :
Geophysics of Sea Ic e , ed. N. U ntersteiner. D ordrecht: M a rtin u s
Nijhoff Pub., pp. 395-463.
Maykut, G.A. and N. U n te rste in e r. 1971. Some re su lts from a tim edependent therm ody nam ic model of sea ice. Jo u rn a l o f G eophysical
Research,. 76:550-1575.
Meeks, D.C., G.A. Poe, a n d R.O. Ram seier. 1974. A study of m icrow ave
em ission properties of sea ice - AIDJEX 1972, A erojet E lectrosystem s
Co., U niversity of W ashington. F inal report, 1786FR-1.
Mellor, M. 1977. E n g in eerin g properties of snow. J o u rn a l o f Glaciology.
19:15-66.
Mellor, M. 1965. O ptical properties of snow. CRREL R esearch R eport 169. 20
pp.
M ikhalevsky, P.N., A.B. Baggeroer, A. Gavrilov, a n d M. Slavinsky. 1995.
E xperim ent tests u se of acoustics to m onitor te m p e ra tu re a n d ice in
the Arctic Ocean. EOS T ransactions, A m erican G eophysical U nion,
76:265.
M isurak, K M . 1994. Section 5.2 Snow geophysical p roperties. I n M isu ra k ,
K.M., D.G. B arber, a n d E.F. LeDrew. SIMMS'94 D ata R eport. E a rth
Observation L aboratory Technical Report, ISTS-EOL-SIMS-TR-94-OOl.
Misurak:, K M . 1993. Section 6.1 Snow geophysical p roperties. In M isu ra k ,
K M ., D..G. B arber, a n d E.F. LeDrew. SLMMS'93 D ata R eport. E a rth
Observation L aboratory Technical Report, ISTS-EOL-SIM S-TR-93-007.
Mote, T.L., and M.R. A nderson. 1995. V ariations in snow pack m elt on the
G reenland ice sh e e t based on passive-m icrow ave m e a s u re m e n ts .
Journal o f Glaciology. 41:51-60.
Nakawo, M. and N .K S inha. 1981. Growth Rate a n d S alinity Profile of
F irst-Y ear Sea Ice in th e H igh Arctic. Journal o f Glaciology. 27:315330.
N ansen, F. 1902. The oceanography of the N orth P o lar B asin, N o rw eg ian
N orth Polar Expedition. 1893-1896, Scientific R esearch. V(EX), 427 p.
N eum ann, G. a n d W .J. Pierson, Jr. 1966. P rinciples
Oceanography. Prentice-H all. New Jersey, USA. 545pp.
o f P hysical
NSEDC. 1996. Snow and Ice. 1996. Boulder, CO. 32pp.
Nyfors, E. 1983. On th e dielectric properties of dry snow in th e 800MHz to
13GHz region. H elsinki U niversity Technological Radio Lab. R eport
S135.
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Oke, T.R. 1987. Boundary L ayer Climates. M ethuen & Co., L td. 372 pp.
Onstott, R.G., Grenfell, T.C., M atzler, C., L u th e r, C.A., and E.A.
Svendsen. 1987. E volution of microwave sea ice sig n a tu re s d u rin g
early sum m er and m idsummer in the m a rg in al ice zone. Journal o f
Geophysical Research. 92:6825-6835.
P erla, R. 199L Real p erm ittivity of snow at 1MHz a n d 0°C. Cold Regions
Science and Technology. 19:215-219.
Perovich, D .K and A .J. Gow 1991. A sta tistica l descrip tio n of the
m icrostructure of young sea ice. Journal o f G eophysical R esearch.
96:16 943-16 953.
Rind, D., R. Healy, C. P a rk in so n , and D. M artinson. 1995. T he role of sea
ice in 2X C 02 clim ate sensitivity. P art I: The to ta l influ en ce of sea ice
thickness and extent. Jo u rn a l o f Climate 8:449-463.
Sakazum e, S. and N. Seki. 1978. On the therm al p ro p erties o f ice and snow
in a low te m p e ra tu re region. Japanese Society o f M echanical
Engineers. 46:1119.
Schem enauer, R.S., B erry, M.O., and J.B. M axw ell. 1981. Snowfall
form ation. In H andbook of snow , eds. D .M .G ray a n d D.H. M ale.
Pergam on Press C an ad a L td, Willowdale, O nt. pp. 127-152.
Sm ith, T.G., and I. Stirling. 1977. V ariation in th e d e n sity of ringed seal
(Phoca Hispida) b irth lay ers in the A m undsen G ulf, NW T. C anadian
Journal o f Zoology. 56:1066-1070.
Steffen et al. 1992. The estim atio n of geophysical p a ra m e te rs u sin g passive
microwave algorithm s. In Microwave Remote S e n sin g of Sea Ice.
AGU Geophysical M onograph 68. ed. F. Carsey. pp. 201-231.
Steffen, K , and J.A . M aslanik. 1988. Com parison of N im b u s 7 S can n in g
M ultichannel M icrowave Radiom eter radiance a n d derived sea ice
concentrations w ith L a n d sa t im agery for th e N o rth W ater area of
Baffin Bay. Journal o f Geophysical Research. 93(C9):10,769-10,781.
Tinga, W.R., W.A.G. Voss, a n d D.F. Blossey. G en era liz ed approach to
m ultiple dielectric m ix tu re theory. IEEE T ra n sa ctio n s on A n te n n a s
and Propagation. 44:3897-3902.
T iuri, M.E., A H . Sihvola, E.G. Nyfors and M.T. H allik ainen. (1984) The
Complex Dielectric C o n sta n t of Snow at M icrowave F requencies. IEEE
Journal o f Oceanic Engineering. OE-9:377-382.
T ucker, W.B., III, D .K Perovich, A.J. Gow, W .F. W eeks, an d M.R.
D rinkw ater. 1992. P hysical properties of sea ice re le v a n t to rem ote
sensing. In M icrowave Rem ote Sensing of Sea Ice. A G U G eophysical
M onograph 68. ed. F. C arsey. pp. 10-28.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tucker, W.B., A .J., and J.A . R ichter. 1984. On sm all-scale v a ria tio n s in
first-year sea ice. Jou rn a l o f Geophysical Research. 89:6505-6514.
Ulaby, F.T., R.K. Moore, a n d A.K. Fung. 1986. Microwave Remote S e n sin g :
Active and Passive: Volum e III: From Theory to A p p lica tio n .
Addison-W esley P u b lish in g Company, M assachusetts.
V ant M.R. 1976. A com bined em pirical and theoretical study of th e
dielectric properties of sea ice over the frequency range 100MHz to
40GHz. Technical report, C arleton University, O ttaw a, C anada.
W alker, A.E., and B.E. Goodison. 1993. D iscrim ination of a w et snow cover
using passive m icrow ave satellite data. A nnals o f Glaciology. 17:307311.
W eeks, W.F. and S.F. Ackley, 1986. The growth, stru ctu re, and p ro p e rtie s
of sea ice. in The Geophysics of Sea Ice, ed. N. U n d erstein er. NATO
Series B: Physics vol. 146. P lenum press, New York, USA. Pp 9-164.
Weeks,W.F. and G. Lofgren 1967. The effective solute distribution d u r in g
the freezing of NaCl solutions. Proceedings of the In te rn a tio n a l
Conference on Low T em p eratu re Science, Physics of Snow a n d Ice. ed.
H. Oura, In stitu te of Low T em perature Science, Hokkaido U n iv ersity ,
Sapporo, japan. Pp. 579-597.
Welch, H.E., M.A. B ergm ann, T.D. Siferd and P.S. A m arualik.
1991.
Seasonal developm ent of ice algae near Chesterfield Inlet, N .W .T .,
Canada. Canadian Jo u rn a l o f Fisheries and Aquatic Sciences.48:23952402.
Welch, H.E. and M.A. B ergm ann. 1989. Seasonal developm ent of ice a lg a e
an d its prediction from environm ental factors n e ar R esolute, N T,
Canada. C anadian Jo u rn a l o f Fisheries and A quatic Sciences. 46:17931804.
W ilheit, T., J. Blin, W. C am pbell, A. Edgerton, and W. N ordberg. 1972.
A ircraft m easu rem en ts of microwave em ission from Arctic se a ice.
Remote Sensing o f the E nvironm ent. 2:129-139.
Yoshida, Z. (1955) Physical stu d ies on deposited snow: T herm al p ro p e rtie s.
In stitu te of Low T em p era tu re Science. Hokkaido Univ., S apporo,
Ser.A.No.27, pp. 19-74.
World Meteorological O rganization. 1970. Snow and Ice.
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendices
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A ppendix A: A cronym s & A breviations
A ID JEX
Arctic Ice D ynam ic Jo in t Experim ent
AM SR
Advanced M icrow ave S canning Radiom eter
C-ICE
C ollaborative-Interdisciplinary Cryospheric E x p erim e n t
DMSP
U nited S ta te s Defense Meteorological S atellite P rogram
ERS-1
E a rth R esources S atellite
ESMR
E lectronically S canning Microwave R adiom eter
FY I
F irst-year ice
GCM
G eneral circu latio n models
IPCC
In terg o v ern m en tal P anel on Clim ate C hange
MYI
M ultiyear ice
NASA
N ational A eronautics Space Agency
OW
Open W ater
SBR
Surface B ased R adiom eter
SIMMS
Seasonal Sea Ice M onitoring and Modelling
SMMR
S canning M u ltich an n el Microwave R adiom eter
SSM /I
Special S en so r M icrow ave/Im ager
SWE
Snow w a ter equivalence
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix B: L ist o f Sym bols
Ki
Downwelling shortw ave radiation
KT
U pw elling shortw ave rad iatio n
W avelength o f free space
a
Albedo
e
Complex dielectric constant
e”
Dielectric loss
e
E m issivity
£'
P erm ittivity
GHz
G ig ah ertz
H
H orizontal p olarization
K
Kelvin
^■se
T herm al conductivity of snow
P
D ensity
Si
Ice salin ity
Sw
W ater salin ity of in p a rts p er thousand (%o)
Tf
F reezing te m p e ra tu re of w a ter
V
V ertical polarizatio n
Vi
Volume of ice
°c
D egrees Celcius
Tb
B rightness T em p era tu re
kg
K ilogram
EM
E lectrom agnetic spectrum
vb
Volume of b rin e
%o
p a rts p er th o u sa n d (ppt)
FOV
Field of view
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A ppendix C: L ist o f T erm s
P e n d u la r R egim e: Liquid w a te r content w ithin a snow pack is general less
th a n 1% by volume.
F u n ic u la r R egim e: Liquid w a ter content w ithin a snow pack is g en eral
greater th a n 7% by volum e.
Em issivity: a ratio of the em ission an object rad ia te s com pared to w hat a
blackbody a t th e sam e tem p era tu re would rad iate.
Perm ittivity: A dielectric
m easure
of the am o u n t
of energy w hich
penetrates into a m edium .
Loss: A dielectric m easu re th a t describes w h at h a p p e n s to energy once it
h a s penetrated a m edium .
GR: The gradient ratio. A comparison of 19GHz a n d 37GHz a t V
polarization.
PR: The polarization
ra tio . A comparison
of 19GHz a t V and H
polarizations.
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A ppendix D: N orm ality P lots from Chapter 3
A*All Sam ples Com bined
•
5
I
' ..............................................
* ■
a
3
•
a
.
i
a
•.
*
••
s
^
.
in
lit
>«•
;*
•
3
a
.
.
.
I
i
.
•
IN
It*
a
‘
1
2
•
I
,
I
•
•
3
3
•
I
•
.
•
>H
It*
..
l
’
••
•
■
•
/
•
I**
•••
•••
•
i
•
S
■
•
1
'
■
«•
IM
M
:
alt
V
III
'
•••
•
m
5
•
.
/
:
3
•
•
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
:
B» AM Subset
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A ppednix E: ANOVA R esu lts from Chapter 3
A* AH Sam ples Com bined
ANOVA Table
SWE vs. TB 19H
CF
Regression
1
Residual
48
Total
49
Sum of Squares
31.431
185.506
21 6 .9 3 7
Mean Square
31.431
3.865
F-Value
8.133
P-Value
.0064
ANOVA Table
SWE vs. TB 19V
CF
Regression
1
Residual
53
54
Total
Sum of Squares
12.662
2 1 6 .5 9 2
2 2 9 .2 5 4
Mean Square
12.662
4.087
F-Value
3.098
P-Value
.0841
ANOVA Table
SWE vs. TB 37H
CF
Regression
1
54
Residual
Total
55
Sum of Squares
42.532
188.240
2 30 .7 7 2
Mean Square
42.532
3.486
F-Value
12.201
P-Value
.0010
ANOVA Table
SWE vs. TB 37V
CF
Regression
1
Residual
54
Total
55
Sum of Squares
35.277
195.495
23 0 .7 7 2
Mean Square
35.277
3.620
F-Value
9.744
P-Value
.0029
ANOVA Table
SWE vs. TB 85H
CF
Regression
1
Residual
45
Total
46
Sum of Squares
2 0 .1 4 7
178.841
198.988
Mean Square
20.147
3.974
F-Value
5.069
P-Value
.0293
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ANOVA Table
SWE vs. TB 85V
CF
1
Regression
54
Residual
Total
55
Sum of Squares
4 1 .8 5 0
188.922
2 3 0 .7 7 2
Mean Square
41.850
3.499
F-Value
11.962
P-Value
.0011
OF Sum of Squares
1
6 6 .7 3 7
13
2 8 .5 3 8
14
9 5 .2 7 5
Mean Square
66.737
2.195
F-Value
30.401
P-Value
<.0001
CF
1
14
15
Sum of Squares
7 1 .6 8 4
2 6 .4 5 2
9 8 .1 3 7
Mean Square
71.684
1.889
F-Value
37.940
P-Value
<.0001
CF
1
15
16
Sum of Squares
9 1 .2 2 3
9.736
100.959
Mean Square
91.223
.649
F-Value
140.551
P-Value
<.0001
CF Sum of Squares
1
90.301
15
10.658
16
100.959
Mean Square
90.301
.711
F-Value
127.090
P-Value
<.0001
B« AM Subset
ANOVA Table
SWE vs. T19H
Regression
Residual
Total
ANOVA Table
SWE vs. T19V
Regression
Residual
Total
ANOVA Table
SWE vs. T37H
Regression
Residual
Total
ANOVA Table
SWE vs. T37V
Regression
Residual
Total
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IMAGE EVALUATION
TEST TARGET ( Q A - 3 )
1.0
m
L. imm
U
I 2-2
tu.
IL.
1*0
U,
L_
1.1
IIIIIM
1.8
1.25
1.4
1.6
150mm
A P P L I E D A IIW IG E . I n c
■.^ = ^ 5
1653 East Main Street
Rochester. NY 14609 USA
Phone: 716/482-0300
Fax: 716/288-5989
O 1993. Applied Image. Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
5 409 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа