вход по аккаунту


Satellite radar imagery for monitoring inland wetlands in boreal and sub-arctic environments.

код для вставкиСкачать
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
Published online in Wiley InterScience
( DOI: 10.1002/aqc.836
Satellite radar imagery for monitoring inland wetlands in
boreal and sub-arctic environments
Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, Vienna, Austria
Spatial Information & Mapping Centre, Badan Rekonstruksi dan Rehabilitasi NAD-Nias, Banda Aceh, Indonesia
1. Knowledge about the distribution and types of wetlands is in high demand by ecosystem
modellers for full greenhouse gas accounting. The scope of this paper is to demonstrate the
suitability of satellite radar data for the delineation of wetlands in the tundra and boreal forest
biomes of central Siberia.
2. An area of more than 3 million km2 in central Siberia was investigated using satellite data. It
covers freshwater ecosystems of the tundra and non-forested peatlands in tundra and boreal forest
biomes. The satellite data represent the growing seasons of 2003/2004.
3. Microwave data were acquired by the Advanced Synthetic Aperture Radar (ASAR) instrument
onboard ENVISAT. The multi-temporal capabilities and resolution (150 m 150 m in WS mode) of
the ASAR wide swath mode enabled the detection of dynamic features >2 ha over this vast area.
Scatterometer (QuikScat) data could be employed to distinguish hydro-periods.
4. Wetland types have been identified on the basis of seasonal changes in backscatter. In a first step
scatterometer data were used to identify the transition period from frozen to unfrozen conditions
over a range of 158 latitude. Inundation patterns and soil moisture changes could be identified for the
different hydro-periods and used to classify wetlands. Results for peatlands have been compared
with Russian forest inventory data which contain information on wetland distribution.
5. The database of permanently inundated areas is an intermediate product which enables the
mapping of wetlands in two ways: (1) identification of seasonal inundation in relation to snowmelt
and high permafrost tables and (2) input for density analysis of permanent small and shallow lakes in
tundra areas which are important freshwater ecosystems as well as a methane source. Differences in
intensity and duration of soil moisture conditions allow the identification of peatlands.
Copyright # 2007 John Wiley & Sons, Ltd.
remote sensing; ScanSAR; scatterometer; peatland; tundra; taiga; Siberia
*Correspondence to: A. Bartsch, Institute of Photogrammetry and Remote Sensing, Vienna University of Technology
GusshausstraXe 27–29, 1040 Vienna, Austria. E-mail:
Copyright # 2007 John Wiley & Sons, Ltd.
Wetlands in the tundra and boreal forest biome play an important role for storage and emission of carbon
and other greenhouse gases such as methane. Their extent and role as a sink or source is still uncertain
(Callaghan et al., 2004a), especially in Siberia (Friborg et al., 2003). Changes of wetlands in those biomes
are caused by climate change, fires and to some extent direct human impact. Increasing air temperatures,
together with changes in snow depth, influence permafrost-related processes ranging from changes in the
thickness of the active layer, enhanced alas thermokarst formation, and complete thawing (Osterkamp and
Romanovsky, 1999). This is especially important in regions of sporadic and discontinuous permafrost
where small changes can have large impacts. It depletes the capacity of peatlands to act as carbon sinks or
storage. Carbon-rich material might become available for decomposition after permafrost melt and drying
out. On the other hand this may cause moisture increase in depressions, which supports carbon
sequestration, and, in the case of inundation, methane emission (Callaghan et al., 2004b). Although current
methane emissions at high latitudes are comparably low (13% of global emissions) the global warming
potential is high, so these regions are very important in terms of climate change (Callaghan et al., 2004b).
Both methane fluxes and dissolved carbon dioxide in surface waters are insufficiently accounted for in
models, owing to limited knowledge on the emissions themselves and the spatial extent of sources.
The effective use of satellite imagery in mapping northern peatlands for monitoring and carbon cycle
modelling has already been emphasised by Gorham (1991) more than a decade ago. Nowadays, active as
well as passive satellite sensors are employed for wetland monitoring all around the world (Ozesmi and
Bauer, 2002) within local studies but rarely for large areas in northern latitudes. Studies on global wetland
dynamics in general have focused on inundation (Prigent et al., 2001) but are based on analysis of
vegetation patterns. This approach has only been validated in the tropics and sub-tropics. Northern
wetlands are mostly unaccounted for both in extent and function.
Information obtained from radar includes not only inundation but also (with decreasing degree of
confidence) freeze/thaw changes, topography, soil moisture, and above-ground phytomass changes. In the
boreal and tundra biome wetland types have been investigated with respect to biogeochemical processes
(Morrisey et al., 1994), seasonal thawing (Murphy et al., 2001), periodic inundation (Töyrä et al., 2002) and
incidence angle and polarization of the radar signal (Baghdadi et al., 2001).
Most space-borne radar remote sensing studies on wetlands either aim to distinguish wetlands from other
land cover, or look at different subtypes within an already identified wetland area (e.g. Thomas et al., 2002).
Classes are selected solely by data capabilities rather than by research requirements. Only a few studies have
addressed the question of whether specific classification schemes used for purposes such as wetland
conservation and greenhouse gas accounting can be applied when analysing radar data. The Canadian
wetlands classification scheme (CWCS) as described in Warner and Rubec (1997) has been used by
Mogghadam et al. (2003) and Morrisey et al. (1994, 1996). The latter also analysed the relationship between
methane emission and wetland types in arctic environments. Another scheme based on multi-temporal
observations of seasonal and periodic flooding events was proposed by Milne et al. (2000). This hydro-period
classification is applicable in regions where large-scale and temporally highly variable flooding occurs.
This study was carried out within the framework of the Siberia II project, which deals with multi-sensor
concepts for greenhouse gas accounting in northern Eurasia (Schmullius et al., 2003). The region of interest,
located in central Siberia and covering more than 3 million km2, is representative of the tundra as well as
the boreal forest biome. ENVISAT (ENVIronmental SATellite), ScanSAR (C-band) and QuikScat (Kuband) scatterometer data are utilized. These space-borne data have so far only been combined for
oceanographic applications (i.e. Bentz et al., 2004). This paper presents an approach which exploits the
capabilities of both sensor types (with differing spatial and temporal resolution) and wavelengths for
wetland identification and monitoring. The classification results are compared with data sets available
within the Siberia II project provided by Russian forest enterprises.
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
Data from two different sensors were combined to address the spatial and temporal variability that is
characteristic for the area of interest which spans several biomes (Figure 1). The ScanSAR and
scatterometer data were first separately pre-processed and classified. The end of freeze/thaw transition in
Figure 1. Location of Siberia-II area (inset map) and landscape groups (source: IIASA, Laxenburg, Austria); case study sites for
wetland types 2 (Norilsk) and 3 (middle taiga) are added.
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
spring was obtained from the QuikScat/SeaWinds scatterometer data (Kidd et al., 2004). Utilizing these
data in a pre-selection step, inundation patterns were identified with ASAR WS images.
Study area and wetland types
The Siberia II project study area stretches south–north from Lake Baikal to the Taymir peninsula. The
northern part belongs to the tundra biome and its lowlands are characterized by vast polygon mire systems.
Palsa mires extend to the south (Botch and Masing, 1983) into the forest tundra. Peat bogs are found in the
middle and southern part within the boreal forest biome at the eastern rim of the West Siberian peat basin.
The wetland types chosen for this investigation represent both mesotrophic wetlands with thin peat layers
as found in the tundra as well as peatlands of the middle taiga vegetation zone (Table 1). The most
northerly area of interest is the entire moderate sub-arctic zone which is part of the tundra biome. Another
specific site is found in the transition zone between sub-arctic moderate conditions to boreal continental.
Seasonal inundation occurs here mostly due to snow melt. Although it is located further south it still
features continuous permafrost whereas the region with the third investigated type (bogs with pools and
ridges; see Table 1) only shows sporadic permafrost (Stolbovoi and McCallum, 2002). Here peatlands occur
in a boreal continental climate with middle taiga vegetation. Seasonal changes in soil moisture, soil
dielectric properties, and seasonal dynamics of vascular plant growth are parameters which influence
microwave backscatter. Because of the topography no large peatland complexes are found east of the
Yenisey river and in the southern part of the boreal biome.
The wetland types considered are listed in Table 1. The Russian wetlands classification scheme (RWCS)
can be related to the Canadian scheme (Bartsch et al., 2004) which has often been used for radar remote
sensing studies at these latitudes. In addition, the classes specified by the Ramsar Convention for natural
inland wetlands have been assigned in Table 1 to enhance comparability. Freshwater and saline lakes are
not dealt with separately since they cannot be distinguished with the chosen methodology and data.
Case study regions
The complete sub-arctic and arctic climate zones of the Taimyr Peninsula have been investigated for
delineation of tundra wetlands (Vt). Peatlands (wetland type 2 and 3) have been extracted for selected
regions only (for location see Figure 1). Permanent inundation is mapped for the whole Siberia II area.
The sub-arctic zone on southern Taimyr is of plain to moderate terrain and dominated by glacial
deposits. Small and shallow lakes are abundant. The water table is close to the surface and varies with thaw
depth of the active layer. These features support a vast mire system which is of comparatively low
importance for current carbon sequestration but methane emissions are increased through coupling with
Table 1. Description of investigated wetland types according to a Russian wetlands classification scheme (RWCS), the international
Ramsar classification scheme (including code) and general trophic conditions
(sub-)arctic mineral sedge mires
bogs of southern tundra and
northern taiga
bogs with pools and ridges of
northern and middle taiga
permanent lakes
Copyright # 2007 John Wiley & Sons, Ltd.
Trophic condition
tundra wetlands
non-forested peatlands, fens
non-forested peatlands
permanent freshwater and
saline lakes (>8 ha)
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
active layer dynamics (Nakano et al., 2000). Methane fluxes in Siberian wet tundra constitute up to 5% of
total heterotrophic respiration (Christensen et al., 1996). The Siberian lake sediments which act as source
originate from the Pleistocene and emission from them accounts for half of annual methanogenesis in this
region (Zimov et al., 1997).
According to Birdlife International (, the Taimyr wetland tundra is one of the
key wetland regions for threatened waterfowl in Asia. These ecosystems are prone to permafrost
degradation owing to climate warming and human impacts (Pavlov and Moskalenko, 2002). Thermokarst
processes are intensified by industrial activities especially in the south-western Taimyr (Stolbovoi and
McCallum, 2002).
The chosen peatland areas correspond to specific forest inventory sites (see the subsection ‘Reference
data’). The Norilsk site (100 km south of the town) is located in the forest tundra between the Yenisey River
and Putorana plateau and covers part of the Khantaiskoe reservoir. Although it is partly covered by forest
(larch and birch) stocking (tree cover with respect to possible stocking) is very low and generally under
40%. This region is severely affected by industrial pollution (especially sulphur) because of its proximity to
the Norilsk heavy metal smelter complex (Zubareva et al., 2003). Influence on aquatic habitats in the
Taimyr lowlands to the north has been detected up to a maximum distance of 100 km from the mining town
(Allen-Gil et al., 2003).
The inventory site of Nizhne Yenisky is located west of the Yenisey River in the middle taiga. Open bogs
are an important land-cover type in this region albeit not dominating. These peatland types (bogs with
pools and ridges; high moor peat and podzols with peat layers) are characteristic of the eastern portion of
Western Siberia (Botch, 1999). Coniferous forest with high stocking covers most of the area. Human impact
is limited to logging which may, together with fires, encourage the re-formation of open wetland areas.
These peatlands are important for discharge, geochemistry and sedimentology of the Yenisey River
(Kremenetski et al., 2003).
Reference data
Ground information on wetlands and water bodies is available at different levels: global, regional and local.
Global databases are the Global Lakes and Wetlands Database (GLWD; Lehner and Döll, 2004) based on
the Digital Chart of the World, and the 1-degree grid by Matthews and Fung (1987). The latter is widely
used for modelling purposes and, compared with other sources on this level, it is the most reliable (Mitra
et al., 2005). However, it is too coarse to serve as a reference database for this study. A Russian national
database which is based on topographic maps exists for water bodies. Both vector-based databases, the
GLWD and Russian data base rely on fairly old sources and contain only features larger than 10 ha and
15 ha, respectively. Within the Siberia II project, digital data (vector layers) with high detail have been
provided via the International Institute of Applied Systems Analysis (IIASA) from the Sukachev Institute
(Krasnojarsk, Russia), which are designed for forest inventory purposes. This information covers the extent
of open bogs and permanent water bodies, and is mostly based on fieldwork and aerial photography
interpretation. Soil data by the Dokuchaev Soil Institute (Moscow, Russia) are also available through
Sensors and data coverage
This study utilised ENVISAT ASAR data acquired in wide swath (WS) mode in vertical polarization (VV).
The pixel spacing was 75 m which corresponded to an approximate spatial resolution of 150 m. Each swath
covered an area of 405 km width. Approximately 100 scenes were processed for each of the summer periods
2003 and 2004. These together provided an almost complete coverage (95%) enabling permanent
inundation (lakes and rivers) mapping for the whole study area. The number of spring and autumn scenes
varied significantly and therefore did not allow a spatially complete analysis of wetland dynamics.
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
SeaWinds is a currently operational scanning dual spot beam scatterometer, launched in June 1999
aboard the NASA QuikScat satellite (Spencer et al., 2000). It operates at 13.4 GHz (Ku-band). The antenna
has a footprint size of roughly 25 km 25 km and scans over a swath of 1800 km, imaging 90% of the
earth’s surface in one day.
In addition, spatially interpolated meteorological data from more than 200 stations provided by the
World Meteorological Organization (WMO) were used for the evaluation of autumn data in regions of
discontinuous permafrost when active layer depth decreases.
ASAR WS data processing
SAR data require georeferencing with respect to the earth’s curvature and terrain. The geocoding of the
ASAR WS data is based on the Range–Doppler approach for performing a backward geocoding, realized
using commercial and in-house software solutions (Bartsch et al., 2004). This approach does not require tie
points, if the sensor geometry is known precisely, which is the case for ENVISAT ASAR. The geometric
accuracy of this geocoding approach mainly depends on the accuracy of the sensor position and velocity
vectors, the measurement accuracy of the pulse delay time and the knowledge of the target height relative to
the assumed Earth model given by a digital elevation model. Digital elevation data of sufficient resolution
are only available below 60 8N (from the Shuttle Radar Topography Mission – SRTM, 100 m 100 m).
However, since wetlands occupy mostly flat regions and the terrain in the study area is moderate in higher
latitudes, the earth-curvature-based correction is sufficient.
Within a normalization step the effects on the backscatter due to varying incidence angle and distance
from sensor (near and far range) were removed. Normalization was carried out by fitting a linear model to
the backscatter data, to get an estimate of the slope in units of decibels per degree incidence angle which
characterizes the decrease of the radar backscatter from near range to far range. The corresponding
incidence angles were calculated as a by-product during the geometric correction. Then the new backscatter
values were calculated for a mean reference incidence angle of 308.
Scatterometer data processing
Backscatter measurements from SeaWinds/QuikScat, currently extending from July 1999 to December
2003, were re-gridded and reformed into time series allocated to a regular, global coverage of grid points,
with a 10 km 10 km grid spacing.
Scatterometers are well suited to monitor freeze–thaw events (Wismann, 2000; Nghiem and Tsai, 2001).
At the onset of snow melt, diurnal changes in the snow pack, from a wet surface or wet snow pack in the
evening and refrozen snow pack in the morning, cause strong diurnal effects in backscatter. Exploiting the
high temporal resolution of the SeaWinds/QuikScat, these diurnal effects within the backscatter time series
were observed, and the dates of onset of thaw, and the duration of diurnal thaw and refreeze period, were
detected (Kidd et al., 2004).
Combination of scatterometer and ScanSAR data
The identification of wetlands requires both spatial and temporal pre-classification. Inundation can mostly
be associated with snow melt. Its patterns differ over the Siberia II region. The growing season can be
subdivided into three hydro-periods. The first phase starts after the end of freeze/thaw transition and ends
with the time when the groundwater table goes back to normal summer conditions. Mid-summer until midautumn falls into the rather stable second period. The third phase covers the time from first temperature
drop below 08C during the night until snow covers the landscape or the ground is mostly frozen. Peatlands
and tundra wetlands can be associated with specific biomes. This information was used in the pre-selection
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
The dates available for onset of thaw derived from QuikScat data (Kidd et al., 2004) were used to select
relevant ASAR WS images for spring and summer periods. This, and the following analysis, were carried
out with a Geographic Information System (ArcGIS). Daily temperature data were used to determine the
first date of freeze in autumn. These dates were assigned to the coordinates of the station and a spatial
interpolation applied by inverse distance weighting (Philip and Watson, 1982) to 10 km 10 km pixel
spacing. Data were further assigned to three landscape groups according to the biome into boreal forest,
tundra (sub-arctic/artic) and a transition zone between them.
Classification of ASAR WS backscatter values
Open water surfaces can be identified using a simple threshold-based classification applied to the
normalized image data. Specular reflection from calm water surfaces results in low backscatter. This
phenomenon enables a straightforward identification of inundation in areas with limited vegetation cover.
Two types of inundation were considered in this investigation: permanent inundation and seasonal
inundation. The threshold for permanent inundation was determined from mid-summer data of known
permanent water surfaces. It was set to 14 dB for the normalized images. Inundation during the spring
period was then identified by change detection: low backscatter surfaces during spring which are above the
set threshold in the summer data are only seasonally inundated. Low values similar to specular reflection
from water can occur in regions of radar shadow; high values can also be the result of foreshortening in
mountainous regions. Both effects occur in terrain where no large wetland complexes are expected.
Therefore, regions with high variation in elevation and steep slopes, respectively, were derived from digital
elevation data (SRTM and GTOPO30) and excluded from the analysis in the southern mountainous part
(Wagner et al., 2003a).
Within regions of low human impact and flat to moderate terrain, a high backscatter signal in C-band is
caused either by double-bounce effect within vegetation in water such as reeds, or by high soil moisture
conditions. The latter backscatter behaviour is characteristic for open bogs in the boreal forest biome.
Changes in soil moisture in relation to the seasonal dynamics in active layer thickness are detectable in
regions where the groundwater table is in general close to the surface. Man-made objects can also cause
double bounce, and thus high backscatter, but these are negligible in the chosen case study areas and in this
region in general.
In summary, thresholds were set to separate areas of low (water), medium (in general forest or tundra)
and high backscatter (double bounce or high soil moisture). These values are chosen independently from
acquisition date or location. This limits the detection of large water bodies such as reservoirs at some dates
owing to wave action but enables efficient processing of the large amount of data.
Post-processing of ASAR WS classification results
Permanent open water bodies play an important role in wetland identification where they are shallow
(Ramsar type Tp). They are abundant in the tundra biome where permafrost is continuous and ice content
high. This means not simply an abundance of water but a high number (density) of small features.
Permanently inundated basins smaller than 8 ha in the sub-arctic regions indicate tundra wetlands
according to the Ramsar classification scheme. They were separated from the derived water layer of the
complete region and subjected to density analyses (50 km search radius). Based on information from
topographic maps and especially Ramsar sites on the Taimyr Peninsula a density threshold was set to
5000 m2 km2 of water surface area from features between 2 and 8 ha for the delineation of Ramsar wetland
type Vt. This analysis was carried out using an available ArcGIS density function. Digitized outlines of the
Ramsar sites were used to mask the results from the density analyses. The threshold could be determined
from the standard deviation of the clipped raster data.
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
Permanent inundation
The minimum detectable lake size when using ASAR WS is approximately 2 ha. The northern part of the
studied region with arctic and sub-arctic environments (tundra biome) has 8.4% (20.383 km2) of water
cover compared with approximately 1% in the boreal parts and more than 39 000 water bodies as identified
by ASAR WS. If smaller detectable lakes (2–8 ha) are taken into account the number of lakes in the subarctic moderate zone rises to more than 150 000 which totals 2500 km2 of additional water surface for
natural lakes. Regions where surface water from these small lakes adds up to more than 5000 m2 km2 are
shown in Figure 2. All Ramsar inland wetlands on southern Taimyr are located within the extracted tundra
wetland zone. The density range for Pura and Gorbita is 5000 to 7000 m2 km2 for lakes between 2 and 8 ha
open water surface.
Seasonal variations
Backscatter behaviour differs over the year owing to seasonal inundation dynamics and processes in the
active layer (especially changes in soil water content) that distinguish bogs from tundra wetlands.
Representative average monthly backscatter values for wetlands and other characteristic land-cover classes
are presented in Figure 3, for (a) the tundra biome and (b) the boreal forest biome. Small lakes in the
moderate sub-arctic zone have high backscatter in winter owing to multiple scattering effects in the ice
cover. In some cases these can drop to the same level as in summer if the ice layer reaches the lake bed
(Duguay and Lafleur, 2003). Permanently inundated basins of the tundra can be clearly identified during
the summer months of July and August. Within the thaw period backscatter values drop for open bogs in
the forest tundra transition zone. Values correspond to those from large water surfaces in the same region
(such as reservoirs). This supports the assumption that spring inundation occurs here and is also important
for the maintenance of this wetland type. During summer backscatter equals that from adjacent areas with
larch and birch stands. Backscatter from tundra is fairly stable over the whole year. Backscatter from open
peatlands in the middle taiga vegetation zone varies by soil type and dielectric properties, respectively (see
Figure 3(b)). High moor peatlands (raised bogs) show fairly high backscatter during summer which peak
towards autumn. This can be explained by high soil moisture which might be conditioned by sporadic
permafrost that can be found especially in these bog types in that region. The autumn peak of peaty podzol
corresponds to seasonal meteorological changes. When temperature drops below 08C in autumn, the soil
dielectric properties change significantly. This causes increased backscatter values, and continues almost
until the ground is fully frozen or covered by snow. Comparison with WMO data showed that this time
period is limited to about 2 weeks in this region. An example classification result for the middle taiga zone
of the Siberia II region is presented in Figure 4. A comparison with forest inventory data from site Nizhne
Yenisky 2 is shown as well. Podzols with peat layers and also wet alluvial soils have very similar values to
the surrounding coniferous forest except in autumn.
This study presents data from only two consecutive years. Spring flooding and autumn changes could only
be detected for one season (2003) owing to a low temporal sampling rate in 2004. Therefore, it could be
assumed that soil-related backscatter changes in the boreal biome might have been specific for 2003. The
backscatter increase in autumn, however, seems to occur regularly as it has already been employed by Im
et al. (2003) to identify seven bog types along the Yenisey River with one Radarsat (C-HH) scene from late
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
Figure 2. Tundra wetland area (Vt, see Table 1) from surface water density analysis, thermokarst features (Stolbovoi and McCallum,
2002) and location of Ramsar sites (1, Brekhovsky islands; 2, Pura and Moritto rivers; 3, Gorbita. Source: Wetlands International –
Russia Program,, April 2005; Krivenko, 1999).
Figure 3. Backscatter characteristics (average values) from normalized ASAR WS data for two different land-cover types for 2003
(winter values from 2004): (a) the sub-arctic zone and (b) the boreal continental (middle taiga) zone; start of thaw in 2003 indicated
with a vertical bar (as derived from QuikScat data).
autumn 1999. These wetland types, however, have been mostly explained by differing tree species cover and
The density of lakes in the tundra biome which can be associated with Ramsar Wetlands of International
Importance is particularly important for wetland conservation. This relationship between the abundance of
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
peat and peaty-gley podzols as derived from ASAR WS
open wetlands from forest inventory
10 m
high moor and
transitional moor peat soils
peat and peaty-gley podzol
wet alluvial soils
inventory site Nizhne Yenisky 2
100 km
Figure 4. (a) Non-forested peatlands (Ramsar type U) in the middle taiga zone (boreal continental) as identified from ASAR WS,
features >10 ha and location of inventory site Nizhne Yenisky 2. (b) Inventory site Nizhne Yenisky 2 (608 400 N, 888 300 E); open bogs
from inventory (all other is middle taiga forest) and peatlands as derived from ASAR WS (only features greater than 2 ha).
small lakes and regions which are determined as suitable habitat for waterfowl could be used to delineate
such areas in the entire circumpolar region. The permanent inundation mapping approach presented here is
based on currently operational satellites, but the input information can be replaced by data from other
similar active microwave sensors since the mapping of open water surfaces is a straightforward task and
thus can be easily applied to other SAR satellite systems.
The chosen multi-temporal approach demands very good data availability as well as processing and data
storage facilities. Constraints differ by wetland type. In order to map permanent inundation and thus
tundra wetlands complete data coverage for the summer period is required. If seasonal inundation is
playing an important role (such as for bogs in the forest tundra), data from a very specific time period are
needed. The revisit interval of the satellite needs to be short enough to ensure that the flooding event is
captured. This high temporal resolution should be available throughout the entire growing season in order
to account for the latitudinal gradient or range of onset and end of snow melt, respectively. The same
applies to peaty podzol in the middle taiga region where the backscatter peaks for a short time in autumn.
The identification of high moor peat (raised bogs) is not limited to these short periods but to the summer
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
time as well. The ENVISAT ASAR instrument complies with these requirements where there are no user
conflicts concerning the acquisition mode.
This study used the wide swath mode of ENVISAT ASAR since it provides data with a sufficient
resolution of 150 m 150 m to capture small features which are important in the tundra region. Coarser
data are provided by the second ASAR ScanSAR mode (1 km 1 km, global monitoring mode) but with a
much better temporal coverage. This yields a high-potential for detection of high-latitude peatland basins.
The soil moisture changes caused by permafrost melt (Callaghan et al., 2004b) may be monitored for
peatlands with radar data since the backscatter is strongly influenced by wetness, as demonstrated for the
middle taiga case study site. Wetlands in rather dry environments, where snow melt patterns and limited
percolation caused by permafrost sustain wetlands such as found in the tundra region, are especially
susceptible to climate change (Price et al., 2005).
The approach described here combines the advantages of the QuikScat and ASAR sensors. The high
temporal resolution of QuikScat allows the detection of diurnal changes of surface properties whereas
ASAR WS provides data with a sufficient resolution to capture features which are important, for example,
in tundra wetland mapping. Both sensors allow monitoring over large regions, whereas QuikScat can be
used to retrieve information at a global scale as is already available for soil moisture from the ERS
scatterometer (Wagner et al., 2003b).
This methodology could be transferred to lowlands of the boreal severe climate zone, where permafrost
ice content is as high as on the Taimyr lowlands and thus numerous small lakes occur. Changes are
expected to occur especially in this region (Smith et al., 2005) but also in Alaska (Stow et al., 2004) through
climate change. These lakes and associated wet tundra soils are also of specific interest for modelling
biogeochemical cycles (Zelenev, 1996; Gal’chenko et al. 2001). Methane emissions occur especially in fens
but also from shallow lakes of the Siberian tundra biome (Zimov et al., 1997).
Possible applications of the wetland maps presented for central Siberia are the delineation of areas of
high ecological importance (e.g. for waterfowl), comparison with information from other sources, and thus
monitoring, and update of land-cover information as used by ecosystem models to improve greenhouse gas
This research has been carried out in connection with two research projects: MISAR (Austrian Science Fund, P16515N10) and Siberia II (5th Framework Programme of the European Commission, Generic Activity 7.2: Development of
Generic Earth Observation Technologies, EVG1-CT-2001-00048). ASAR WSM data are available by courtesy of ESA,
and SRTM3 by USGS. Thanks also to Anatoly Shvidenko and Ian McCallum from the International Institute of
Applied Systems Analysis (Laxenburg, Austria) for extensive support regarding the Russian ground data.
Allen-Gil SM, Ford J, Lasorsa BK, Monetti M, Vlasova T, Landers DH. 2003. Heavy metal contamination in the
Taimyr Peninsula, Siberian Arctic. The Science of the Total Environment 301: 119–138. DOI: 10.1016/S00489697
Arzandeh S, Wang J. 2002. Texture evaluation of RADARSAT imagery for wetland mapping. Canadian Journal of
Remote Sensing 28: 653–666.
Baghdadi N, Bernier M, Gauthier R, Neeson I. 2001. Evaluation of C-Band SAR data for wetlands mapping.
International Journal of Remote Sensing 22: 71–88.
Bartsch A, Kidd R, Pathe C, Shvidenko A, Wagner W. 2004. Identification of wetlands in central Siberia with
ENVISAT ASAR WS data. Proceedings of the ENVISAT Symposium, 6–10 September 2004, Salzburg. ESA:
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
Bentz CM, Lorenzzetti JA, Kampel M. 2004. Multi-sensor synergistic analysis of mesoscale oceanic features:
Campos Basin, South-Eastern Brazil. International Journal of Remote Sensing 25: 4835–4841. DOI: 10.1080/
Botch MS (ed.). 1999. Wetlands in Russia: Volume 2, Important Peatlands. Global Series 2, Wetlands International:
Botch MS, Masing VV. 1983. Mire ecosystems in the U.S.S.R. In Mires: Swamp, Bog, Fen and Moor. Regional Studies;
Ecosystems of the World 4B, Gore AJP (ed.). Elsevier: Amsterdam; 95–152.
Callaghan TV, Björn LO, Chernov Y, Chapin T, Christensen TR, Huntley B, Ims RA, Johansson M, Jolly D,
Jonasson S, Matveyeva N, Panikov N, Oechel W, Shaver G, Schaphoff S, Sitch S, Zöckler C. 2004a. Climate change
and UV-B impacts on Arctic tundra and polar desert ecosystems: key findings and extended summaries. Ambio 33:
Callaghan TV, Björn LO, Chernov Y, Chapin T, Christensen TR, Huntley B, Ims RA, Johansson M, Jolly D, Jonasson
S, Matveyeva N, Panikov N, Oechel W, Shaver G, Schaphoff S, Sitch S. 2004b. Effects of changes in climate on
landscape and regional processes, and feedbacks to the climate system. Ambio 33: 459–468.
Christensen TR, Prentice IC, Kaplan J, Haxeltine A, Sitch A. 1996. Methane flux from northern wetlands and tundra.
Tellus B 48: 652–661.
Duguay CR, Lafleur PM. 2003. Determining depth and ice thickness of shallow sub-arctic lakes using space-borne
optical and SAR data. International Journal of Remote Sensing 24: 475–489. DOI: 10.1080/01431160210144507.
Friborg T, Soegaard H, Christensen TR, Lloyd CR, Panikov NS. 2003. Siberian wetland: where a sink is a source.
Geophysical Research Letters 30: 2129–2132. DOI: 10.1029/2003GL017797.
Gal’chenko FV, Dulov LE, Cramer B, Konova NI, Barysheva SV. 2001. Biogeochemical processes of methane cycle in
the soils, bogs, and lakes of Western Siberia. Microbiology 70: 215–225.
Gorham E. 1991. Northern peatlands: role in the carbon cycle and probable responses to climate warming. Ecological
Applications 1: 182–195.
Im ST, Gorondzankina SM, Kharuk VI, Ranson KJ. 2003. Radarsat data for Siberian plain ecosystems classification.
Proceedings IGARSS ’03, 21–25 July 2003, Toulouse. IEEE: Piscataway, NJ; 2562–2564.
Kidd R, Bartsch A, Wagner W. 2004. Development and validation of a diurnal difference indicator for freeze–thaw
monitoring in the Siberia II project. Proceedings ENVISAT Symposium, Salzburg.
Kremenetski KV, Velichko AA, Borisova OK, MacDonald GM, Smith LC, Frey KE, Orlova LA. 2003. Peatlands of
the Western Siberian lowlands: current knowledge on zonation, carbon content and late quaternary history.
Quaternary Science Reviews 22: 703–723. DOI: 10.1016/S0277-3791(02)00196-8.
Krivenko VG (ed.). 1999. Wetlands in Russia: Volume 1, Wetlands of International Importance. AEME Publication 52,
Wetlands International: Moscow.
Lehner B, Döll P. 2004. Development and validation of a global database of lakes, reservoirs and wetlands. Journal of
Hydrology 296: 1–22. DOI: 10.1016/j.jhydrol.2004.03.028.
Matthews E, Fung I. 1987. Methane emission from natural wetlands: global distribution, area, and environmental
characteristics of sources. Global Biochemical Cycles 1: 61–86.
Milne AK, Horn G, Finlayson M. 2000. Monitoring wetlands inundation patterns using RADARSAT multitemporal
data. Canadian Journal of Remote Sensing 26: 133–141.
Mitra S, Wassmann R, Vlek PLG. 2005. An appraisal of global wetland area and its organic carbon stock. Current
Science 88: 25–35.
Moghaddam M, McDonald K, Cihlar J, Chen W. 2003. Mapping wetlands of the North American boreal zone from
satellite radar imagery. Proceedings IGARSS ’03, 21–25 July 2003, Toulouse. IEEE: Piscataway, NJ; 500–502.
Morrissey LA, Livingston GP, Durden SL. 1994. Use of SAR in regional methane exchange studies. International
Journal of Remote Sensing 15: 1337–1342.
Morrissey LA, Durden SL, Livingston GP, Stearn JA, Guild LS. 1996. Differentiating methane source areas in arctic
environments with multitemporal ERS-1 SAR data. IEEE Transactions on Geoscience and Remote Sensing 34:
Murphy MA, Martini IP, Protz R. 2001. Seasonal changes in subarctic wetlands and river ice breakup detectable
on RADARSAT images, southern Hudson Bay lowland, Ontario, Canada. Canadian Journal of Remote Sensing 27:
Nakano T, Kuniyoshi S, Fukuda M. 2000. Temporal variation in methane emissions from tundra wetlands in a
permafrost area, northeastern Siberia. Atmospheric Environment 34: 1205–1213.
Nghiem SV, Tsai WT. 2001. Global snow cover monitoring with spaceborne Ku-band scatterometer. IEEE
Transactions on Geoscience and Remote Sensing 39: 2118–2134.
Osterkamp TE, Romanovsky VE. 1999. Evidence for warming and thawing of discontinuous permafrost in Alaska.
Permafrost and Periglacial Processes 10: 17–37.
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
Ozesmi SL, Bauer ME. 2002. Satellite remote sensing of wetlands. Wetlands Ecology and Management 10: 381–402.
DOI: 10.1023/A:1020908432489.
Pavlov AV, Moskalenko NG. 2002. The thermal regime of soils in the north of Western Siberia. Permafrost and
Periglacial Processes 13: 43–51.
Philip GM, Watson DF. 1982. A precise method for determining contoured surfaces. Australian Petroleum Exploration
Association Journal 22: 205–212.
Price JS, Branfireun BA, Waddington JM, Devito K. 2005. Advances in Canadian wetland hydrology, 1999–2003.
Hydrological Processes 19: 201–214.
Prigent C, Matthews E, Aires F, Rossow WB. 2001. Remote sensing of global wetland dynamics with multiple satellite
data sets. Geophysical Research Letters 28: 4631–4634.
Schmullius C, Hese S, Knorr D. 2003. Siberia-II – a multi sensor approach for greenhouse gas accounting in northern
Eurasia. Petermanns Geographische Mitteilungen 147: 4–5.
Smith LC, Sheng Y, MacDonald GM, Hinzman LD. 2005. Disappearing arctic lakes. Science 308: 1429.
Spencer MW, Wu C, Long DG. 2000. Improved resolution backscatter measurements with the SeaWinds pencil-beam
scatterometer. IEEE Transactions on Geoscience and Remote Sensing 38: 89–104.
Stolbovoi V, McCallum I. 2002. CD-ROM Land Resources of Russia, International Institute for Applied Systems
Analysis and the Russian Academy of Science: Laxenburg, Austria.
Stow DA, Hope A, McGuire D, Verbyla D, Gamon J, Huemmerich F, Houston S, Racine C, Sturm M, Tape K,
Hinzman L, Yoshikawa K, Tweedie C, Noyle B, Silapaswan C, Douglas D, Griffith B, Jia G, Epstein H, Walker D,
Daeschner S, Petersen A, Zhou L, Myneni, R. 2004. Remote sensing of vegetation and land-cover change in arctic
tundra ecosystems. Remote Sensing of Environment 89: 281–308.
Thomas V, Treitz P, Jelinski D, Miller J, Lafleur P, McCaughey JH. 2002. Image classification of a northern peatland
complex using spectral and plant community data. Remote Sensing of Environment 84: 83–99. DOI: 10.1016/
Töyrä J, Pietroniro A, Martz LW, Prowse TD. 2002. A multi-sensor approach to wetland flood monitoring.
Hydrological Processes 16: 1569–1581.
Wagner W, Luckman A, Vietmeier J, Tansey K, Balzter H, Schmullius C, Davidson M, Gaveau D, Gluck M, Le Toan
T, Quegan S, Shvidenko A, Wiesmann A, Jiong Yu JJ. 2003a. Large-scale mapping of boreal forest in Siberia using
ERS tandem coherence and JERS backscatter data. Remote Sensing of Environment 85: 125–144.
Wagner W, Scipal K, Pathe C, Gerten D, Lucht W, Rudolf B. 2003b. Evaluation of the agreement between the first
global remotely sensed soil moisture data with model and precipitation data. Journal of Geophysical Research –
Atmospheres 108: 4611. DOI: 10.1029/2003JD003663.
Warner BG, Rubec CDA. 1997. The Canadian Wetland Classification System. National Wetlands Working Group,
Wetlands Research Centre, University of Waterloo: Waterloo, Ontario.
Wismann V. 2000. Monitoring of seasonal thawing in Siberia with ERS scatterometer data. IEEE Transactions on
Geoscience and Remote Sensing 38: 1804–1809.
Zelenev VV. 1996. Assessment of the average annual methane flux from the soils of Russia. Publication Number WP96-51, International Institute for Applied Systems Analysis: Laxenburg, Austria.
Zimov SA, Voropaev YV, Semiletov IP, Davidov SP, Prosiannikov SF, Chapin FS, Chapin MC, Trumbore S, Tyler S.
1997. North Siberian lakes: a methane source fuelled by Pleistocene carbon. Science 277: 800–802.
Zubareva ON, Skripal’shchikova LN, Greshilova NV, Kharuk VI. 2003. Zoning of landscapes exposed to technogenic
emissions from the Norilsk mining and smelting works. Russian Journal of Ecology 34: 375–380.
Copyright # 2007 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. 17: 305–317 (2007)
DOI: 10.1002/aqc
Без категории
Размер файла
341 Кб
environment, monitoring, radar, wetlands, satellite, imagery, arctic, sub, inland, boreal
Пожаловаться на содержимое документа