Characterization of Freeze-Thaw Process Heterogeneity with Spaceborne Radar
by Erika V. Podest and Kyle C. McDonald, Jet Propulsion Laboratory
The landmass of the boreal high latitudes encompasses almost 30 percent of the global land area. Within these regions, a major portion of the Earth’s carbon is stored in vegetation and in seasonally frozen and permafrost soils. The annual freeze/thaw cycle drives the length of the growing season in these landscapes and is a major factor determining annual productivity and associated exchange of CO2 with the atmosphere. Accurate characterization of these processes can improve regional assessment of vegetation productivity and seasonal carbon dynamics. Satellite microwave remote sensing is sensitive to landscape freeze/thaw state and is an effective tool for this purpose since large inaccessible areas can be monitored on a temporal basis without limitations imposed by weather or daylight conditions. We have applied time series spaceborne Synthetic Aperture Radar (SAR) imagery from ESA’s ERS-1/2 (C-band, 5.4 GHz, VV pol.), and JAXA’s JERS-1 (L-band, 1.3 GHz, HH pol.) to assess the freeze/thaw state of the landscape and estimate growing season length.
Variations in seasonal freeze/thaw processes can be spatially and temporally complex. Landscape complexity can affect the timing of seasonal freeze/thaw transitions as a result of local-scale variations in land cover, snow cover, or topography, including elevation and slope aspect. We examined spatial and temporal characteristics of the seasonal freeze/thaw transitions over study areas in the Bonanza Creek and the Kenai Peninsula regions (see accompanying figure). The study areas encompass a variety of landcover classes and regions of low relief and moderate to complex topography within two different climatic zones, boreal continental and boreal maritime. The Bonanza Creek study area included two separate sites, a floodplain wetland site and a moderate topography site. The Kenai Peninsula study area encompasses a region of complex topography. Ancillary validation data, including data from long-term ecological study and other monitoring stations, regional land cover classifications and digital elevation model (DEM) information, were available for each site. JERS-1 SAR data employed were drawn from the data holdings generated as part of the North American component of the Global Boreal Forest Mapping project. The SAR data were radiometrically corrected and, where needed, terrain corrected using ASF’s radiometric and terrain correction software. Imagery were processed to 100 meter spatial resolution.
Results show that both C-band ERS-1/2 and L-band JERS-1 backscatter enabled characterization of freeze/thaw transition heterogeneity. However, the presence and status of snow cover strongly influence detection and monitoring of freeze/thaw status and can be a confounding factor in determining vegetation thaw status.
A change detection algorithm that examines the time series progression of radar backscatter relative to seasonal reference states was developed. The freeze/thaw classification algorithm was applied to the multi-temporal SAR data to develop state maps of each study area. Land cover composition, land cover classifications, and DEM information were merged with the SAR maps to examine spatial variability in freeze/thaw state as related to microclimate variations that arise from slope aspect, elevation, and land cover.
At Bonanza Creek, wetland vegetation in the floodplain study area thawed later and froze later than other vegetation classes in the domain. At the moderate topography study area, coniferous vegetation was seen to thaw earlier in springtime than the other vegetation classes. Slope aspect also gave rise to freeze/thaw transition complexity, with south facing slopes thawing earlier and freezing later than north facing slopes. At the Kenai Peninsula study area, snow melt dominated the backscatter seasonal response of both SARs. The progression of the associated seasonal thaw and freeze was clearly visible in the form of a “thaw wave” progressing from lower to higher elevations.
Although current SARs are capable of characterizing the spatial complexity associated with this important biophysical variable, they lack the temporal revisit capability necessary to accurately constrain the seasonality of the associated processes. Current lower resolution satellite microwave sensors (e.g., QuikSCAT, SSM/I) have the potential to resolve the timing of these seasonal transitions at the pan-Arctic scale, however, their ability to accurately resolve the finer (
Click here to download a copy of the newsletter featuring this article