Study on Landscape Freeze/Thaw Classification and its Spatial Scale Effects using Satellite L-band radar observations over Alaska
Abstract
Spatial and temporal variability in landscape freeze-thaw (FT) status at higher latitudes and elevations significantly impacts land surface water mobility and surface energy partitioning, with major consequences for regional climate, hydrological, ecological and biogeochemical processes. With the development of new generation space-borne remote sensing instruments, future L-band missions including the NASA Soil Moisture Active and Passive (SMAP) mission will provide new operational retrievals of landscape FT state dynamics at relatively fine (3 km) spatial resolution. We applied theoretical simulations of L-band radar backscatter using first-order radiative transfer models with two-layer and three-layer modeling schemes to develop a modified seasonal threshold algorithm (STA) and FT classification over Alaska using finer scale (100 m resolution) satellite Phased Array L-band Synthetic Aperture Radar (PALSAR) observations. An Alaska FT map for April, 2007 was generated from PALSAR observations and showed a regionally consistent, but finer FT spatial pattern than an alternative surface air temperature based classification derived from global reanalysis data. Validation of the STA based FT classification against regional soil climate stations indicated approximately 80% and 70% spatial classification accuracy in relation to respective in situ station air temperature and soil temperature measurement based FT estimates. The STA FT classification method is found to be reliable for most of the major Alaska land cover types except for barren land. An investigation of relative spatial scale effects on FT classification accuracy indicates that the relationship between pixel size and relative FT spatial classification error follows a general logarithmic function. The optimum resolution for accurate FT classification is expected to depend on the landscape FT spatial heterogeneity. However, our results indicate that the regional FT spatial scaling error is less than 12.8% and 14.6% at respective 3km and 10km pixel scales relative to the baseline (100m) fine scale FT classification results.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFM.C21D0687D
- Keywords:
-
- 0758 CRYOSPHERE Remote sensing;
- 0718 CRYOSPHERE Tundra;
- 0702 CRYOSPHERE Permafrost