Remotely Sensed Active Layer Thickness (ReSALT) from InSAR data near Toolik Lake in Northern Alaska

Friday, 19 December 2014: 2:10 PM
Albert C Chen, Stanford University, Geophysics, Los Altos Hills, CA, United States, Lin Liu, Chinese University of Hong Kong, Earth System Science, Hong Kong, Hong Kong, Kevin M Schaefer, University of Colorado, National Snow and Ice Data Center, Boulder, CO, United States, Andrew Parsekian, University of Wyoming, Laramie, WY, United States, Elchin E Jafarov, National Snow and Ice Data Center, Boulder, CO, United States, Howard A Zebker, Stanford University, Stanford, CA, United States and Tingjun Zhang, University of Colorado, Boulder, CO, United States
Toolik Field Station is built on spatially continuous permafrost on the north slope of Alaska. Seasonal surface subsidence and uplift occurs in permafrost regions due to thaw settlement and frost heave as the active layer thaws and refreezes.

Using L-band (23.6 cm wavelength) InSAR data from ALOS-PALSAR acquired between 2006 and 2010, we use a small-baseline subset (SBAS) method to estimate seasonal surface subsidence and retrieve fine-resolution maps of active layer thickness (ALT) for a ~25x25 km area surrounding Toolik Field Station (located at 68.63°N, -149.60°E). We compare these remotely sensed ALT (ReSALT) results with in situ data from: 1) the Circumpolar Active Layer Monitoring (CALM) network showing mean ALT of ~40-50 cm in the region surrounding Toolik Field Station, corresponding to seasonal subsidence of 1 to 2 cm, and 2) mechanical probing measurements of ALT, obtained during field work in the study area in August 2014.

We also solve for secular subsidence trends from the InSAR data. The trends are close to zero in most places, but larger subsidence trends in some isolated areas could be due to thermokarst processes (long-term thawing of ice-rich permafrost). We note, however, that downslope motion due to gelifluction cannot be separated from vertical thermokarst-related deformation without incorporating InSAR measurements from multiple look angles.

Two key limitations to our method are the spatial variability of volumetric soil moisture content and the accuracy of the DEM needed to correct for topographic effects. We investigate the use of bulk volumetric water content inferred from ground-penetrating radar (GPR) data to improve the ReSALT retrieval algorithm. We also quantify the effect of DEM accuracy on ReSALT uncertainties, leads to requirements for DEM accuracy in InSAR-based ALT retrieval.