Multispectral Remote Sensing Technologies Applied to Assess Recent Aeolian Activity and Thaw Lake Changes in Kobuk River Valley, Alaska

The Kobuk River Valley, within the continuous permafrost zone at 67° north latitude in Kobuk Valley National Park, northwestern Alaska, is an aeolian-influenced fluvial system with boreal forest, thaw lakes, polygonal ground, loess deposits, and active and stabilized cold-climate dune fields. We used multispectral remote sensing data to study recent geomorphic and hydrogeologic processes acting on this landscape, and developed a new methodology for remote monitoring of active morphological processes using precise orthorectification, coregistration, and subpixel correlation. Ubiquitous thaw lakes (scales range from 0.2 to 1.5 km) in the valley at 15 to 50 m amsl are evidence of longstanding, local permafrost degradation. As permafrost thaws, land subsidence occurs, developing topographic lows in which thaw lakes may form. While further degradation at lake margins may initially increase lake size, continued permafrost degradation may lead to increased net infiltration and reduction or total depletion of lake surface area. Unsupervised classification performed on Landsat TM, ASTER, and AVNIR-2 imagery quantified surface water (i.e., thaw lake) areal changes within two distinct land cover areas (wetlands and Quaternary noncarbonate deposits). The estimated 14 and 20 percent decrease in surface water area from 1985 to 2008 is attributed, in part, to climate warming effects on permafrost stability. For related results attained using ALOS polarimetry data, see Hooper et al. (this meeting). Using our new multispectral data displacement analysis method, based on the Coregistration of Optically Sensed Images and Correlation (COSI-Corr) technique, we retrieved the subpixel displacements between ASTER and SPOT 5 images to estimate the migration rates and directions for the 62 km2, dominantly transverse and barchanoid Great Kobuk Sand Dunes over a recent 5-year period. This method corrects offsets due to sensor distortions, attitude drifts, and orbital errors, enabling data coregistration with a 1/50 pixel accuracy. We constructed a 15-m-resolution digital elevation model from ASTER stereo data for use in orthorectification. Our calculations suggest the mean migration rate for stabilized portions of the dune field is 0.2±0.16 m/yr, and the most likely migration rate for active portions of the dune field is within the range of 0.5-1.5 m/yr. Traditional image processing revealed subtle spatial variations in sand dune reflectance that may correlate to moisture content, and thus to dune activity. The presumed presence of permafrost beneath the dune system coupled with documented ephemeral niveo-aeolian deposits, a bimodal wind regime, and the Waring Mountains topographic barrier likely leads to the low migration rate of these cold-climate dunes. Further permafrost degradation due to climate warming may lead to a drier active layer with impacts on local vegetation and increased aeolian activity in the Kobuk River Valley.