Factor Analysis of Multitemporal Lake Mapping on the Alaskan North Slope

Mapping land surface changes over time requires analysis of critical factors that obscure the true change. We present an analysis of the factors that influence multitemporal lake mapping in the Arctic Coastal Plain (ACP) of Northern Alaska. The ACP has many thousands of dynamic lakes which can be repeatedly mapped by relatively straight forward remote sensing techniques. However, the comparability of maps from different dates depend on factors that are lake-specific, sensor-specific, or processing-specific. Factors specific to each lake include lake size, shape, water level, and topography/bathymetry, while sensor dependent factors include resolution, spectral bands, and radiometric sensitivity. Processing specific factors include resampling and orthorectification. We found that water level variations together with near shore topography/bathymetry directly lead to lake shoreline variability seen from satellite imagery. Shallow near shore waters are underestimated in area when shorter wavelength spectral bands are used due to their deeper water penetration. In order to avoid erroneous change detection, similar spectral bands should be used across different sensors for consistency, and image rectification should be performed in the vector domain after lake identification.