Accurate simulation of early cold-season soil freezing requires accurate characterization of landscape-scale snowcover conditions, which was addressed in this study by gapfilling the MODIS SCE record to mitigate data loss from pervasive cloud cover and other factors. The gap-filled MODISSCE products were then combined with other ancillary datato downscale the MERRA-2 reanalysis snow depth data,as one of the main driver data sets for the permafrost soilmodel. The accuracy of the gap-filled MODIS SCE productwas cross-checked using the two MODIS sensors (Terra andAqua); the downscaled snow depth data were evaluated using in situ SNOTEL observations across Alaska. The modelsimulated soil-freezing process and ALT dynamics were conducted over a smaller Arctic Alaska domain and evaluatedusing a diverse set of regional observations.