Verification of simulated snow cover in an Arctic basin using satellite-derived snow-cover maps

Abstract

AbstractTime series of Earth observation (EO) data (Landsat Thematic Mapper (TM), U.S. National Oceanic and Atmospheric Administration Advanced Very High Resolution Radiometer (NOAA AVHRR) and European Remote-sensing Satellite synthetic-aperture radar (ERS SAR)) were obtained for a 2250 km2 mountainous basin in northern Sweden to validate snow-cover area (SCA) estimates produced by a conceptual model (HBV) during three melt seasons. SCA depletion curves derived for each image type, arid coincident images, reveal that the SCA estimate varies with the sensor. Discrepancies betweenTM and AVHRR appear to be an effect of spatial resolution. However, differences betweenTM and SAR are not simply related. Since more AVHRR thanTM data were available, a TM-equivalent SCA was derived from AVHRR by relating TM SCA to AVHRR channel 1 reflectance. The TM-equivalent SCA was used to test SCA simulated by HBV for the 1992 melt season. Although the modelled and TM-equivalent SCA were in reasonable agreement, the modelled SCA declined faster than the TM-equivalent SCA. Partial recalibration of model parameters controlling snowpack accumulation improved the match between the modelled and EO-derived SCA decline. The recalibrated parameters were verified using SCA maps generated for the 1996 and 1998 melt seasons. The adjusted parameter sets had little effect on the Nash-Sutcliffe R2 runoff fit but improved the volume fit in all three years.