Search

Summary MODIS snow data are appropriate for a wide range of eco-hydrological studies and applications in the fields of snow-related hazards, early warning systems and water resources management. However, the high spatio-temporal resolution of the remotely sensed data is often biased by snow misclassifications, and cloud cover frequently limits the availability of the MODIS-based snow cover information. In this study, we applied a four-step methodology that aims to optimize the accuracy of MODIS snow data. To reduce the cloud fraction, 8-day MODIS data from both the Aqua and Terra satellites were combined. Neighborhood analysis was applied as well for this purpose, and it also contributed to the retrieval of some omitted snow. Two meteorological filters were then applied to combine information from station-based measurements of minimum ground temperature, precipitation and air temperature. This procedure helped to reduce the overestimation of snow cover. To test this technique, the methodology was applied to the Rhineland-Palatinate region in southwestern Germany (approximately 20,000 km 2 ), where cloud cover is especially high during winter and surface heterogeneity is complex. The results show that mean annual cloud coverage (reference period 2002–2013) of the 8-day MODIS snow maps could be reduced using this methodology from approximately 14% to 4.5%. During the snow season, obstruction by clouds could be reduced by even a higher degree, but still remains at about 11%. Further, the overall snow overestimation declined from 11.0–11.9% (using the original Aqua–Terra data) to 1.0–1.5%. The method is able to improve the overall accuracy of the 8-day MODIS snow product from originally 78% to 89% and even to 93% during cloud free periods. [ABSTRACT FROM AUTHOR]