An improved methodology to map snow cover by means of Landsat and MODIS imagery

In this article we propose a methodology to determine snow cover by means of Landsat-7 ETM+ and Landsat-5 TM images, as well as an improvement in daily Snow Cover TERRA-MODIS product (MOD10A1), between 2002 and 2005. Both methodologies are based on a NDSI threshold > 0.4. In the Landsat case, and although this threshold also selects water bodies, we have obtained optimal results using a mask of water bodies and generating a pre-boundary snow mask around the snow cover. Moreover, an important improvement in snow cover mapping in shadow cast areas by means of a hybrid classification has been obtained. Using these results as ground truth we have verified MODIS Snow Cover product using coincident dates. In the MODIS product, we have noted important commission errors in water bodies, forest covers and orographic shades because of the NDVI-NDSI filter applied to this product. In order to improve MODIS snow cover determination using MODIS images, we propose a hybrid methodology based on experience with Landsat images, which provide greater spatial resolution.