Snow Cover Quantification in the Central Andes Derived from Multi-Sensor Data

Thursday, 18 December 2014
Edward Cornwell1, Gonzalo Cortés2, James P McPhee1 and Steven A Margulis3, (1)University of Chile, Santiago, Chile, (2)University of California Los Angeles, Los Angeles, CA, United States, (3)UCLA, Los Angeles, AP, United States
In this work we analyze time series of MODIS MOD10A1 and Landsat fractional snow cover images over the central-southern Andes (30°-34°S) with the objective of quantifying snow cover persistence over the region. Data from 2000-2011 is analyzed, and a snow cover persistence index (SCPI) is computed for each pixel at different resolutions over four test watersheds. The annual SCPI is calculated as the number of acquisitions during a year on which a pixel is covered by a fraction of snow greater than a defined threshold. We calculate the SCPI at the nominal Landsat resolution (30m) and at the MODIS resolution (500m) for each of the pixels of the different watersheds and for different thresholds. Furthermore, for each watershed, the correlation of SCPI with slope, aspect and elevation is calculated for Landsat at 30m, and aggregated at 100m and 500m, and for MODIS at 500m. Both MODIS and Landsat show significant differences in annual SCPI for the 500m resolution. The difference in correlation between SCP and physiographic variables is largest when comparing Landsat 30m to MODIS, suggesting that sensor resolution is an important limiting factor for the analyzed region. Trend analysis using both products show differences in terms of changes in SCPI for the 2000-2011 period, however the conclusions are not robust due to the short period analyzed.