Date of Award
Master of Science
Geological and Atmospheric Sciences
Kristie J. Franz
Many regions of the US rely on spring snow melt forecasts for water resource planning and flood prediction. With the potential for climate change to alter temperature and precipitation patterns during the cold season, snow information (both modeled and observed) will become increasingly important. Given limited snow and meteorological observations, the US National Weather Service streamflow forecasting system has relied on ground-based measurements of temperature and precipitation as input to a lumped, empirically-based snow model (the SNOW17) to track winter snowpack processes for decades. With the advent of satellite-based data sources and more powerful computing capabilities, the potential now exists to advance the forecasting system beyond this traditional modeling approach. Three possible areas of advancement in operational snow modeling are the use of: (1) a spatially distributed snow model, (2) direct input of satellite observations into the model, and (3) calibration of the snow model to satellite observations that are not available from ground-based monitoring sites or are unavailable at the watershed scale. The current study will investigate these three topics. We hypothesize that the application of snow data from the MODIS TERRA satellite, which provides spatially distributed hydrologic information in remote areas that are not generally monitored, will improve snow modeling for better spring streamflow predictions. Two different data applications were tested. First, the use of snow covered area data to calibrate the distributed SNOW17 model was investigated. Secondly, the application of MODIS snow albedo as input to an energy balance snow melt model was tested. The study area is the North Fork of the American River located in central California. The study period spans October 1st, 2000 through September 30th, 2009. Distributed temperature and precipitation time series are created from station data and PRISM (Daly et al., 2010) using inverse distance weighting and application of hourly precipitation trends.
Results showed that a multi-step calibration approach using both remotely sensed snow cover information and stream flow discharge produced, on average, better streamflow simulations during the spring melt period than model parameters used in the operational system. In addition, the MODIS albedo appears to underestimate the snow surface albedo leading to erroneous early melting in the mode
Logan Ray Karsten
Karsten, Logan Ray, "Investigation of MODIS snow cover products for use in streamflow prediction systems" (2011). Graduate Theses and Dissertations. 12101.