A comparison of modeled, remotely sensed and measured snow water equivalent in the northern Great Plains

Flood forecasting in the northern Great Plains is highly dependent on quantity and melt rate of the seasonal snow cover. This paper applies common snow water equivalent (SWE) algorithms using the Special Sensor Microwave/Imager (SSM/I) instrument across the northern Plains during the early 1990s. Various filtering techniques are applied to eliminate those situations in which the microwave SWE algorithms are known to be ineffectual. The microwave SWE estimates are compared to first-order SWE observations and the SWE output of an energy and mass balance model of the snowpack (SNTHERM). It is well know that microwave SWE algorithms underestimate SWE during periods of snowmelt. Current methods for determining snowmelt using the SSM/I appear to be insufficient for this application. After significant melt-freeze metamorphism, it appears the microwave SWE algorithms overestimate SWE by an order of magnitude. Methods for filtering these events or modifying existing SWE algorithms are examined. The comparisons are made at several first order stations across the northern Plains. The importance of correcting for under cache of accumulated snowfall is demonstrated in attempting to properly model SWE.