547 The Evaluation and Verification of a Distributed Snow Model with Time, Space, and Elevation Variables

Thursday, 14 January 2016
Jongkwan Kim, NOAA, TUSCALOOSA, AL; and M. Smith

Handout (4.6 MB)

In the mountainous western U.S., snow accumulation and melt are critical components of the water cycle. In this study, we compare simulated and observed snow variables using traditional and new spatial similarity measures. We use the SNOW17 model within the Hydrology Laboratory Research Distributed Hydrologic Model (HL-RDHM) developed by NOAA-National Weather Service (NWS) to simulate snow cover and snow water equivalent over the Durango River Basin in Colorado. The study basin is in the mountainous western U.S. area and consists of 108 HRAP grid cells. Simulated snow information is produced on 4km HRAP grids with the temporal resolution of 6 hours for a 5-year period (Water Years 2001-1005) using a priori parameters provided by NOAA-NWS. The simulated snow information is compared to various types of snow observations such as in-situ and satellite remotely sensed information. In-situ snow observations include snow water equivalent data from 3 Snow Telemetry (SNOTEL) stations. Satellite observations of snow archived at the National Snow and Ice Data Center (NSIDC) are used as remotely sensed snow observations. The main purpose of this research is to compare the spatially distributed snow simulations with various observations as considering time, space, and elevation variables. For the consideration of spatial patterns, both traditional measurements and similarity functions are employed to calculate error gaps between snow computations and in-situ or remotely sensed observations. We use the Hausdorff Distance (HAUS) and Earth Mover's Distance (EMD) for the analysis of spatial patterns. HAUS allows considers various factors such as time, location, and elevation. EMD measures the volume that must be displaced to make one gridded field equal to another. This research shows the usefulness and availability of HAUS and EMD similarity error functions for evaluating spatially distributed snow models.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner