550 A Comparison of Weather and Research Forecast Model Precipitation Forecasts versus Observed SNOTEL Precipitation

Thursday, 14 January 2016
Vincent Patrick Holbrook, Idaho Power Company, Boise, ID; and D. Blestrud and M. L. Kunkel

Handout (1.2 MB)

The Idaho Power Company (IPC) is a hydroelectric based utility serving customers throughout southern Idaho and eastern Oregon. Winter snowpack is critical to IPC operations and the company uses a commercially developed River Forecast System (RFS) to predict flows along the Snake River and its tributaries. Observed precipitation across the region is needed to initialize the model. SNOw TELemetry (SNOTEL) sites spread across a 5 state region from Oregon to Wyoming are a major component of the observation network. A second input to the RFS is seven day precipitation forecasts obtained from a 1.8 kilometer resolution Weather and Research Forecast (WRF) model. The combination of observed precipitation, forecast precipitation, flows in the river, reservoir storage, and other variables is combined by the RFS to forecast river flows. Optimized hydroelectric operations are a result of this process and benefit IPC in the form reliability and effective use of its most cost effective generation resource and IPC customers benefit from low cost electricity. The goal of this research was to determine the accuracy of WRF precipitation forecasts for the SNOTEL sites in order to adjust model forecast points to provide the RFS the best precipitation forecast possible. This study directly compares observed SNOTEL precipitation for the Nov 1 through Mar 31 time frame during 4 years between 2011 and 2015 versus cumulative WRF model 24 hour precipitation forecasts (midnight to midnight). The Nov 1 through Mar 31 time period is when the majority of accumulated Snow Water Equivalent (SWE) occurs with peak SWE at SNOTEL sites generally around the end of March. For the 5 state region composing this study, the average precipitation forecast by the WRF was within 5% of the average precipitation observed at the SNOTEL sites. Model error was not consistent either by year or SNOTEL location. The absolute value of the error at individual SNOTEL sites was over 50% at some locations and varied by year. A discussion of these results will be given. In addition, a comparison of high resolution precipitation gauge data collocated at two SNOTEL sites will highlight under catch issues of SNOTELs.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner