8.7 An alternative approach to predicting snowfall across the Sierra Nevada

Tuesday, 31 August 2010: 5:15 PM
Alpine Ballroom A (Resort at Squaw Creek)
Mark S. Raleigh, University of Washington, Seattle, WA; and J. D. Lundquist

In the Sierra Nevada the majority of annual precipitation falls during the winter months, with up to 67% of annual precipitation arriving as snowfall. Snowpack provides a critical input into California's water supply during the dry summer months. One tool that is currently employed by scientists to estimate snowfall in the Sierra Nevada is the PRISM model, which estimates precipitation between observation stations through geostatistical techniques, empirical relationships, and climatological records. However, the spatial distribution of snowfall in the Sierra is not always well predicted by PRISM in years that deviate from mean climatological values. For example, during water year 2002 PRISM overestimated mean annual precipitation across observations in the northern Sierra Nevada, and in many cases these accumulations were overestimated by a factor of two or three. Conversely, during water year 2007, PRISM underestimated annual precipitation at most observation stations in this region, with errors as high as 50%. Currently, deviations from monthly and annual precipitation climatology can occur due to barrier jets and fluctuations in ENSO. With the anticipated impacts from a warming climate, more frequent deviations from mean climatological snowfall accumulations are possible, and thus a more robust approach is needed to estimate snowfall accumulation in the Sierra Nevada.

An alternative method that simulates snowfall accumulation is the reconstruction of snow water equivalent (SWE), which is the amount of water stored in a snowpack. The peak accumulation of SWE can be reconstructed by summing estimated snowmelt prior to the date of snowpack disappearance, which can be observed by various methods, such as satellite-borne remote sensing instruments. Spatial distributions of reconstructed peak SWE can then be related to one or more weather stations to scale daily snowfall data prior to peak SWE. The SWE reconstruction approach is advantageous over PRISM because it does not modify snowfall estimations based on climatology. While the SWE reconstruction method can be only applied once the snowpack has melted out, knowledge about snowfall distributions during past winters are still needed by scientists to evaluate meteorological forecasts, to analyze the impact of snowpack on forests and ecology, and to monitor annual changes in mountain snowpack due to a warming climate.

While some studies have employed SWE reconstruction, few have attempted validation of the reconstruction method and none have compared the reconstruction method to other precipitation estimation methods, such as PRISM. In this study, the SWE reconstruction method will be tested at control sites and then used to derive spatial distributions of snowfall. Spatial estimates of seasonal snowfall generated from PRISM, the SWE reconstruction method, and observations will be compared in the northern Sierra Nevada from water years 2002 to 2009. Particular focus will be centered on years that deviate from climatology, such as water years 2002 and 2007. This research will determine where and when estimated snowfall from PRISM and the SWE reconstruction method do not match, and will propose explanations for these differences.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner