Evaluation of Precipitation Scaling Using Snow Cover Disappearance Date

Abstract

Snowfall is a critical component of the hydrologic cycle of mountainous regions in the Western United States as snowfall accounts for the majority of annual precipitation. Point scale precipitation measurements of snowfall are often inaccurate due to wind effects, quality control and improper gage calibration, and spatial estimates between these precipitation gauges are challenging in complex terrain. There is a need to improve spatial distributions of winter precipitation for snow, hydrology, and climate models but existing methods of precipitation estimation struggle to accurately simulate precipitation on finer spatial scales. As applied in previous studies, the snow depletion date and a snowmelt runoff model can be used jointly to provide an improved indication of winter precipitation. Peak snow water equivalent (SWE) can be reconstructed at a location by calculating the total melt from ablation energetics and subtracting the solid precipitation during the melt season. We call this SWE reconstruction technique the Snow Depletion Date Scaling (SDDS) method. The SDDS method has seen application in the Himalayas, the Swiss Alps, and the Rio Grande headwaters in the San Juan Mountains of Colorado but the method has yet to receive rigorous testing of its accuracy. Raleigh will describe a study in which he evaluated the SDDS method at 144 surface meteorological stations in Washington, Oregon, and California between water year 1996 and 2004. The SDDS method is implemented with the Snow-17 model by iteratively adjusting a snow correction factor (SCF) until the simulated snow disappearance date matches the observed disappearance date. The preliminary results using Snow-17 indicate that the SDDS method underestimates peak SWE by 17% and SWE volume by 16% on average, however, significant spread exists in the results. Lower elevation stations tended to overestimate SWE while higher elevation stations generally underestimated SWE. A sensitivity analysis indicated that these results did not change when the date of snow depletion was incorrect by ±2 days. Further work is required with the Snow-17 since the sites were not calibrated to region-specific Snow-17 state variables. Future work also includes evaluating the SDDS method with a full energy balance snow model that requires less calibration and may represent ablation energetics more accurately, thus improving the performance of the SDDS method. In the application setting, the snow depletion date can be observed by remote sensing satellites or buried temperature sensors, and the SDDS method can be implemented using simulated or measured data and a snow model.