How Does Knowing Snowpack Distribution Help Model Calibration and Reservoir Management?

Abstract:

Well calibrated hydrologic models are a necessary tool for reservoir managers to meet increasingly complicated regulatory, environmental and consumptive demands on water supply systems. Achieving these objectives is difficult during periods of drought, such as seen in the Sierra Nevada in recent years. This emphasizes the importance of accurate watershed modeling and forecasting of runoff. While basin discharge has traditionally been the main criteria for model calibration, many studies have shown it to be a poor control on model calibration where correct understanding of the subbasin hydrologic processes are required. Additional data sources such as snowpack accumulation and melt are often required to create a reliable model calibration.

When allocating resources for monitoring snowpack conditions, water system managers often must choose between monitoring point locations at high temporal resolution (i.e. real time weather and snow monitoring stations) and large spatial surveys (i.e. remote sensing). NASA’s Airborne Snow Observatory (ASO) provides a unique opportunity to test the relative value of spatially dense, temporally sparse measurements vs. temporally dense, spatially sparse measurements for hydrologic model calibration.

The ASO is a demonstration mission using coupled LiDAR and imaging spectrometer mounted to an aircraft flying at 6100 m to collect high spatial density measurements of snow water content and albedo over the 1189 km² Tuolumne River Basin. Snow depth and albedo were collected weekly throughout the snowmelt runoff period at 5 m² resolution during the 2013-2014 snowmelt.

We developed an implementation of the USGS Precipitation Runoff Modeling System (PRMS) for the Tuolumne River above Hetch Hetchy Reservoir, the primary water source for San Francisco. The modeled snow accumulation and ablation was calibrated in 2 models using either 2 years of weekly measurements of distributed snow water equivalent from the ASO, or 2 years of 15 minute snow depth and snow water equivalent from the 2 weather stations within the basin. The models were then validated against 90 years of streamflow. This work demonstrates the relative value of spatially versus temporally dense snow measurements in model calibration, and the possible management benefits for collecting each type of data.