Recorded presentation4.6a
Remote Sensing for Improved Water Management—the PNWRC Water Resources Project
Dennis P. Lettenmaier, University of Washington, Seattle, WA; and M. McGuire, K. Humes, V. P. Walden, B. Harshburger, R. Qualls, W. Zhao, M. Wigmosta, R. Leung, R. Hruska, D. Garen, T. Perkins, and R. Abromavich
Mountain snowmelt contributes over eighty percent of the water supply for the Western U.S. Accurate forecasts of snowmelt runoff are important to many local, state, and Federal agencies that have interests in agriculture, hydropower, and recreation, among other water uses. Water resource managers depend on accurate water supply predictions to allocate a finite supply of water to often competing demands. Although quantitative forecasts of seasonal snowmelt runoff have been made at many locations in the West for over 50 years, these forecasts are limited by accurate knowledge of winter season precipitation and snow accumulation in remote areas, which presently are estimated via in situ networks like the NRCS’ SNOTEL. The NASA-funded Pacific Northwest Regional Colaboratory, a consortium of regional universities and government laboratories, has as its goal the use of geospatial methods and remote sensing data for improved resource management in the region. One particular aim of PNWRC is to improve water management in the Pacific Northwest through more widespread use of remote sensing data. The initial PNWRC water resources demonstration area is the Snake River basin, where a suite of hydrologic models and remote sensing products is being evaluated to determine the potential to improve on current operational streamflow forecasting methods which are based solely on in situ observations. The general strategy is to combine nowcasts of soil moisture, snow water content, crop water requirements, and other hydrologic variables derived from dynamic hydrologic models with remote sensing estimates of snow properties. The suite of models used includes the high resolution quasi-distributed Snowmelt Runoff Model (SRM), the high resolution Distributed Hydrology Soil Vegetation Model (DHSVM), and the Variable Infiltration Capacity (VIC) macroscale hydrologic model. The two high resolution models are applied to two small drainage basins with drainage areas several hundred km2 in the upper Snake River basin, while the VIC model is applied to the entire Snake River basin. An important goal for the modeling of the small drainage basins is to evaluate the trade-off between model complexity and model performance for agencies with responsibility for streamflow forecasting in basins of that size. Various downscaling techniques are used to adapt the 1/8 degree driving data previously developed from gridded in situ data for the VIC model to spatial scales appropriate to SRM and DHSVM. Long-term retrospective runs are performed with each of the models to produce model climatologies and a basis for a retrospective evaluation of seasonal streamflow forecast skill absent updating. For winters 2001-2 and 2002-3, we evaluate the impact of both replacing and augmenting our updating scheme based on SNOTEL data with corrections from course spatial resolution MODIS snow-cover data and, on a more experimental basis using the VIC model, an AMSR snow water equivalent product. In addition to seasonal streamflow forecasts based on an adaptation of the Extended Streamflow Prediction (ESP) method, we implement and evaluate experimental reservoir forecasts produced through use of the SNAKESIM Snake River reservoir management model driven with VIC model output for forecasts made in the winters of 2001-2 and 2003-4 for the following summers. In addition, real-time winter 2003-4 forecasts produced using the VIC model and associated forecasts of reservoir contents during summer 2004 are illustrated. Recorded presentation
Session 4, Applications Workshop
Wednesday, 14 January 2004, 8:30 AM-5:15 PM, Room 401
Previous paper Next paper