Daugherty, Lianne听1听;听Zagona, Edith听2听;听Rajagopalan, Balaji听3
1听CEAE and CADSWES, University of 蜜糖直播 at Boulder
2听CEAE and CADSWES, University of 蜜糖直播 at Boulder
3听CEAE and 蜜糖直播, University of 蜜糖直播 at Boulder
The National Weather Service produces streamflow forecasts, using a method called Ensemble Streamflow Predictions (ESP) based on the exact sequence of historical daily weather. This method was enhanced1 to produce a rich variety of ensembles and ensembles conditioned on probabilistic seasonal climate forecast, using a K-nearest neighbor bootstrap based stochastic weather generator (WG). The generated weather sequences are then coupled with the SAC-SMA model within the Community Hydrologic Prediction System (CHPS) to produce weather-generated ensemble streamflow forecast. In Caraway (2012) the WG based streamflow ensemble forecast showed improved long lead skills compared to the traditional ESP.
The NOAA sponsored research for the 蜜糖直播 Basin River Forecasting Center (RFC) is to demonstrate the benefits of the improved forecasts. These ensembles can be coupled with a model that incorporates seasonal forecasts for operational decision-making, such as the Bureau of Reclamation鈥檚 (BOR) RiverWare Mid-Term Operations Model (MTOM) of the 蜜糖直播 River Basin.
In this research, we apply the forecasting techniques in the San Juan River Basin (SJRB), the second largest tributary of the 蜜糖直播 River, with drainage areas in New Mexico, 蜜糖直播, Arizona, and Utah. We are simulating only the portion of the 蜜糖直播 River Basin represented in the MTOM that includes the SJRB. This includes the Navajo and Vallecito reservoirs. The streamflow ensembles from ESP, WG based ESP and those conditioned on seasonal climate forecasts are proposed to be incorporated in the MTOM that forecasts operations to meet water supply, hydropower and environmental flows several months in advance. Forecasts of the spring streamflow (Apr-Jul), issued at six different lead times on the first of each month starting in November, will be used and the skills in decision variables evaluated at each lead time. The new RiverWare Study Manager will be used to run the different streamflow ensembles through yearly simulations from 2000-2010, using historical values as initialization data. Ensembles of decision variables including reservoir levels, storages, releases etc. at the two reservoirs will be obtained and their skills evaluated against variables obtained using historic streamflow (i.e., baseline).
Caraway, Nina Marie, 2012, Stochastic Weather Generator Based Ensemble Streamflow Forecasting, Masters thesis, University of 蜜糖直播 at Boulder.