Long-range streamflow forecasting using ENSO information: Application to the Columbia River Basin
Long-range forecasting of streamflow is important to hydrologists and water resource planners for the maintenance of a diverse ecosystem and for optimal operation of water resource systems. In the Western U.S., there is a significant lag relationship between streamflow and the El Nino-Southern Oscillation (ENSO). The Southern Oscillation Index (SOI) and the Wright Sea Surface Temperatures (SSTw) are used here to study the lag relationship between Western U.S. streamflow and ENSO indices. A long-range seasonal streamflow forecasting model is developed and tested on the Columbia River Basin, Washington State. A probabilistic streamflow forecast is made from an optimal linear combination of SOI Phase, SSTw Phase, and SOI Linear Discriminant Analysis (LDA), and SSTw LDA forecasts (See Figure 1). The forecast takes the form of a probability of the streamflow occurring in specified categories (i.e., below normal, normal, and above normal). The optimal linear combination of the four models is the forecast that minimizes the mean square error (Half-Brier score) and is the forecast with the best overall skill. This final combination forecast is referred to as a "consensus forecast." This consensus model is used on seven Columbia River basin streamflow stations that cover the period 1933-1992. An additional station, Columbia River at The Dalles, covers a period from 1911 to 1992. It was found that three to seven month lead time predictor periods of spring-summer runoff, based on SOI and SSTw data, had better skill than climatology. A plot of the observed streamflow (in percentiles) versus the forecast probabilities of below normal streamflow is shown in Figure 2. For the streamflow station, Columbia River at the Dalles, 20 points correspond to a streamflow forecast where the probability is high (> 40%, greater than climatology) for the below normal category. Of these 20 points, 12 correspond to observed streamflow that is below normal (i.e., successful forecast). This work demonstrates the potential of using ENSO indicators for long-range streamflow forecasting in the Columbia River basin.
Climate | Environmental Sciences | Fresh Water Studies
Use Find in Your Library, contact the author, or use interlibrary loan to garner a copy of the article. Publisher copyright policy allows author to archive post-print (author’s final manuscript). When post-print is available or publisher policy changes, the article will be deposited
Piechota, T. C.,
Dracup, J. A.
Long-range streamflow forecasting using ENSO information: Application to the Columbia River Basin.