Ongoing research by the Climate Impacts Group at the University of Washington focuses on the use of recent advances in climate research to improve streamflow forecasts at seasonal-to-interannual, decadal, and longer time scales. Seasonal-to-interannual climate forecasting capabilities have advanced significantly in the past several years, primarily because of improvements in the understanding of, and an ability to forecast, El Niño/Southern Oscillation (ENSO) at seasonal/interannual time scales, and because of better understanding of longer time scale climate phenomena like the Pacific Decadal Oscillation (PDO). These phenomena exert strong controls on climate variability along the Pacific Coast of North America.
The streamflow forecasting techniques we have developed for Pacific Northwest (PNW) rivers are based on climate forecasts that facilitate longer lead times (as much as a year) than the methods that are traditionally used for water management (maximum forecast lead times of a few months). At interannual time scales, the simplest of these techniques involves resampling meteorological data from previous years identified to be in similar climate categories as are forecast for the coming year. These data are then used to drive a hydrology model, which produces an ensemble of streamflow forecasts that are analogous to those that result from the well-known Extended Streamflow Prediction (ESP) method. This technique is a relatively simple, but effective, way of incorporating long-lead climate information into streamflow forecasts. It faithfully captures the history of observed climate variability. Its main limitation is that the sample size of observed events for some climate categories is small because of the length of the historic record. Furthermore, it is unable to capture important aspects of global change, which may interact with shorter term variations through changes in climate phenomena like ENSO and PDO. An alternative to the resampling method is to use nested regional climate models to produce the long-lead climate forecasts. Success using this approach has been hindered to some degree by the bias that is inherent in climate models, even when downscaled using regional nested modeling approaches. Adjustment or correction for this bias is central to the use of climate model output for hydrologic forecasting purposes. Approaches for dealing with climate model bias in the context of global and meso-scale are presently an area of active research. We illustrate an experimental application of the nested climate modeling approach for the Columbia River Basin, and compare it with the simpler resampling method.
At much longer time scales, changes in Columbia River flows that might be associated with global climate change are of considerable concern in the PNW, given recent Endangered Species Act listing of certain salmonid species, and the increase in water demand that is expected to follow increases in human population in the region. Many of the same general challenges associated with the spatial downscaling of climate forecasts are present in these long-range investigations. Additional uncertainties exist in the ability of climate models to predict the effects of changing greenhouse gas concentrations. These uncertainties tend to dominate the results, and lead us to use relatively simplemethods of downscaling seasonal temperature and precipitation to interpret the implications of alternative climate scenarios on PNW water resources.
ENSO-Conditioned Weather Resampling Method for Seasonal Ensemble Streamflow Prediction