Long Term Forecasting of Streamflows

Forecasting Models Using Climatic Precursors. Long-term forecasting of hydrologic variables requires climate forecasts. As mentioned in the section on long-term precipitation forecasting, the increase in predictive skills of the models to forecast climatic anomalies based on the ENSO phenomena have provided renewed impetus for hydrologic forecasting.

There have been a significant number of contributions to hydrologic forecasting using climatic precursors. For example, the hydrologic forecasting model proposed by Liu et al. (1997, 1998) used multiple ENSO forecasts to produce forecasts of seasonal precipitation, streamflow, and other variables. This is essentially a combined data fusion and forecasting system that incorporates ENSO forecasts, persistence-based forecasts, and up-to-date observations. The system assimilates past observations, hindcasts, and projects with both multiple model outputs and persistence into its merged forecast. The error bounds on the forecast are also propagated. Liu and co-workers applied the approach to forecast droughts in Texas (Liu et al., 1997) and precipitation in the Equatorial Pacific and streamflows in Colombia (Liu et al., 1998). An example of these applications is given in Figure 4. Other applications include Berri and Flamenco (1999) who used a regression model with climatic precursors to forecast seasonal volumes in the Rio Diamante and Chiew et al. (1998) who studied the relationship between ENSO and rainfall, drought and flows in Australia, and its potential for forecasting. For example, they found that spring runoff in southeast Australia may be predicted several months in advance. Other examples are the work of Guetter and Georgakakos (1996) on the relationship between Iowa River flows and ENSO and of Kayha and Dracup (1993, 1994) on the Southwest flows.

An interesting example of the use of seasonal climate outlooks of expected air temperature and precipitation probabilities in hydrologic forecasting is found in the work of Croley and colleagues at the NOAA GREL (Croley, 1996, 1997; Croley and Kunkel, 1996). In their work historical meteorology record segments are used with hydrological and other models to simulate hydrological scenarios. The historical meteorological records are weighted to be compatible with NOAA's climate outlooks. An example of the use of the procedure is shown in Figure 5. Hamlet and Lettenmaier (1999) proposed a method to incorporate both ENSO and PDO signals in the long-term forecasting of streamflows in the Columbia River. The climate forecasts, classified in six categories as a function of ENSO and PDO, are used to generate climatic scenarios that serve as input to a hydrologic model.

0 0

Post a comment