Advisories
Stream flow at a specific site such as a reservoir, hydroelectric plant or water treatment plant intake is often the variable of direct interest to managers. Such on-site stream flow predictions may be derived
from precipitation fields that are predicted by Global Climate Model's (GCM) after suitable downscaling and hydrological modeling. Where historical stream flow records are available, one may circumvent these steps by using the historical relationship between the global sea surface temperatures (SST) and stream flow. This approach works best when stream flow is aggregated into seasons during which similar climatic conditions prevail over the river basin. For each of these seasons, a sensitivity analysis of SST’s and stream flow data can be conducted to identify the predictors of significance. Thereafter, multivariate techniques may be used to develop a prediction scheme that generates forecasts at the lead–time.
These scatter plots point to a significant correlation between the stream flow at Peradeniya and SST indices in previous months. These relationships were exploited to predict the seasonal stream flow from 1994 to 1999 below based on SST.