Weighted-Climate Parametric Hydrologic Forecasting

Abstract

This paper briefly summarizes an existing nonparametric method for using meteorology probability forecasts in operational hydrology and extends it for parametric estimation. The methodology builds a sample of possibilities for the future, of climate series from the historical record, which is weighted to agree with selected forecasts of meteorology probabilities. The nonparametric method concentrates on isolated event probabilities rather than on the entire probability distribution for various variables. It sometimes assigns the same weight to all climate series in selected categories, resulting in the same relative frequency for those climate series. By changing to a parametric approach, one determines entire probability distributions that match available forecast meteorology probabilities. This allows a continuous distribution of probability across a variable, not always possible with the nonparametric approach. This paper illustrates both the nonparametric and the parametric methods with an example, comments on both approaches, and evaluates both in a selected comparison.