A Forecast Model of Hydrologic Single Element Medium and Long-Period Based on Rough Set Theory

On the basis of rough set theory, this paper presents the single element medium- and long-term classification forecast model that uses historical data of a hydrologic series as forecast factors. The minimal rule set, i.e., forecast pattern set, is achieved according to the principle of maximal attribute significance and rules frequency. Maximal support strength is put forward and applied to predict by using the model. The model is applied to forecast annual runoff of Dahuofang reservoir. The result indicates that the forecast model based on rough set can describe the relationship between forecast factors and forecast object efficiently and accurately. This model, which is composed of simple solution rules, can be easily understood and applied. Copyright Kluwer Academic Publishers 2004