RAINFALL PREDICTION USING INNOVATIVE GREY MODEL WITH THE DYNAMIC INDEX

Taiwan's special climate and landforms are affected by summer typhoons, with 78% of its rainfall occurring during the summer and autumn months. The range and the severity of disasters has increased in recent years, thanks in part to climate change, which has caused an unstable rainfall. Accurate rainfall predictions help to forecast rivers' water levels. This study proposes a new rainfall prediction model based on features of the rainfall system. In order to overcome the drawbacks found in the original grey model, a few corrections have been made to the new model. First, the dynamic index transformation is used to generate an exponentially smooth sequence. When the new model was applied to data from eight different typhoons, the results revealed that the mean peak rainfall error, compared to the original sequences, is close to 0. This technique can also effectively increase the accuracy of maximum rainfall predictions. Next, the grey model's background value was improved through integration. This technique can correct any delay in the peak rainfall as predicted by the conventional model, and make the predicted and actual values closer. Finally, we used the Fourier series and the exponential smoothing technique to correct periodical and random errors. The new model is called the Dynamic index Exponential Fourier Grey Model (DEFGM (1,1)). By examining different indicators, the mean coefficient of efficiency of the DEFGM (1,1) was found to be close to 1, which is indicative of a relatively good overall performance. With this tool, the predictability of rainfall systems during typhoons is more accurate, and disaster prevention measures can be made in advance.