受氣候變遷之影響，特異（極端）天氣現象發生的頻率及強度明顯的提高， 過去事件（如莫拉克颱風）引發之強降雨完全超乎數值預報、統計預報及人工預 報之能力範圍，氣象界傾全力提供各種方發提高類似極端事件預報之準確度。本 研究利利用模糊邏輯之類神經網路(Aritifical Neural Network)概念，輸入與降雨 有關（如地形、雷達、衛星、模式預報產品、氣候統計預報等等）之因子以推估 高密度網格（如1km*1km)點之降雨量圖（rain map），可以用來推估都市區域、 山地地區、或集水區之降雨量）。 Owing to the impact of climate change, the occurrence frequency and intensity of extreme weather events has been greatly increased. For example, like typhoon Morakot 2009, the rainfall amount was far beyond the predictability of numerical more and man forecast. The operation units are taking the full power to improve the forecast ability upon these extreme events. In the study, proving the AWOS station rainfall as the truth, a neural network technique is used to estimate the rainfall amount using the radar data, satellite data, and model output as the input. After training, a stable network is obtained which can create a rain-map at the horizontal resolution of 1km x 1km. The total rainfall amount over the urban or watershed area can be obtained.
