初始場的水氣資訊及其分布對於對流的形成與發展至關重要，而紅外線探測儀則能提供大氣垂直方面的資訊。已有研究顯示，同化AHI輻射資料與三層可降水量（LPW）產品能有效改善颱風路徑和強降雨事件的預報。然而，雲層的複雜性會顯著影響輻射資料的傳輸，因此在進行資料同化之前，必須利用雲遮（cloud mask）產品來篩選觀測資料，以進一步提升同化效果，從而提高天氣預報的準確性。由於午後熱對流的尺度較小且變化迅速，過去的預報研究多集中於長時間尺度與大範圍的天氣系統。本研究採用WRF模式和GSI同化系統，針對台灣典型的午後對流個案，評估同化AHI觀測資料對短時強降雨預報的影響。研究結果顯示，同化晴空區域的亮溫資料能顯著改善環境場中的水氣與溫度分布，並提高降水預報的準確度。在各實驗組中，經過雲遮處理後的資料同化效果最為顯著。表示雲遮技術在提升觀測資料品質及資料同化中扮演重要的角色。;Moisture information and its distribution in the initial conditions are crucial for developing convection, with infrared sounders providing valuable vertical atmospheric data. Previous studies have shown that assimilating AHI radiance data and three-layer precipitable water (LPW) products can effectively improve the forecasts of typhoon tracks and heavy rainfall events. However, the complexity of clouds significantly impacts radiance data transmission, making cloud mask products essential for filtering observational data before assimilation to enhance the process and improve forecast accuracy. Afternoon convection is characterized by its small scale and rapid evolution, yet past forecasting research has often focused on larger-scale weather systems over longer time scales. This study employs the WRF model and GSI assimilation system to evaluate the impact of assimilating AHI observational data on short-duration heavy rainfall forecasts for a typical afternoon convection case in Taiwan. The results demonstrate that assimilating brightness temperature data from cloud-free areas significantly improves the moisture and temperature distributions in the environmental fields, leading to better precipitation forecast accuracy. Data assimilation using cloud-masked observations showed the most substantial improvements, highlighting the critical role of cloud masking in enhancing data quality and improving the prediction of heavy rainfall events.
