提升數值天氣預報模式之預報力的方式，一直都在精進，其中一種改善方式為提供更合理的初始條件及邊界條件，而若使用星載之高光譜紅外線探空儀及微波探空儀，除了能夠提供三維大氣溫濕度狀態，亦可補足傳統觀測在時間及空間分布之不足。於實務上進行資料同化時，又可區分為使用輻射強度(radiances)資料與反演產品(retrieved products)，使用前者可兼顧即時性之需求，而使用後者與傳統觀測性質相近，原理也較為直觀。 過往研究曾指出，衛星遙測資料不論是輻射強度或反演產品，需考慮可能存在的系統性偏差問題。本篇研究使用美國「聯合繞極軌道衛星系統(JPSS)」中Suomi-NPP衛星上的先進技術微波探空儀(ATMS)與高光譜紅外線探空儀(CrIS)觀測資料，也將其反演產品藉由NCEP再分析資料(FNL)進行大氣熱力參數品質控制與偏差修正，結果可與歐洲中期預報中心(ECMWF)再分析資料比較後，可得到較低系統性偏差的三維大氣溫溼度剖線。 為進一步探討反演產品與輻射強度之差異及修正成效，研究中使用WRF Model及Gridpoint Statistical Interpolation(GSI)資料同化方法，探討尼伯特颱風(2016)之環境、路徑、強度變化與定量降水預報。結果顯示如能透過資料同化技術使用經誤差修正後的溫溼度反演產品於多個實驗組中，將能有較顯著的颱風預報結果，並且提高定量降水預報(QPFs)之技術得分。;Improving numerical weather prediction (NWP) model are discussed uninterruptedly. One of ways to improve the performances of NWP model is providing more reasonable initial conditions and/or boundary conditions. The hyperspetrum infrared sounder and microwave sounder make up conventional observations on special and temporal distribution with three dimensional observations of atmospheric temperature and moisture. Both of radiances and retrieved products can be assimilated into NWP. The radiances usually used in operational center with its immediacy. The use of retrieved products more simple and similar to conventional observations. Some previous works indicate that systematic bias is found in satellite observed radiance/retrieved products. In this study, we try to reduce the uncertainty of retrieved products from Advanced Technology Microwave Sounder (ATMS) and Cross-track Infrared Sounder (CrIS) onboard NOAA/JPSS Suomi-NPP. The bias correction mothed of moisture by refer to FNL data can makes bias close to zero. We assimilate radiance and retrieved products in NWP during the period of Typhoon Nepartak (2016). The Weather Research and Forecasting (WRF) Model and Gridpoint Statistical Interpolation (GSI) system are adapted to investigate typhoon track, intensity, environmental fields and Quantitative Precipitation Forecasts (QPFs). The results show it has better forecasts and the skill scores of QPFs after implementing the bias correction.
