在OSSE架構下利用Gal-Chen(1978)提出的熱力反演方法，將雷達觀測風場帶入動量方程、解一個與壓力有關的包桑方程，得到標準化氣壓擾動量和虛雲位溫擾動量的水平平均及其水平平均偏差量(π^'-〈π^' 〉、θ_c^'-〈θ_c^' 〉)，經由公式轉換推得氣壓和溫度的擾動量在水平平均及其水平平均偏差量(P^'-〈P^' 〉、T^'-〈T^' 〉)。引入海洋上在雷達觀測區域內含有壓力、溫度垂直剖線的GPS RO探空資料，可得知一個三維的溫度、壓力分佈，再經由水汽調整後同化到模式中。本文根據同化變數多寡和同化雷達個數、區域範圍實驗設計探討對預報結果的影響。同化變數多寡實驗以同化U、V、W風場、回波場和經由熱力反演得知變數的氣溫和壓力，預報結果最好。同化雷達個數與範圍實驗中，以不同的雷達個數和不同的同化範圍各別進行實驗，發現不同同化區域彼此之間的差異，同化雷達各數少的差異大於同化雷達各數多的差異，且同化區域涵蓋越多天氣系統，預報結果達到最佳。 This study is conducted on an Observation System Simulation Experiment (OSSE) framework. The purpose is to investigate the feasibility that by using the traditional thermodynamic retrieval technique with a combination with the GPS RO data, it is possible to recover the complete temperature and pressure fields from three-dimensional wind field. The latter can be obtained through multiple-Doppler synthesis. By assimilating the wind and thermodynamic fields into a model, one can improve the accuracy of the model quantitative precipitation forecast (QPF). Experimental results indicate that the best results are achieved when variables such as u, v, w, qr, t and p are all assimilated into the model. Assimilating qr has an immediate modification on the locations of the convections. The impact of the radar coverage provided by different radar stations in Taiwan on the model QPF is also explored.
