本研究計畫將於中央氣象局未來作業用全球模式系統(FV3GFS)架構下，進行混成資料同化系統的建置與開發，強化具氣象局特色之全球預報系統。計畫第一年將以氣象局的GSI資料同化系統為基礎，開發混成增益矩陣資料同化系統(HGDA)，並以單點觀測實驗比較各同化系統特性，藉此評估同化系統開發所需之參數及適合作業用的同化策略。第二年將藉由循環分析實驗，以敏感度實驗測試最佳化結合權重，並評估建立隨波譜特性調整最佳結合權重之迫切性。第三年將致力於QR-HGDA同化系統的建置，並將修正量的正交向量診斷應用於FV3GFS-GSI同化系統，釐清EnVAR同化系統的修正特性。此三年期研究計畫成果預期可完成FV3GFS-HGDA系統的建置與效益評估，並可培養國內自主研發與強化全球模式的維護能力。 ;CWB has advanced its global operational system toward the next-generation system, the FV3GFS, coupled with a Gridpoint Statistical Interpolation (GSI) hybrid ensemble-variational (EnVar) data assimilation system (FV3GFS-GSI). This system is planned to be operational in 2021. Different from hybridizing the background error covariance in EnVar, a three-year project is planned to explore a different possibility of hybrid data assimilation and aims to develop the hybrid gain data assimilation (HGDA) system using the ensemble and variational DA components from the FV3GFS-GSI system. In the first year, the major task is to implement the HGDA system, including the feasible steps to combine the sub-DA components of the FV3GFS-GSI system. A single observation experiment will be conducted to investigate the properties of each DA system, including the FV3GFS-GSI, the FV3GFS-HGDA and its component DA systems. The results of a single observation experiment would help evaluate the assimilation strategy and optimize the setup of the HGDA system. For the second year, cycling experiments will be conducted to examine the applicability and feasibility for the FV3GFS-HGDA under a semi-operational environment. A series of sensitivity tests are necessary to justify the optimal combination weight. Except for a single weight value, a spectrum-dependent combination weight would be a potential option for improving the hybrids if needed. In the third year, we will focus on implementing the QR-HGDA algorithm to evaluate the possibility of removing the ad-hoc determined combination weight in hybrids. The newly derived orthogonal analysis increment through QR-factorization can be used to understand the characteristics of the increments in the current FV3GFS-GSI system. Through the three-year study, we are able to develop a hybrid DA system and such experiences can feedback to the FV3GFS-GSI development. The optimal goal of this study is to enhance CWB’s ability in DA research and global forecast system in Taiwan.
