本文使用WRF 3DVAR,將GPS掩星觀測折射率資料及傳統觀測資料同化於WRF中尺度模式,以了解同化折射率對於數值模擬結果的影響。文中使用CHAMP衛星折射率觀測資料,選取兩個颱風個案海棠(2005)以及敏督利(2004)進行模擬。我們將每一個颱風作數組模擬,分為沒有及有同化GPS掩星折射率資料,及在模擬期間加入GPS掩星折射率觀測資料同化(即cycling 3DVAR)。同化後,在差異增量上,折射率在水氣方面有較大的回饋,但在溫度方面回饋較小。WRF模擬結果顯示,同化GPS掩星觀測折射率資料對於颱風路徑模擬的影響較小,對降雨模擬則有改善。CHAMP衛星的掩星觀測只能提供水氣跟溫度,缺乏風場資訊,因此我們另使用QuikSCAT衛星所提供的海面風場作為觀測資料同化於模式中。同時同化GPS掩星資料及QuikSCAT風場資料的WRF模擬結果顯示,對路徑模擬誤差不大的個案,同化QuikSCAT風場後可改善累積降雨分布特徵。
This study utilizes WRF 3DVAR to assimilate GPS radio occultation (RO) refractivity and conventional observations into model for assessment on the impact of these refractivity data on numerical prediction with WRF. The RO observations from CHAMP were used for assimilation. Two typhoons impinging Taiwan were chosen for simulations with no assimilation, with initial GPS RO observations and the GPS observation at later prediction times (i.e., cycling). With assimilation of GPS refractivity, the moisture field has relatively larger initial increments than the temperature field. The simulation results show less impact on typhoon track prediction but more on rainfall prediction. For sake of lack of wind information associated with CHAMP RO observations, the QuikSCAT near-surface wind observations over the ocean were assimilated into the model for comparisons. With both GPS RO refractivity and QuikSCAT wind assimilated, the distributions of accumulated rainfall for the WRF prediction have been improved.