本研究運用兩個剖風儀觀測(UHF和L-band)驗證垂直速度的準確性，所選取的個案為2017年9月12日至13日的泰利颱風及2018年7月10日的瑪麗亞颱風，使用三維變分方法發展的多都卜勒風場反演系統(WInd Synthesis System using DOppler Measurement，簡稱WISSDOM) 加入一個S-band雷達(五分山雷達)及兩個C-band雷達(桃園機場及中央大學的雷達)來反演垂直速度。結果顯示，WISSDOM所反演的垂直速度和剖風儀所量測的相當吻合，且WISSDOM和L-band剖風儀垂直指向的模式所觀測到的垂直速度有最高的相關性及最小的方均根誤差，此WISSDOM反演的結果總體而言相較North (2017)等人在Oklahoma利用ARM雷達網驗證的研究來得好。 此外，為了提升WISSDOM在垂直速度的反演能力，將L-band剖風儀所觀測的風場設定750公尺的影響範圍並利用高斯分布的權重同化進WISSDOM中，首先利用UHF剖風儀進行此方法的風場驗證，結果顯示加入剖風儀後能有效改善WISSDOM在較低高度的風場，並且發現加入剖風儀後影響不僅在半徑750公尺內，實質可影響整個模式網格，使用桃園機場雷達進行風場驗證，結果顯示WISSDOM同化L-band剖風儀風場後能提升和桃園機場雷達徑向風的相關性並降低誤差，證明此同化方法能確切改善WISSDOM的風場。 最後，測試WISSDOM在高解析度(100公尺)於複雜地形上的風場反演能力，本項為OSSE實驗，假定兩座雷達於模式外進行反演測試，首先針對高解析度無明顯地形區域進行測試，發現WISSDOM能將風場良好的反演出來，再來將區域移置陽明山進行複雜地形測試，結果顯示WISSDOM也能反演出於複雜地形上高解析度的風場結構。
;The retrieval of vertical velocity using Doppler radar data always contains large uncertainty. In this study the quality of the vertical velocity retrieved by a 3DVar-based multiple-Doppler wind analysis system named WISSDOM (WInd Synthesis System using DOppler Measurement) is investigated. Data from one S-band (RCWF) radar and two C-band radars (RCTP and NCU) in northern Taiwan are utilized. The accuracy of the vertical velocity from WISSDOM is examined using two profilers (L-band and UHF) observations. The inter-comparison shows that WISSDOM-derived vertical velocity agrees reasonably well with the profiler data. The highest correlation and the smallest RMSD (Root-Mean-Square Deviation) take place when comparing against the L-band profiler data obtained under vertical beam scanning mode. The WISSDOM-derived results are generally better than those reported in a previous study using ARM scanning radar network in Oklahoma. The winds from the L-band profiler are also merged into WISSDOM, resulting in an overall improvement of the three-dimension wind field at low levels. UHF and RCTP are applied to verify the accuracy of the method. It is shown that WISSDOM assimilating the L-band can reduce RMSD and enhance correlation with UHF and RCTP. The result suggests that the method can get better three-dimension wind field. Finally, WISSDOM retrieved wind field in different terrain with high resolution (100m) is tested. A true wind field is simulated by WRF and two virtual radars are put outside the domain in an OSSE experiment. Both the domain set in a plain and in a mountain show that WISSDOM can retrieve a well-organized wind field.
