張能復2006-07-262018-06-282006-07-262018-06-282000-07-31http://ntur.lib.ntu.edu.tw//handle/246246/22042本研究的目的在於建立一個適用於台灣地區的 三維風場診斷模式。為了解決複雜地形對風場的影 響並同時滿足質量守恆的條件我們使用了變分分析 的技巧。在第一代變分分析中，因為對垂直風速分 量並無限制，風會選擇翻越地形而非繞流。這與空 氣污染事件發生時大氣密度成層的現象並不相符， 因此我們發展了第二代變分架構。然而，二代變分 法的統御方程式非常複雜，若使用一般常用的超鬆 弛法(Successive Over-Relaxation,SOR)求解， 收斂速率很慢，不具實用價值。本研究中，我們嘗 試以幾種不同的數值解法來求解，如強隱性法 (Strongly Implicit Procedure,SIP)、多重網 格法(Multigrid method)等等。我們發現在各種 方法中，以SIP 法求解最有率。尤其求解複雜地形 的風場時，SIP 法的優勢更明顯。In this research, we tried to establish a diagnostic wind field model to help solve the problem about pollutant dispersion. We took the variational analysis technique in order to let the wind field be satisfied with the continuity equation. In the first generation of variational analysis developed by Chang(1983), there is no constraint on the vertical wind vector so that the wind will cross over the terrain rather than around the obstacle. This does not match with the phenomenon that the air flow is a highly stratified fluid, thus we developed the second generation of variational analysis. Since the governing equation of the second generation of variational analysis is very complicated, the convergence rate isslow when using successive over-relaxation (SOR) method to solve it. In this research, we use seven different methods, such as Strongly Implicit Procedure(SIP) ,Full Multi-grid-method(FMG),Successive-OverRelaxation(S OR)method, etc. ,to solve the governing equation of second generation of varational analysis. After a series of checking procedures, we have found that SIP method is the most efficient one among these seven methods. Especially when complex terrain is considered, the superiority of SIP method is even more significant. In the cases we have tested, the numerical convergence rate of using SIP method is at least seven times faster than SOR method.application/pdf243311 bytesapplication/pdfzh-TW國立臺灣大學環境工程學研究所風場地形診斷模式強隱性法(SIP)多 重網格法鬆弛法(SOR)變分分析wind fieldterraindiagnostic modelSIPMulti-Grid methodSORvariational analysisreporthttp://ntur.lib.ntu.edu.tw/bitstream/246246/22042/1/892211E002012.pdf