在中尺度氣象模擬中常需使用土地分類資料和各類土地的地表參數，然而詳細的土地使用資料不易取得，過去RAMS模式常利用AVHRR遙測資料所推求的GLCC（全球土地覆蓋特性資料）作為其預設的輸入資料，然而GLCC所使用的遙測資料不但較舊，解析度較低，而且對都市地區的判讀較不準確。現今更新的全球土地使用資料庫是MOD12Q1地表覆蓋資料庫， MODIS的遙測資料不但較新，且具有較高的解析度，其分類演算法也較準確。在本研究中我們發現這兩個資料集在台灣有很大的差異，MODIS的都市範圍比GLCC大許多，如果和台灣人口分佈，建築物面積比較，也發現MODIS的資料較為合理。除了地表分類資料外，我們也發現不同地表參數也會對模式模擬的結果產生影響，尤其是在都市地面粗糙度上，RAMS所設定的地面粗糙度過低，與台灣都市建築物高度有很大的出入。因此，本研究除了比較兩種不同土地使用資料對RAMS模式模擬結果影響，也針對修正地面粗糙度後的模擬結果進行探討。 變更土地使用資料後，台灣西部沿岸的都市幅員大幅度增加，在地面氣象場模擬結果中，可以發現地表類型為都市與農田兩者間之溫度差異約在2℃左右，且都市空氣中水氣含量較農田降低了0.002 kg/kg-air，都市熱島效應非常顯著。而提升RAMS的都市地面粗糙度，在風速的模擬上有良好的結果，地面風速減弱3m/s，模擬值與實測值大幅度的趨近。 The land use dataset and surface parameters for different land categories are important inputs for mesoscale meteorological models. The Global Land Cover Characterization (GLCC) dataset generated form AVHRR data were widely used by mesoscale meteorological models such as RAMS. However, the GLCC data have several drawbacks. They are of out of date, low resolution, and inaccurate, especially for urban classification. Another newer dataset is MOD12Q1. This dataset uses new data, has higher resolution and more accurate classification for urban area. There are significant differences between these two dataset for Taiwan area. The urban areas in MODIS dataset are much larger than that of GLCC dataset. When compare with the distributions of population and building density in Taiwan, it is found that MODIS dataset is more reasonable. We also noted that the surface parameters, especially surface roughness, are important for urban meteorological simulation. Since the buildings are higher in Taiwan area, the default setting of surface roughness in RAMS is too low when it was used in Taiwan. In this study, we investigate the effects of different land use dataset and urban roughness on the results of meteorological simulations. When the MODIS dataset was used, the surface temperatures in urban areas increased about 2℃ and the water contents in urban areas decreased about 2g/kg-air. Significant urban heat island effects were noted. If the surface roughness were increased, the surface wind speeds may decrease 3m/s in urban areas and have better agreement when compared with observations.
