對於模擬台灣這種小區域的氣候特性而言，高解析度的區域模式可表現出台灣本身複雜的地理型態對區域氣候的影響，但模式所需地表變數的觀測資料相當缺乏。由於台灣地區並無合適的土壤觀測資料可供使用，因此利用陸面模式(NOAH LSM)以off-line的模擬方式，得到台灣地區的土壤溫度及溼度分析場，並將此土壤資料作為MM5模式下邊界條件的土壤初始場，模擬台灣地區在2001年5月到8月的氣候狀態，探討高解析度的土壤資料對於模擬台灣地區氣候的影響。本研究依照在模擬過程中是否使用off-line土壤資料與地面觀測資料同化的差異，做六組的實驗組模擬。藉由比較不同實驗的結果，探討氣象場與土壤之間的交互作用。 對台灣地區而言，ERA 40的土壤分析場在模式初期存在明顯誤差(土溫偏高且土溼偏低)，此土壤初始場的誤差，依照由淺至深的土壤層，分別約需模擬一個月至三個月以上的時間，方可使土壤狀態達到平衡；使用off-line LSM的土壤資料可改善土壤初始場誤差的缺點。模式對台灣地區氣候模擬的結果顯示，在台灣西部地區的夏季降水模擬偏少，土壤溼度有可能偏低，使得白天地表溫度場的模擬為暖偏差，氣溫的日變化振幅過大。在台灣東部地區的夏季降水模擬偏多，土壤溼度有可能偏高，使得白天地表溫度場的模擬為冷偏差。 各實驗組對地表及2公尺溫度場的模擬比較，顯示是否加入off-line LSM的土壤資為造成實驗組彼此間溫度場模擬差異的主要原因。地面資料同化與初始土溫調整對長期地表及2公尺溫度場模擬的影響較小。在降水偏少的情形下，使用off-line LSM的土壤資料可以改善白天地表最高溫模擬的暖偏差，以及氣溫日變化振幅過大的情況；輸入地面觀測資料同化可以改善2公尺氣溫模擬的冷偏差或暖偏差，也間接改善了對地表溫度的模擬。模式可合理模擬出台灣地區夏季的降水空間分布，受到台灣地形效應影響，夏季降水主要分布在東部及南部，降雨量的最大值在山區，且降水模擬在東部地區有偏多的現象。 Using high-resolution regional climate model can perform regional climate’s characters in Taiwan which has complex geography. The land-surface observations provided for model simulation are short of in Taiwan, and therefore we utilize off-line land surface model(NOAH LSM) to simulate soil data in Taiwan. We put soil data in Taiwan in regional model(MM5) with low boundary conditions to simulate climate from May to Aug in 2001 and study the impact of high-resolution land surface data by off-line land surface model on regional climate. There are six experiments with different soil data and data assimilation to study the interaction between atmosphere and soil. In Taiwan, MM5 with uncorrected soil data provided by ERA 40 which soil state is drier and warmer need to run for three month at least to ensure convergence to soil state equilibrium. MM5 with soil data provided by off-line land surface model can avoid the error of initial soil state. Surface verification statistics indicated that MM5 overestimated maximum surface temperature and diurnal air temperature cycle over west region of Taiwan because it simulated less precipitation in summertime than observation, and its soil state was drier. MM5 underestimated maximum surface temperature over east region of Taiwan because it simulated more precipitation in summertime than observation, and its soil state was wetter. MM5 would have a great impact on surface temperature and air temperature simulation due to different soil initialization. MM5 with soil data provided by off-line land surface model would correct the warm bias of maximum surface temperature and improve to simulate the daily changes of air temperature. MM5 with data assimilation would improve to simulate maximum surface temperature and air temperature in the daytime both. MM5 simulated precipitation pattern over Taiwan in summertime reasonably. Heavy rainfall area was located in the east and south area of Taiwan mostly. However, MM5 simulated precipitation too much in the east area of Taiwan.
