本研究採用兩種多變量地理統計方法,包括反距離權重法與克利金法,探討莿竹林小集水區降雨量空間分布特性;選定國立屏東科技大學水土保持戶外教室西南側莿竹林小集水區,利用4組自計式雨量器、2組普通式雨量器及6組簡易式雨量器,平均設置於集水區內以形成雨量站網,雨量資料收集時間由2009年7月至2011年3月。隨後首先探討雨型與降雨量內插方法之關係,再依據各方法所建立小集水區內降雨量空間分布差異性,檢討合適的雨量站網之設置,進而求得小集水區具代表性之降雨量。研究發現,集水區內氣流行進受谷地地形束縮之影響,導致林外降雨量大於林內降雨量;而根據平均標準誤差與均方根誤差方法,來計算推估降雨量與實際降雨量之差異性,發現克利金法較反距離權重法準確。相對來說,若降雨量估測值與實際值間之平均標準誤差與均方根誤差值越大,代表雨量站具有代表性,因此推得小降雨量時,谷地之雨量站較重要,而暴雨時則是谷地中下游設置雨量站較佳。
This study adopted two multivariate geo-statistical methods, including Inverse Distance Weighted Method (IDWM) and Kriging Method (KM), to discuss the space distribution of rainfall depth in a small watershed with thorny bamboo plantation. The experiment watershed is located in the soil and water conservation outdoor classroom of National Pingtung University of Science and Technology. The network of rainfall stations were formed by four sets of self-recording rain gauge, two sets of general rain gauge, and four sets of simple rain gauge from. The average distribution method is used to set the rainfall station and the experimental time of rainfall depth is from July, 2009 to March, 2011. The relationship between rainfall type and the interpolation method of rainfall depth was discussed firstly. The representative rainfall depth was obtained by the location discussion of rain gauge locations, which can be determined by the difference between various methods of space distribution of rainfall depth. After the detailed analysis, the outer rainfall depth is larger than that of inner due to the influence of the terrain beam reduction of valley. This study used the average standard error and root mean square error to calculate the difference between the estimation rainfall (ER) and the actual rainfall (AR), and KM is more accurate than IDWM. Relative speaking, a large difference between ER and AR represents the importance of rain gauge location. Based above assumptions, this study found that the rain gauge located in the valley is more important than other places for small rainfall, and the downstream of valley is more important than others for torrential rainfall.