| 摘要: | 臺灣氣象站分佈密度很高，山地氣象站分佈密度相對較少，但是山地降雨規模頻率高，其時空變異也較顯著；這使得降雨時空分佈的推估，有相對較高的不確定性。本研究根據中央氣象局25座氣象站2000-2010年來的日雨資料，透過ArcGIS上5種空間推估方法，推估雪山山脈北段及其宜蘭周邊地區的降雨分佈，檢視、比較各種方法的有效性，進而討論山地降雨時空分佈特性。
論文使用的19座氣象站2000-2010年的日雨資料，適合作為模式檢驗與區域降水討論。這些區域降水的推估成果，也被用來和其他資料有缺失的氣象站資料比較，一併討論各種空間推估的有效性。ArcGIS上5種空間推估方法包括反距離權重法(Inverse-distance-weighting)、簡單克利金(Simple Kriging)、普通克利金(ordinary kriging)、趨勢面分析(Trend Surface Analysis) 、曲面擬合(Spline-surface fitting)。它們分別用以推估年均溫、月均溫、平均年雨量與2000~2010逐年年雨量在選山山脈北段地區的分布狀況。
研究成果顯示，擁有完整溫度雨量的氣象站資料用來推估地區的年雨量，以結合DEM參數的複合克利金法較為適宜；氣象站的增加，但資料可能有缺值的情況下，以普通克利金法較為適宜；使用年平均雨量，以趨勢面法，較為適宜。
Taiwan mountain measuring climate stations distributed more scarce, spatial and temporal variability of rainfall significantly scale; this makes the estimate spatial and temporal distribution of rainfall, there is high uncertainty. In this study, according to the Central Weather Bureau 25 Stations (2000--2010 years) of rain data, estimate method through the ArcGIS five space, estimate rainfall and Xueshan of the northern section of the surrounding area Ilan distribution, to view, compare various methods effectiveness, and thus discuss temporal distribution of rainfall mountain.
Papers used 19 Stations 2000--2010 year Days of Rain opposite data integrity test as a model for discussion with regional precipitation. The results of the regional precipitation estimate, has also been used, and other information are more incomplete station data were compared, together with discussion of the effectiveness of various spatial estimate. ArcGIS Spatial estimate on five kinds of methods including inverse distance weighted method (IDW), Simple Kriging (S_Kriging), ordinary Kriging (O_ kriging), trend surface analysis (Trend), Spline-surface fitting (Spline). They are used to estimate the average annual temperature, monthly average temperature, average annual rainfall from 2000 to 2010 with yearly distribution of annual rainfall in the northern section of the election Mountains region.
We find that the three phenomena relatively complete weather station data to estimate rainfall in the region, with the composite Kriging method is more appropriate, the station using the following information may be more, but there is lack of value, ordinary Kriging method is more appropriate , use more than a decade the average annual rainfall in inverse distance weighted method, is more appropriate. |
