降雨量預測的準確性在防災上有著重要的意義，颱風來襲時容易造成在短時期內降下大量豪雨，這樣突然產生大量降雨的情形容易引發土石崩塌。根據中央氣象局1911-2019年以來的紀錄，颱風登陸次數以臺灣東岸的宜蘭至花蓮間為最多，花蓮至成功第二，且全臺灣1726條土石流潛勢溪流中就有170條位於花蓮，排名全台第三。水保局自2005年起針對土石流潛勢溪流附近的村里長等在地熱心民眾進行土石流防災專員培訓，土石流防災專員在災害應變期間的任務中，包含進行自主雨量觀測，是提升防災執行效率的參考。
本研究的目的是希望比較不同空間插值方法來得到較適合花蓮縣的空間插值方式，並瞭解防災專員測量的雨量資料與估計值上的差異，最後參考防災專員與估計值間的差異作為未來雨量站增設地點之參考。
本研究先利用2015至2019年間花蓮縣雨量站的颱風累積雨量及最大時雨量資料，來比較反距離權重法(Inverse Distance Weighted, IDW)、薄板樣條插值法(Thin Plate Splines, TPS)及克利金法(Kriging)等常用的空間插值方法的優劣，並透過交叉驗證來計算實際值與估計值的均方根誤差(root mean square error, RMSE)和平均絕對誤差(mean absolute error, MAE)來進行驗證。在比較驗證的這十個颱風中，在克利金法的表現較好的颱風數量最多。
將土石流防災專員測量的雨量資料及雨量站資料利用克利金法計算的土石流防災專員所在地的雨量資料進行比較，並參考前人研究將土石流防災專員與雨量站的資料趨勢特徵分成6種，分別是最後一致(+)、最後一致(-)、約略相同、完全不同(+)、完全不同(-)及不規則。基於土石流防災專員量測雨量的功能，主要在彌補雨量站不足並作為提前疏散依據，因此，如果雨量站的數值大於或等於防災專員所量測的地方，則以雨量站資料為主即可。將雨量資料分析所得的結果顯示，除了少數幾個土石流防災專員的資料，大部分的土石流防災專員測量的雨量資料是與雨量站具有補足或提前疏散依據的意義的。

The accuracy of rainfall prediction is of great significance in disaster prevention. Typhoon will cause heavy rain in a short time period, and sudden heavy rainfall may cause landslides. According to the records of the Central Weather Bureau (CWB) since 1911-2019, the number of typhoons made landfall between Yilan and Hualien is the most one, and between Hualien and Chenggong ranked secondly. Of the 1,726 potential debris flow torrent in Taiwan, 170 are located in Hualien, ranked third in Taiwan.
Since 2005, Soil and Water Conservation Bureau has provided debris flow disaster prevention volunteer training to residents near the potential debris flow torrent. Debris flow disaster prevention volunteers' tasks during the typhoon include autonomous rainfall observation. This is a reference to improve the efficiency of disaster prevention execution.
This paper hopes to compare different spatial interpolation methods to obtain a suitable spatial interpolation method for Hualien. Understand the difference between the rainfall data measured by debris flow disaster prevention volunteer and the rainfall data estimated by the spatial interpolation method. Refer to the rainfall data measured by debris flow disaster prevention volunteers to find the location of additional rainfall stations in the future.
The typhoon accumulated rainfall and maximum hourly rainfall data from the Hualien County Rainfall Station between 2015 and 2019 has been applied to compare the advantages and disadvantages of commonly used spatial interpolation methods such as Inverse Distance Weighted (IDW), Thin Plate Splines (TPS) and Kriging, and further to calculate the root mean square error (RMSE) and mean absolute error (MAE) of the actual value and the estimated value through cross-validation for verification. Among the ten typhoons, the number of typhoons that performed well in Kriging was the largest.
the Kriging method performs better in the root mean square error and the mean absolute error of the accumulated rainfall. The Kriging method is slightly better than Thin Plate Splines in the root mean square error and mean absolute error for the maximum hourly rainfall, and Inverse Distance Weighted performs poorly.
The rainfall data measured by the debris flow disaster prevention volunteer was compared to the rainfall data of the rainfall station data calculated by the Kriging method. According to previous studies, the rainfall data trend characteristic of the debris flow disaster prevention volunteer and the rainfall station are divided into 6 types, which are Final Agreement (+), Final Agreement (-), About the same, Completely Different (+), Completely Different (-) and Irregular. Since the function of the rainfall data measured by the disaster prevention specialist is mainly to be the early warning or complement the rainfall station, if the value of the rainfall station is greater than or equal to the place measured by the debris flow disaster prevention volunteer, the rainfall station data can be used as the main data. The results of the analysis of rainfall data show that, except for the data of a few debris flow disaster prevention volunteer, most of the rainfall data measured by the debris flow disaster prevention volunteer have the significance of complementing the rainfall station or being the early warning.

第壹章 緒論 1
壹、 研究動機 1
貳、 研究目的 4
第貳章 文獻回顧 5
壹、 降雨量空間插值 5
貳、 土石流防災專員 8
第參章 材料與方法 13
壹、 雨量站資料 13
貳、 颱風資料 17
參、 空間插值方法 25
肆、 土石流防災專員 30
第肆章 結果與討論 33
壹、 插值方法比較 33
貳、 防災專員比較 40
參、 討論 164
第伍章 結論與建議 165
壹、 結論 165
貳、 建議 166
第陸章 參考文獻 167
附錄 170

Adhikary, S. K., Muttil, N., & Yilmaz, A. G. (2017). Cokriging for enhanced spatial interpolation of rainfall in two Australian catchments. Hydrological processes, 31(12), 2143-2161.
Bookstein, F. L. (1989). Principal warps: Thin-plate splines and the decomposition of deformations. IEEE Transactions on pattern analysis and machine intelligence, 11(6), 567-585.
Burrough, P. A. (1986). Principles of Geographical. Information Systems for Land Resource Assessment. Clarendon Press, Oxford.
Chen, F.-W., & Liu, C.-W. (2012). Estimation of the spatial rainfall distribution using inverse distance weighting (IDW) in the middle of Taiwan. Paddy and Water Environment, 10(3), 209-222.
Chen, S.-C., & Wu, C.-Y. (2014). Debris flow disaster prevention and mitigation of non-structural strategies in Taiwan. Journal of Mountain Science, 11(2), 308-322.
Chen, Y. C., Wei, C., & Yeh, H. C. (2008). Rainfall network design using kriging and entropy. Hydrological Processes: An International Journal, 22(3), 340-346.
Correa, C. C. G., Teodoro, P. E., Cunha, E. d., Oliveira-Júnior, J. d., Gois, G., Bacani, V., & Torres, F. (2014). Spatial interpolation of annual rainfall in the State Mato Grosso do Sul (Brazil) using different transitive theoretical mathematical models. International Journal of Innovative Research in Science, Engineering and Technology, 3(10), 16618-16625.
Gentile, M., Courbin, F., & Meylan, G. (2013). Interpolating point spread function anisotropy. Astronomy & Astrophysics, 549, A1.
Griffith, D. A. (1987). Spatial autocorrelation. A Primer. Washington DC: Association of American Geographers.
Gunarathna, M., Nirmanee, K., & Kumari, M. (2016). Are geostatistical interpolation methods better than deterministic interpolation methods in mapping salinity of groundwater. Int. J. Res. Innov. Earth Sci, 3(3), 59-64.
Jung, Y., Kim, H., Baik, J., & Choi, M. (2014). Rain-Gauge Network Evaluations Using Spatiotemporal Correlation Structure for Semi-Mountainous Regions. Terrestrial, Atmospheric & Oceanic Sciences, 25(2).
Luo, W., Taylor, M., & Parker, S. (2008). A comparison of spatial interpolation methods to estimate continuous wind speed surfaces using irregularly distributed data from England and Wales. International Journal of Climatology: A Journal of the Royal Meteorological Society, 28(7), 947-959.
Mitas, L., & Mitasova, H. (1999). Spatial interpolation. Geographical information systems: principles, techniques, management and applications, 1(2).
Plouffe, C. C., Robertson, C., & Chandrapala, L. (2015). Comparing interpolation techniques for monthly rainfall mapping using multiple evaluation criteria and auxiliary data sources: A case study of Sri Lanka. Environmental Modelling & Software, 67, 57-71.
Rezca Kurnia, P. (2018). 雨量站與防災專員量測雨量之比較研究. (碩士), 國立東華大學, 花蓮縣. Retrieved from https://hdl.handle.net/11296/wp7ky8
Sui, D. Z. (2004). Tobler's first law of geography: A big idea for a small world? Annals of the Association of American Geographers, 94(2), 269-277.
尹孝元、李鎮洋、詹錢登、林美聆 (2015). 台灣土石流災害管理決策支援系統. [Nationwide Decision Support System for Debris Flow Disaster Management in Taiwan]. 工程環境會刊(35), 65-84. doi:10.6562/jee.2015.35.4
王天華. (2018). 西南地區降水量插值方法比較——以麗江市為例. 长江科学院院报, 35(10), 21-24.
任芝花、馮明農、張洪政、鞠曉慧、王穎. (2007). 自動與人工觀測降雨量的差異及相關性. 應用氣象學報, 18(3), 358-364.
朱求安、張萬昌、余鈞輝. (2004). 基于GIS的空间插值方法研究. [The Spatial Interpolations in GIS]. 江西師範大學學報（自然科學版）, 28(2), 183-188.
行政院農業委員會水土保持局. 土石流潛勢溪流統計. Retrieved from https://246.swcb.gov.tw/Info/Potential_Statistics
行政院農業委員會水土保持局. 土石流警戒基準值訂定方法. Retrieved from https://246.swcb.gov.tw/Info/Debris_Function
行政院農業委員會水土保持局. 災害預報. Retrieved from https://246.swcb.gov.tw/DisasterApplication/Disasterdeal_DisasterForecast
行政院農業委員會水土保持局. (2019). 土石流潛勢溪流劃設作業要點. Retrieved from https://www.swcb.gov.tw/Laws/laws_more?id=35102840f34e441e809cfbffd7bbb055
李翠平、李仲學、余東明. (2007). 基于泰森多边形法的空间品位插值. [Ore Grade Interpolation Based on Thiessen Polygon Method]. 遼寧工程技術大學學報（自然科學版）, 26(4), 488-491.
汪青靜、許崇育、陳華. (2016). 克里金日降水插值的不同变异函数比较分析 Comparison and Analysis of Different Variogram Functions Models in Kriging Interpolation of Daily Rainfall. Journal of Water Resources Research, 5(05), 469.
沈哲緯、蕭震洋、羅文俊. (2012). 花蓮縣土石流潛勢溪流地文特性初探. [A Preliminary Study of Geomorphologic Factors Characteristics of Potential Debris Flow Torrents in Hualien County]. 水保技術, 7(2), 96-105. doi:10.30040/jswct.201204.0004
林宛儒. (2010). 應用地理統計方法於雨量站網調整之研究－以臺北水源特定區為例. (碩士), 逢甲大學, 台中市. Retrieved from https://hdl.handle.net/11296/sar7uf
花蓮縣政府. (2018). 花蓮縣行政區域圖. Retrieved from https://www.hl.gov.tw/Detail/7845c534335d48a0a2d6b61268b69cee
張松林、張崑. (2007). 全局空间自相关Moran指数和G系数对比研究. [Comparison between General Moran's Index and Getis-Ord General G of Spatial Autocorrelation]. 中山大學學報（自然科學版）, 46(4), 93-97.
張強、涂滿紅、馬舒慶、楊志彪、羅永春. (2007). 自动雨量站降雨资料质量评估方法研究. 应用气象学报, 18(3), 365-372.
張壹壹. (2019). 動員土石流防災專員回傳雨量在防災應變之成效分析. (碩士), 國立東華大學, 花蓮縣. Retrieved from https://hdl.handle.net/11296/hhtkh9
陳巧涓. (2014). 土石流防災專員培訓之成效分析. (碩士), 國立中興大學, 台中市. Retrieved from https://hdl.handle.net/11296/kpmd7m
陳怡君、吳榮平. (2017). 防災社區自主性評估之探討. 台灣社區工作與社區研究學刊, 7(2), 1-44.
陳炳森、李建平、曾宇良、莊翰華. (2013). 農村防災性社會資本建構創新策略研析~ 以土石流災害為例. 農業推廣文彙, 89-122.
陳茹蕙. (2007). 以連續機率分佈熵及克利金建構雨量站網. (碩士), 國立臺北科技大學, 台北市. Retrieved from https://hdl.handle.net/11296/6957a6
黃明萬. (2001). Kriging方法於地質圖製作之應用. (碩士), 國立交通大學, 新竹市. Retrieved from https://hdl.handle.net/11296/m9m5c8
萬龍、馬芹、張建軍、付艷玲、張曉萍. (2011). 黄土高原降雨量空间插值精度比较-KRIGING 与 TPS 法. 中国水土保持科学, 9(3), 79-87.
詹錢登、李明熹. (2004). 土石流發生降雨警戒模式. 中華水土保持學報, 35(3), 275-285.
靳國棟、劉衍聰、牛文傑. (2003). 距离加权反比插值法和克里金插值法的比较. 长春工业大学学报.
趙娜、焦毅蒙. (2018). 基于 TRMM 降水数据的空间降尺度模拟. 地球信息科学学报, 20(10), 1388-1395.
劉怡君、陳亮全. (2015). 防災社區之回顧與課題. 災害防救科技與管理學刊, 4(2), 59-81.
劉怡君、曾敏惠. (2014). 本土防災社區的推動與深化. 災害防救科技與知識專欄.(檢索日期 2017.04. 23), 取自 http://www. ncdr. nat. gov. tw/upload/epaper/086. pdf.
劉雨、劉廣磊、姜自武、龔佃選. (2015). 径向基函数插值配置点的自适应选取算法 An Adaptive Method for Choosing Collocation Points of RBF Interpolation. Advances in Applied Mathematics, 5, 8.
謝惠紅、鄭士仁、劉璟燁、周世昌、 蕭博仁. (2006). 夏季日降雨量空間分佈特性之研究. [Characterization of Spatially Distributed Summer Daily Rainfall]. 農業工程學報, 52(1), 47-55. doi:10.29974/jtae.200603.0005