台灣因地形陡峭，地質脆弱再加上每年夏季常有颱風來襲，使得台灣山區容易發生崩塌，崩塌不只威脅人類生命安全還會使水庫淤積、使用壽命降低、水質變差。水庫為都市重要水源之一，然而台灣的水庫集水區大部分位於易發生崩塌的山區，若要達成永續水資源的目標，崩塌將成為水庫集水區治理的重要課題。
本研究利用9年(2004~2012)福衛二號影像判釋崩塌，以空間統計指標分析高淤積率水庫集水區崩塌地分佈特性與優先治理區位。根據水利署於2012年的報告中選出淤積率前三高水庫為研究對象，依序為曾文水庫、南化水庫、石門水庫。首先，進行集水區內崩塌與環境因子的分佈統計，建立三大水庫9年崩塌統計資料，並瞭解崩塌與環境因子的關係。其次，採用景觀指標(Landscape Metrics)、熱點分析及年間分析等空間統計方法，將崩塌圖層進行量化，以了解九年來水庫集水區崩塌地的時空分佈特性。最後，整合本研究所有統計成果與水庫淤砂率進行相關及逐步回歸分析，選出集水區優先治理區位，使崩塌資料轉化為利於集水區治理之資訊。

環境因子分析結果顯示各水庫集水區崩塌最易發生在岩性強度V及坡度7級坡之區位。曾文及石門在距水系距離小於400公尺內易發生崩塌，南化則主要分佈在距水系距離400~800公尺間；曾文、南化的易發生崩塌坡向為東、東南、南、西南，石門則是東、東北、東南；土地使用類型分別是水利使用地、裸露地及森林使用地；高程為1250~1750公尺、750~1000公尺及3250~3500公尺，此外三個水庫集水區中僅有曾文水庫集水區的崩塌受斷層影響較大。
景觀指標分析結果顯示，過去9年間曾文及南化水庫集水區內的崩塌平均面積、標準差有逐年增加之趨勢，破碎度卻逐年下降，表示崩塌在整體集水區的分佈由零散逐漸聚集甚至結合為大面積的崩塌。石門水庫集水區則是以2008年崩塌面積標準差、平均面積最大，崩塌分佈的破碎程度最小，表示該年崩塌分佈的零散程度較其它年小。透過熱點分析計算可以算出三個集水區的崩塌熱點，熱點分佈較靠近水系位置，表示高崩塌密度群集區發生在水系附近。
根據年間分析新生率、重現率、復育率的計算結果，可以排列出過去9年間對水庫集水區造成重大影響之風災事件，曾文水庫集水區依序為莫拉克(2009/08/05)、柯羅莎(2007/10/04)、海棠(2005/07/16)；南化水庫為莫拉克(2009/08/05)、卡玫基(2008/07/16)、艾利(2004/08/23)；石門水庫集水區則是艾利(2004/08/23)、莫拉克(2009/08/05)、柯羅莎(2007/10/04)。最後整合本研究的指標與水庫淤砂率選出減緩水庫淤砂的優先治理區位，曾文水庫集水區為平均崩塌面積(MPS)高的崩塌熱點；南化水庫集水區為復育率低的濱水區；石門水庫集水區為總崩塌面積(TA)高的崩塌熱點。透過以上的優先治理區位將可提高集水區治理效率，達到整治最少面積卻能有效減緩水庫淤砂率之目標。而本研究之結果將可作為集水區管理單位在進行管理及整治策略上的參考依據。

Taiwan is an island with steep sloping, vulnerable terrain that receives heavy rainfall during typhoon seasons; therefore landslides occur easily in mountain regions. Landslides not only cause loss of human life but also sedimentation in reservoirs that causes problems including loss of capacity and effects water quality. Reservoirs are the primary water source for cities, however most catchments are located in mountain regions which are prone to landslides, therefore landslides have become an important issue in regards to reaching the goal of water resource sustainability in Taiwan.
In this study, landslide data was mapped by using FORMOSAT-2 images (2004~2012) then determining spatial statistics indices to analysis the landslide distribution characteristics and site for priority management in the catchment. Here we selected the top three reservoirs with high sedimentation rates, that included Tseng-Wen, Nan-Hua and Shih-Men as a study area according to a report from the Water Resource Agency 2012. First, landslide data was with environmental factors to realize the relationship between landslide distribution and environmental factors, then using a landscape metric, a hotspot analysis was conducted to detect changes and to quantify landslide data to further realize the Spatio-Temporal distribution of landslides from 2004 to 2012. Finally, it is possible to determine the priority site for management in the catchment by integrating the results with the sedimentation rate of the reservoirs and stepwise regression.
The results show that landslides occur easily with weak rock mass strength, hilly slopes throughout the study areas. Landslides were primarily distributed on areas with a distance within 400 meters of drainage systems in Tseng-Wen and Shih-Men, within 800 meters in Nan-Hua；with an aspect facing east, south east, south and southwest in Tseng-Wen and Nan-Hua, east, northeast and northeast in Shih-Men；with land use of water conservancy in Tseng-Wen, bare land cover in Nan-Hua, forest in Shih-Men ; with elevation from 1250 to 1750 meters in Tseng-Wen, 750 to 1000 meters in Nan-Hua, 3250 to 3500 meters in Shih-Men. Additionally the landslides induced by fault zones in Tseng-Wen.
According to landscape metrics, the landslide area average and standard deviation has increased in tendency but the landslide fractures have decreased in tendency from year to year in the Tseng-Wen and Nan-Hua reservoirs. This result indicates the landslide distribution from dispersion to concentration, and even merge with each other. In Shih-Men the landslide area average and standard deviation reached the maximum but fractures reach to the minimum in 2008. This shows the landslide distribution is the most concentrated during the past year. Using the hotspot analysis, this study found that the landslide hotspot area which is near the drainage system in all the study areas, which reveals that the area with high landslide cluster is near the drainage system.
The landslide change detection shows typhoon events which cause the disturbance significant by calculating a generative ratio, reactive ratio and recovery ratio. The top three typhoon events are Morakot(2009/08/05), Krosa(2007/10/04), Haitang(2005/07/16) in Tseng-Wen, Morakot(2009/08/05) Kalmaegi(2008/07/16), Aere(2004/08/23) in Nan-Hua, Aere(2004/08/23), Morakot(2009/08/05), Krosa(2007/10/04) in Shih-Men. Finally we integrated the results with the sedimentation rate of the reservoirs and stepwise regression to decide a site for priority management in catchment. The results show the priority site for management is the hotspot area with high mean patch size(MPS) Tseng-Wen, watershed with a smaller recovery ratio in Nan-Hua, hotspot area with high total landslide area(TA) in Shih-Men. Management of these priority sites will increase the efficiency of catchment integrity. This study can be used as a reference by administrators of reservoirs in drafting relevant strategies and management policies.

Chen, C.-W., Lee, C.-C., Chen, C.-H. and Tseng, C.-P. The integration of nautical hazard assessment and harbor GIS models to the Taichung Port area in Taiwan. Natural Hazards 67(2). 275-294. (2013)
Cressie, N.A.C. Statistics for spatial data Wiley, New York. (1993)
D'Ambrosio, D., Iovine, G., Spataro, W. and Miyamoto, H. A macroscopic collisional model for debris-flows simulation. Environmental Modelling & Software 22(10). 1417-1436. (2007)
Dahal, R.K., Hasegawa, S., Nonomura, A., Yamanaka, M., Masuda, T. and Nishino, K. GIS-based weights-of-evidence modelling of rainfall-induced landslides in small catchments for landslide susceptibility mapping. Environmental Geology 54(2). 311-324. (2007)
Đặng, T.-H. and Chen, B.-H. Optimization in esterification of palmitic acid with excess methanol by solid acid catalyst. Fuel Processing Technology 109. 7-12. (2013)
Florinsky, I.V. and Kuryakova, G.A. Determination of grid size for digital terrain modelling in landscape investigations—exemplified by soil moisture distribution at a micro-scale. International Journal of Geographical Information Science 14(8). 815-832. (2000)
Frohn, R. and Hao, Y. Landscape metric performance in analyzing two decades of deforestation in the Amazon Basin of Rondonia, Brazil. Remote Sensing of Environment 100(2). 237-251. (2006)
Getis and Ord, J.K. The Analysis of Spatial Association by Use of Distance Statistics. Geographical Analysis. (1992)
Jordan, P. The use of sediment budget concepts to assess the impact on watersheds of forestry operations in the southern interior of British Columbia. Geomorphology 79(1-2). 27-44. (2006)
Kawabata, D. and Bandibas, J. Landslide susceptibility mapping using geological data, a DEM from ASTER images and an Artificial Neural Network (ANN). Geomorphology 113(1-2). 97-109. (2009)
Lin, G.-W., Chen, H., Petley, D.N., Horng, M.-J., Wu, S.-J. and Chuang, B. Impact of rainstorm-triggered landslides on high turbidity in a mountain reservoir. Engineering Geology 117(1-2). 97-103. (2011)
Lin, Y.P., Chang, T.K., Wu, C.F., Chiang, T.C. and Lin, S.H. Assessing impacts of typhoons and the Chi-Chi earthquake on Chenyulan watershed landscape pattern in central Taiwan using landscape metrics. Environ Manage 38(1). 108-125. (2006)
Lin, Y.P., Teng, T.P. and Chang, T.K. Multivariate analysis of soil heavy metal pollution and landscape pattern in Changhua county in Taiwan. Landscape and Urban Planning 62(1). 19-35. (2002)
Liu, C.-C. and Chen, P.-L. Automatic extraction of ground control regions and orthorectification of remote sensing imagery. Optics Express, 17. 7970-7984. (2009)
Liu, C.C. Processing of FORMOSAT-2 daily revisit imagery for site surveillance. IEEE Trans. Geosci. Remote Sensing 44(11): 32063214. (2006)
Liu, C.C., Liu, J.G., Lin, C.W., Wu, A.M., Liu, S.H. and Shieh, C.L. Image processing of FORMOSAT‐2 data for monitoring the South Asia tsunami. International Journal of Remote Sensing 28(13-14). 3093-3111. (2007)
Loucks, D.P. River Models. Reference Module in Earth Systems and Environmental Sciences, from Encyclopedia of Ecology. 3069-3083. (2008)
Łowicki, D. Prediction of flowing water pollution on the basis of landscape metrics as a tool supporting delimitation of Nitrate Vulnerable Zones. Ecological Indicators 23. 27-33. (2012)
Naveh, Z. and Lieberman, A.S. Landscape Ecology : Theory and Applications. Spronger-Verlag. (1984)
Pandey, A. and Dabral, P.P. Landslide Hazard Zonation using Remote Sensing and GIS: a case study of Dikrong river basin, Arunachal Pradesh, India. Environmental Geology. (2007)
Paudel, P.P., Omura, H., Kubota, T. and Inoue, T. Spatio-temporal patterns of historical shallow landslides in a volcanic area, Mt. Aso, Japan. Geomorphology 88(1-2). 21-33. (2007)
Prasannakumar, V., Vijith, H., Charutha, R. and Geetha, N. Spatio-Temporal Clustering of Road Accidents: GIS Based Analysis and Assessment. Procedia - Social and Behavioral Sciences 21. 317-325. (2011)
Risser, P.G., Karr, J.R. and Forman, R.T.T. Landscape Ecology: Drections and Applications.. Illinois Natural History Survey. (1984)
Saba, S.B., van der Meijde, M. and van der Werff, H. Spatiotemporal landslide detection for the 2005 Kashmir earthquake region. Geomorphology 124(1-2). 17-25. (2010)
SAITO, H., NAKAYAMA, D. and MATSUYAMA, H. Preliminary study on mountain slope partitioning addressing the hierarchy of slope unit using DEMs with different spatial resolution. Proceedings of Geomophometry. (2011)
Shou, K.J., Hong, C.Y., Wu, C.C., Hsu, H.Y., Fei, L.Y., Lee, J.F. and Wei, C.Y. Spatial and temporal analysis of landslides in Central Taiwan after 1999 Chi-Chi earthquake. Engineering Geology 123(1-2). 122-128. (2011)
Tsai, Z.-X., You, G.J.Y., Lee, H.-Y. and Chiu, Y.-J. Use of a total station to monitor post-failure sediment yields in landslide sites of the Shihmen reservoir watershed, Taiwan. Geomorphology 139-140. 438-451. (2012)
Vink, A.P.A. Landscape Ecology and Land Use. D.A. Davidson (Editor), Longman, London,264 pp. (1983)
Weng, M.-C., Wu, M.-H., Ning, S.-K. and Jou, Y.-W. Evaluating triggering and causative factors of landslides in Lawnon River Basin, Taiwan. Engineering Geology 123(1-2). 72-82. (2011)
Wilson, J., Repetto, P. and Snyder, R. Effect of data source grid resolution, and flow-routing method on computed topographic attributes. In: Wilson, J.P., Gallant, J.C. (Eds.),Terrain Analysis: Principles and Applications. Wiley, New York, NY, . (2000)
王思翰. 台灣運輸製造業群聚版圖變遷分析. 國立政治大學地政研究所碩士論文，台北市. (2007)
全球觀測與資料分析中心. 運用衛星影像於全島崩塌地判釋與災害分析計畫。. 行政院農業委員會林務局委託計畫系列. (2012)
朱豐沂. 以環境指標優選集水區崩塌治理區為之研究. 國立中興大學水土保持學系碩士學位論文. (2012)
李錫堤, 費立沅, 李錦發, 李彥良, 林銘郎, 董家鈞 and 張瓊文. 高屏溪流域之山崩土石流災害潛勢分析. 流域地質與坡地災害研討會. (2009)
林正杰. 運用多期福衛二號高時空分辨率影像建立崩塌災害預警模式. 國立成功大學環境工程研究所碩士論文. (2012)
許添容. 台灣地區鄉鎮市區生育率的空間與群集研究. 國立政治大學統計研究所碩士論文. (2003)
陳宏宇. 台灣山脈剝蝕過程、現今侵蝕堆積作用與長時間沈積物總量收支估算. 國立臺灣大學
(2006)
黃信誠. 空間統計簡介. 自然科學簡訊 12. 101-104. (2000)
黃漢淨. GIS與RS於崩塌地空間分布之應用. Taiwan Forestry Journal 34. (2008)
楊數榮. 台灣常用山崩分類系統. THE 14TH CONFERENCE ON CURRENT RESEARCHES IN GEOTECHNICAL ENGINEERING IN TAIWAN. (2011)
經濟部中央地調所. 環境地質與地質災害敏感區測繪作業準則. 都會區及周緣坡地整合性環境地質資料庫建置計畫. (2006)
經濟部水利署. 九十九年蓄水設施營運統計 (2012)
葉春國, 廖學誠, 詹進發, 黃正良 and 薛美莉. 地景指標與溪流水中硝酸鹽濃度關係之探討-以水里溪上游集水區為例. 中華水土保持學報 43(2): 109-122. (2012)
鄔建國. 景觀生態學－格局、過程、尺度與等級. 五南圖書出版公司. (2003)
鄭曉昀. 鄉村景觀資源評估與規劃-以平溪、雙溪地區為例. 台灣大學地理環境資源研究所碩士論文. (2004)
賴世偉, 林家榮, 鄭旭涵 and 林昭遠. 石門水庫集水區暴雨誘發型之崩塌地潛感分析. 水土保持學報 40. 163-179. (2008)
賴致瑜. 台北市住宅竊盜犯罪地圖製作與犯罪區位分析. 國立台灣大學地理環境資源研究所碩士論文. (2006)
魏秀珍. 以小集水區崩塌地密度探討影響曾文水庫流於邊坡穩定之因子. 國立成功大學資源工程學系碩士論文. (2006)
譚子健. 應用聲光模式及福衛二號遙測影像研究曾文水庫水質之時空分佈. 國立成功大學環境工程學系碩士論文. (2007)