台北山坡地災損潛勢分析與歷史災害驗證

Developing and Verifying a Composite Index of Disaster Potential Evaluation in Hillside Areas of Taipei

沈哲平(Zhe-Ping Shen)

指導教授 ： 陳偉堯

國立臺北科技大學/工程學院/土木與防災研究所/碩士(2007年)

https://doi.org/10.6841/NTUT.2007.00218

摘要

台灣地區山勢陡峭，水流迅急，位於亞太平洋地震帶上，地理環境特殊，颱風豪雨及地震頻繁，使山崩與土石流成為台灣山坡地最常見的地質災害。台灣山坡地面積約為263 萬餘公頃，約佔全島面積73%，近年來人口增加，平地利用已趨飽和，使得國人不斷的轉往山坡地開發，而山坡地的超限利用和坡地的不當開發，輕則破壞當地生態，重則影響地質結構而引起嚴重災情，自1976-2007 年間，臺北市與臺北縣之歷史山坡地災害事件高達867 筆，故坡地災害的防治與預防極為重要。
吾人以長寬各一公里之網格覆蓋臺北市及臺北縣區域(合稱台北地區)，藉由台北地區周邊相關圖層與山坡地歷史災害資料的彙整，建置土地利用、土石流潛勢溪流、降雨量、坡度、坡向、崩塌地、活動斷層、土壤、地層分佈、順向坡、人口密度、道路密度等12 個評估項目，使用Fortran 開發災害潛勢分析系統簡化繁雜的GIS 處理及數據分析，藉由指標彙整、數據分析、立體建模及類神經網路分析之一體化來輔助處理各項研究資料。
本研究使用專家座談評分法、專家問卷層級分析法、多變數迴歸分析及類神經網路分析評估綜合指標模型的有效性，利用歷史災情資訊進行災害點位與指標分數兩者間的數據探討和評估系統之驗證與優化，結果顯示類神經網路分析所模擬之臺北市山坡地歷史災情和真實資料的判定係數為0.793，其所建構之評估模型的災害潛勢預測力較其他三者為佳，且達統計上的顯著水準，本研究並以Google Earth 作為整合平台，使用平面色塊及立體柱狀方式呈現台北地區山坡地災害潛勢的分佈情形，可提供決策單位對於山坡地災害潛勢分析之參考。

關鍵字

地理資訊系統 ； 山坡地災損潛勢分析 ； 類神經網路

並列摘要

Located in the circum-Pacific seismic zone, Taiwan is characterized geomorphically by steep mountains and rapid waters and climatically by frequent typhoons and pouring rain, making landslides and debris flows the most common geologic hazards in Taiwan’s hillside regions that boast a total area of 2,630,000 hectares or 73% of the island’s entire area. The rapid increase in population in recent years has prompted people to develop hillside land. However, over-development of slope areas may bring about drastic impacts, damaging local ecological systems at the least and causing geologic disasters at its worst. In fact, the greater Taipei area (Taipei City and County) has witnessed as many as 867 slope hazards during the past 30 years, rendering prevention and management of hillside disasters a crucial priority.
This study accordingly used 1 km x 1 km grids to analyze the disaster potential of hillside areas in the greater Taipei area. Based on related geographical data and historical records and consulting the evaluation criteria proposed in Disaster Prevention and Ecological Engineering Study of Hillside Regions surrounding Major Cities by Soil and Water Conservation Bureau, 12 factors were identified for developing a Composite Index of Disaster Potential Evaluation in the hillside districts of the greater Taipei area. The 12 factors are respectively land use, distribution of streams with potential for debris flow, rainfall, slope gradient, slope orientation, landslide, fault type, soil type, soil formation, population density, road density, consequent slope. Fortran application programs were adopted to simplify the complicated GIS data processing and to provide an integrated model for assisting various analysis tasks.
Scores obtained by the proposed index were further verified by the collected historical data, and the geovisualization software of Google Earth was then used as the GIS platform to display the distribution of at-risk hillside districts in the greater Taipei area. Results of our study can be expected to provide related government agencies with useful references in decision making. Moreover, while factor scoring, data analysis, evaluation, display, and other tasks needed to be performed separately in previous methods, our proposed methode has the advantage of completing all the tasks with one single program. The model is further able to adjust the weight assigned to each factor to help reach more accurate evaluation.

並列關鍵字

GIS ； Disaster Potential Evaluation ； Neural Network

參考文獻

[20] 陳圭宏（2007），臺北地區農業氣象環境與災害發生潛勢，作物、環境與生物資訊期刊，第四卷，第一期，1-10 頁。

[26] 黃南輝，黃建忠，陳宏宇，李錫堤，潘國樑（2002），環境地質敏感度分析自動化系統開發，地工技術，第90 期，55-62 頁。

[1] Davis, J. C.(2002).Statistics and Data Analysis in Geology 3 edition. John Wiley & Sons, New York, PP.656

[4] Gray, D. H. and Sotir, R. B. (1996). Biotechnical and Soil Bioengineering Slope Stabilization, A Practical Guide for Erosion Control, Wiley, 378pages

[5] Ishizuka, H., Sasaki, K., Matsuura, S., Kamiya, M., Sunahara, H., and Esaki, H. (2006) Collecting Adaptive Data for Isolated Wireless Sensors with Patrol Nodes in Live E!. MDM 2006: 136