國內外關於災害脆弱度的研究已漸臻成熟，各領域對於脆弱度空間分佈的討 論亦極為廣泛。考量到人口等離散型統計資料是以行政區或統計區為基礎，以及 災害潛勢等相關空間圖層套疊的問題，許多相關研究在脆弱度的空間分析上是以 如縣市、鄉鎮市區或村里等行政區為空間單元。此類型的分析操作相對容易且所 得的結果也易於解讀，但對屬於統合性質的統計資料多是以均勻分佈的面量方式 轉化為圖層，因此在直接套疊於災害潛勢的幾何圖形時產生困難。此外，以圖層 套疊進行分析時也多是應用單一組的準則權重，較少將各空間單元於指標圖層在 不同情境組合下的差異列入考量。 本研究運用專家問卷結果，藉由模糊德爾菲 (FDM) 與層級分析法 (AHP) 篩選與確立「個人、社會與地區背景」、「天然災害潛勢」、「災害應對能量」三項 脆弱度評估標的之19 個指標與相對權重，並將對應指標的圖層轉化為相同解析 度的網格，對於統合資料也透過空間內插的方式產生對應的連續面網格圖層以利 於分析。於圖層的套疊分析上，採用循序權重平均法 (OWA) 結合多準則評估 (MCE) 與線性權重組合 (WLC) 的概念，以循序權重與準則權重針對六種OWA 組合進行分析，產生由趨險到避險的不同災害脆弱度評估結果，這些由不同境況 產生的分析可供相關政策擬訂與防救災資源配置的參考。

Disaster vulnerability and vulnerability spatial distribution are both widely discussed in various domains. Since population or other census statistics are discrete data based on administrative or statistical zones, and considering the overlaying issues of different scales or formats of hazard maps, many studies on vulnerability spatial analysis are based on administrative spatial units, such as counties or townships. This spatial analysis is relatively easy to process and interpret the results, but the aggregated statistical data are usually transformed into choropleth map and difficult to overlay direct on hazard maps to analyze the spatial attributes of vulnerability. Additionally, most overlay processes utilize a single set of criteria weights, and do not consider the differences among various scenarios. This study adopts the results from expert surveys, and applies the Fuzzy Delphi Method (FDM) and Analytic Hierarchy Process (AHP) to establish 19 indicators and criteria weights according to 3 categories, namely “personal, social, and local background”, “natural hazard potential” and “disaster coping capacity”. The corresponding maps of the indicators are processed to unify their format and resolution raster layers, and the aggregated data is transformed into corresponding continuous surface type maps by spatial interpolation for overlay analysis. Ordered Weighted Averaging (OWA), incorporating Multi-Criteria Evaluation (MCE) and Weighted Linear Combination (WLC), is applied for overlay process for 6 OWA scenarios based on criterion and OWA weights to produce different results of disaster vulnerability assessment from risk-averse to risk-taking. Analytical results from different scenarios can be provided as the references for policy making and deployment of disaster prevention and rescue resources.