區域災害系統評估不同空間尺度之災害風險研究 - 以臺灣臺東縣坡地環境為例

Abstract

災害是人類資源經營系統與自然環境間交互作用的產物。世界銀行指出臺灣曝露於三種自然災害下之人口與面積均為世界之冠。在臺灣，過去十年來自然現象所導致的災害損失逐年上升，因此深入了解災害發生的原因有其必要性，且藉由了解災害發生之因，落實有效的預備、減災措施，可將災害損失降低。首先，需收集並確認災害的點位、分析災害的孕災環境因素以推估相鄰的環境特性，找出環境敏感地帶。再者，災害對於居民會產生多大的風險程度，並不只能考慮孕災環境與致災因子可能造成的威脅，還必須考慮承災體的曝露與脆弱程度，這扮演著相當關鍵的角色。近年來，國內外整合型災害研究日漸盛行，其研究成果常將國家尺度指標直接應用至地方縣市尺度評估，因此無法突顯區域特性，掩蓋次國家尺度等級之變化，成果難以深入了解災害風險之因。故本研究欲整合坡地災害兩大類群研究的風險評估，並考量不同空間尺度所需選擇指標不同之問題。目的就是以區域災害系統論，提供一個了解人與環境之間互動的有利分析方式，並結合GIS平臺以空間化的方式展現其成果。 孕災環境敏感度部分，本研究採用坡度、高程、坡向、地質、邊坡縱向曲率與橫向曲率、土地利用、河流距離、斷層距離、道路距離、NDVI，共11項指標，以及各種降雨型態指標評估致災因子。結果發現，不同地區因為環境的特性不同，影響塊體崩壞潛在因子的重要程度將會有所不同，但NDVI與坡度二者指標不受空間尺度與區域塊體崩壞特性影響，適用各種大小尺度範圍。 在承災體曝露量與脆弱度部分，本研究將建物、道路與耕作地做為承災體曝露之型態；脆弱度部分選用低收入戶、身心障礙人口、離婚與喪偶率、教育程度、扶養比、醫療資源、聯外道路、人口密度等適合地方至全國尺度之指標，以及評估最小尺度個體、村落適應能力之災害識覺與調適行為。結果發現，大尺度縣市評估難以準確評估各鄉、各部落的脆弱程度。 在風險評估方面，本研究將極端現象影響下的自然危害程度、承災體曝露量，以及承災體脆弱程度標準化，加總平均得到範圍介於0-1之間的風險值。整體而言，不同空間尺度的分析成果確實不同，以大尺度評估容易掩蓋小尺度的區域特性。以三鄉尺度評估結果來看，海端鄉應是風險最高地區，金峰鄉與大武鄉互有高低。事實上並非如此，大武鄉因區域面積小，易受大尺度均化影響，而忽略其風險程度。實際上若以村尺度而言，大武鄉大鳥村的風險值與海端鄉利稻村相當。最後，本研究發現若大小空間尺度的劃分一致(如以主流域做為劃分)，則鄉與村區域特性相近，風險變化之間的差異小，若不一致則反之。故在進行災害研究時，若未重視不空間尺度與區域特性之差異，其風險結果將會嚴重偏離災害的真實性，難以探究災害之因。 本研究提出的概念是地理學者長期關注的人地互動關係，概念的重新詮釋，以及透過GIS的應用，表現區域災害風險在空間與時間的變化性，對於學科未來在災害管理上能有重大的貢獻。整合自然與人文的概念角度出發，對於中央政府由在地思考擬定合適的防災政策、規劃都具有實質的應用價值。

A natural hazard may be defined as a systematic interaction between human resource management and extreme or rare natural phenomena. The World Bank has reported Taiwan as the most vulnerable country in the world in terms of the percentages of exposed areas and multiplicity disaster. In Taiwan, losses from extreme natural phenomena have escalated in the past decade. There is a noticeable change in policy, with more emphasis on loss reduction through mitigation, preparedness, and recovery programs. Effective mitigation of losses from hazards requires hazard identification, analysis of the time and spatial characteristics of mass wasting, and assessment of the sensitivity of potential hazard environment along the side of study area. The degree to which populations are vulnerable to hazards, however, is not solely dependent upon proximity to the source of the threat or the physical nature of the hazard—exposure and vulnerability of residents also play a significant role in determining vulnerability. Integrated study of potential environment with residents’ vulnerability is now a popular topic around the world. Although the results could usually found national-level indices were applied to sub-national level directly, it is impossible to apply a national-level index to a smaller scale of analysis, or indeed to aggregate risk across scales, because the processes that cause risk are manifest differently at each scale.The regional disaster system provides us with a method of analyzing relationships between residents and environment, and with spatially presented results through GIS. The formation of a disaster is usually complex and need to be studied comprehensively. In this study, we used regional disaster system to interpret causes of disaster in terms of different spatial scales. Eleven geographic factors are included in the sensitivity analysis of potential hazard Environment as covariates: slope, elevation, aspect, land-use, profile curvature, plane curvature, lithology, distance to streams, distance to roads, distance to fault, and NDVI. Different patterns of precipitation are used to analyze the factors of cause to extreme natural phenomena. We found that because of the different environmental characteristics in different regions, the weights of the variables of mass wasting are also different. However, NDVI and slope are the most important variables regardless of spatial scales and characteristics of environment. In this study we use three social exposure indexes and eight vulnerability indexes characteristics to assess residents’ vulnerability form place scale to country scale in our study area. Besides, we used perceptions and adjustment indexes to assess the capability of adjustment for scales of individual to place. We found that county-scale assessment was difficult to assess the vulnerability of place (village) scale accurately. For risk assessment, we produce a risk value ranging from 0 to 1, which is a standardized value of summating mass wasting sensitivity, exposure of residents and vulnerability of residents. Our results suggest that large scale assessment obscured the characteristics of smaller scales. For the township scale, the most vulnerable one is Hai-Duag Township. In fact, the risk isalso very high in Da-Niau village of Da-Wu township, but Da-Wu township area was too small and easily equalized by large scale assessment. Furthermore, we found that when the standards of dividing boundaries for spatial scales are the same, their risks are very similar, and vice versa. Thus, in order to explore the causes of disasters, we suggested that researchers should pay attention to spatial scales and characteristics of regions. This paper is a re-interpretation of the long-standing “human-environment or nature-society” research in geography. A geographic information system was utilized to present spatial and temporal variations for regional disaster system. The conceptualization of regional disaster system provides a powerful selling point for the salience of the discipline to public policy, urban planning and disaster management and other related disciplines.