棲蘭野生動物重要棲息環境之脆弱度評估

Assessment of Environmental Vulnerability in Major Wildlife Habitats in Chilan, Taiwan

本研究整合遙測及地理資訊系統（geographic information system）技術，並搭配模糊邏輯（fuzzy logic）理論與主成分分析（principal component analysis, PCA），進行棲蘭野生動物重要棲息環境之脆弱度評估。首先，由 2001~2011 年地景變遷分析獲取植生退化區位，將其視為脆弱環境樣本，隨後與 13 項環境變數進行相對頻度分析並建立模糊隸屬函數，再以 PCA 縮減變數及分配各成分權重，藉以建立分析模式，推估脆弱度之空間分布。研究結果顯示，研究區有 12% 的地景受到改變，其中植生退化的脆弱環境占 4.3%，根據模式推估，區內脆弱度可分為輕微、次輕微、普通、次嚴重及嚴重等 5 個等級，嚴重等級主要分布於中高海拔、坡度陡峭且鄰近水系或崩塌地兩側之脆弱區位；輕微等級者，則多位處坡度平緩之中低海拔區位，植被穩定且遠離脆弱環境，抵禦干擾能力相對較佳。 Vulnerability of major wildlife habitats in Chilan was analyzed using remote sensing, geographic information system, fuzzy logic theory, and principal component analysis (PCA). First, we obtained samples of vulnerability based on landscape change analysis from 2001 to 2011. Fuzzy membership functions were then defined for 13 environment factors using the relative frequency of vulnerability. We developed a vulnerability evaluation model through PCA, which reduced the number of environment factors and determined the weight of principal components. The results demonstrate that 12% of the landscape is dynamic and includes the degraded vegetation zone (4.3%). Vulnerability was classified into five levels: slight, light, medial, heavy and extreme level. According to spatial distribution of vulnerability, most areas classified as extreme level are located in relatively high altitude, steep terrain and vulnerable surrounding environment. The preferential place of slight level was found at low altitude, flat terrain, great anti-interference ability and areas largely covered by vegetation.

Vulnerability of major wildlife habitats in Chilan was analyzed using remote sensing, geographic information system, fuzzy logic theory, and principal component analysis (PCA). First, we obtained samples of vulnerability based on landscape change analysis from 2001 to 2011. Fuzzy membership functions were then defined for 13 environment factors using the relative frequency of vulnerability. We developed a vulnerability evaluation model through PCA, which reduced the number of environment factors and determined the weight of principal components. The results demonstrate that 12% of the landscape is dynamic and includes the degraded vegetation zone (4.3%). Vulnerability was classified into five levels: slight, light, medial, heavy and extreme level. According to spatial distribution of vulnerability, most areas classified as extreme level are located in relatively high altitude, steep terrain and vulnerable surrounding environment. The preferential place of slight level was found at low altitude, flat terrain, great anti-interference ability and areas largely covered by vegetation.