本研究旨在應用航測技術和多變值統計方法，由時空層面建立森林地景之干擾指標，以供監測林業試驗所蓮華池試驗林之民墾地干擾效應之用。研究過程中主要考慮民墾地干擾所造成的地景變遷，先數化1923年和1998年二期的土地利用圖，並量化與地景干擾有關的8個地景指數，接著以多變值統計方法檢定二期地景結構的差異，以判別民墾地在森林地景的影響效應。此外，在研究中並分析單一地景指數的變化，以二項式logit迴歸分析進一步篩選出受影響較大的地景指數並據此建立干擾指標。研究結果指出，由地景指數的多變值統計分析顯示，1923年到1998的地景結構的確受到民墾地的影響。由迴歸分析可知，8個常用的地景指數，以邊緣密度(ED)和Shannon均勻度指數(SHEI)二者受到干擾較大，本研究同時計算ED和SHEI的影響權重，建立一綜合之干擾指標。經由檢驗地景單元受到干擾與否的結果顯示，該指標在評估地景干擾效應的準確度高達92.9%。因此，本研究發展的干擾指標可供監測蓮華池試驗林森林地景的變遷和評估干擾效應之用。
This study was to develop a disturbance index for monitoring landscape change under both spatial and temporal perspectives. The Lian-Haw-Chih Experimental Forest, located in the central part of Taiwan, was selected due to its wide and long-term disturbances through unlawful cultivated forestlands. The processes of quantified landscape indices and multivariate methods were used to discriminate the spatial variations of landscape structure between 1923 and 1998 and binomial logit regression analysis was applied to develop the model of disturbance index. The results showed that landscape structure was affected by unlawful cultivated forestlands according to spatial discrimination using quantified landscape indices and multivariate methods. Moreover, the model of disturbance index, mainly composed of edge density and Shannon's evenness indices, displayed an accuracy of 92.9% in assessing the model. We concluded that the model of disturbance index developed in this study was feasible from spatial and temporal perspective, and it can be used for monitoring landscape change and assessing the disturbance effect.
