台灣地區經常受到許多天災所影響,所影響的不僅是地形也包括了地形資料,而地形資料在天然災害後,往往缺乏即時有效的更新。對於天然災害後的地形改變,我們無法立即清楚知道在天然災害後所改變的植生分佈狀況。傳統上,植生分佈資訊皆以人為方式到現地調查獲取資料。有鑑於此,本研究使用「最佳統計分類法」分類出植生分佈區域,藉由遙測技術降低人力上的需求,加上「K-means分群法」之分類成果來評估最佳統計分類法的可行性,再搭配常態化差異植生指標(NDVI)得知植生數值分佈情況,從準確的分類區域並配合植生數值來探討植生分佈的情形。本研究區域為南投縣竹山地區,其衛星影像拍攝年份分別為1999年、2002年及2007年,使用最佳統計分類法的植生分類結果顯示NDVI指標數值分類在樹林平均數值約為0.4、分類在農地平均數值約為0.2。最佳統計分類法的有效分類配合NDVI指標,可清楚瞭解南投竹山地區植生分佈區域情形。
Taiwan area is usually struck by natural perils. Not only the terrain but also the geographic data are significantly influenced. The terrain data often lack of effective and instant update after one natural peril occurs. Toward the landform change after natural perils, we cannot immediately identify the exact condition in respect of the vegetated cover. The information of vegetated cover is traditionally obtained through personal site survey. As a result, this study utilized the optimum statistical classifiers method to classify the vegetation area, and the study can decrease the manpower demand through remote-sensing technology. Furthermore, the k-means clustering method is employed to appraise the feasibility of the optimum statistical classifiers method, as well as NDVI to derive the figures of vegetation. The classified area together with vegetation figures could therefore be utilized to discuss the condition of vegetated cover. The location chosen for this study is in the Tzu-Shan area of Nantou County; satellite images were respectively taken in 1999, 2002 and 2007. The result of vegetation classification through the optimum statistical classifiers method showed the average value of NDVI in woods was around 0.4, and that of farmland was around 0.2. By the optimum statistical classifiers method in collocation with NDVI indexes, researchers can clearly realize the condition of vegetation cover in the Tzu-Shan area of Nantou County.