林型分佈製圖為人工林經營管理提供關鍵性的訊息,經由航照判釋及配合地面調查來進行林型製圖,耗費許多人力及時間,因此需要發展一套較自動化的林型製圖方法。本研究探討空載LiDAR資料於高山地形對於林型分類之可行性,選定阿里山地區紅檜、柳杉、闊葉混淆林三種林型和裸露地進行分類,分類方法係利用LiDAR強度和多重回波資料進行分類,採用最大概似法對於LiDAR單一影像、LiDAR套疊影像、LiDAR主軸轉換影像進行林型圖之繪製,研究結果顯示LiDAR主軸轉換影像可達到最高85.5%分類總體精確度,總體Kappa值達0.81,此項技術對於林型圖之繪製具有高度可用性。
Mapping the distribution of forest type provides critical information for plantation management. Forest types mapping through with the photogrammetry that was interpreted from the aerial photograph and with ground survey, it requires a lot of technical manpower and is time consuming. Therefore, it is necessary to develop more automatic methods for forest type mapping. This study evaluates the possibilities of airborne LiDAR data for automatic mapping forest types in mountain areas. Three kinds of forest types (Red cypress, Sugi, and mixed hardwood) and bare land are chosen for classification analysis in Alishan, the investigated area. The classification methods are based on LiDAR multiple return and intensity data. In this study, we used LiDAR single image, LiDAR stack image and LiDAR principal component transformation image for Maximum Likelihood Classifier (MLC) to map the forest types. The highest overall classification accuracy reaches 85.5% by LiDAR principal component image, and the overall Kappa statistics reaches 0.81. This result shows the high feasibility for forest type mapping.