利用SPOT衛星影像探討阿里山事業區林型光譜特徵

Analyzing Spectral Characteristic of Forest Stands in Alishan Working Circle Using SPOT Imageries

陳建璋(Jan-Chang Chen) ； 吳守從(Shou-Tsung Wu) ； 陳建文(Chien-Wen Chen) ； 陳朝圳(Chaur-Tzuhn Chen)

《作物、環境與生物資訊》 6卷2期 (2009/06) Pp. 124-133

https://doi.org/10.30061/CEB.200906.0005

摘要

本研究主要以阿里山事業區內六個林型（1.天然闊葉林型、2.天然混淆林型、3.人工針葉林型、4.人工闊葉林型、5.人工混淆林型、6.竹林）為試驗區域，選取2005年四幅SPOT衛星影像，藉由植生指標的計算探討六林型之光譜特徵。研究結果顯示，在不同林型光譜特徵的萃取上，植生指標之計算可協助區分不同林型之光譜特徵。在四個時期的SPOT影像中，不同植生指標的比較亦可清楚的將六個林型之光譜特性區隔出來。本研究透過衛星影像監督性分類之結果顯示，六林型之分類準確率可達到78.44%。

關鍵字

遙感探測 ； 反射光譜 ； 植生指標 ； 影像分類

並列摘要

The objectives of this study were to analyze the spectral characteristics (SCs) of different composite forest stands in Alishan working circle by using four satellite images taken from SPOT in 2005. The vegetation indices were calculated from spectral data of SPOT imageries. The composite forest stands were classified into six categories, namely, natural hardwood forest, natural mixed forest, softwood plantations forest, hardwood plantations forest, plantations mixed forest and bamboo forest. To extract SCs from SPOT imageries, a processing scheme was developed, in which the method of differential sampling timings of SPOT images and spatial distribution of different forest stands were included. The results indicated that different vegetation indices gave vary degrees of accuracy, and may be used to separate the spectral characteristics of forest stands. The overall accuracy for the supervised classification was 78.44%.

並列關鍵字

Remote sensing ； Spectral reflectance ； Vegetation index ； Classification