福山地區SPOT多期影像植生綠度分析

An Analysis of Vegetation Greenness of Fu-shan with SPOT XS Multitemporal Data

謝漢欽(Han-Ching Hsieh) ； 鄭祈全(Chi-Chuan Cheng)

《林業試驗所研究報告季刊》 10卷1期 (1995/03) Pp. 105-120

https://doi.org/10.7075/BTFRI.199503.0105

摘要

本研究之目的在於利用1986年至1990年福山地區四個不同時期之SPOT XS衛星影像，進行各期影像之常態化差異植生指數分析，據以評估綠色植生量的變遷。於影像前處理時，特別採用衛星影像對數值地形起伏影像之簡易有效的幾何糾正程序，使得精度在次像元範圍內。應用本研究之輻射常態化處理程序，分別去除了影響接收反射輻射值之路徑輻射、天空光效應及地形陰影。並引用影像分類程序分析常態化之效果，從分析中得知四期影像的反射輻值可由已常態化區域推算，該區之常態化差異植生指數指數值可代表各期影像的綠度。由綠度分析結果得知，四期影像的常態化差異植生指數值介於0.7至0.8之間差異顯著，皆比未經常態化處理者平均提高了0.3，顯見常態化去除了陰影之效果。依四期影像之常態化差異植生指數植之評斷，福山地區天然闊葉林相有所變化，其常態化差異植生指數值受到衛星掃描偏角及不同時期植生生理狀況之影響。

關鍵字

多期影像 ； 綠度 ； 常態化差異植生指數 ； 幾何糾正 ； 變遷分析

並列摘要

The objective of this study was to use normalized difference vegetation index (NDVI) detecting the vegetation changes in the Fu-Shan district of Taiwan Forestry Research Institute with four multitemporal SPOT XS images, gathered on different dates in the time from 1986 to 1990. Before the greeness analysis of these four date images, the effective image preprocessing proccdures such as improved geometric correction and simple radiation normalization methods were applied. In this study, utilizing the SPOT image to DTM relief image geometric correction method to rectify SPOT images into the same map projection had been proved an easy and useful way. The average RMS accuracy of each corrected image was below one pixel size. The path radiance, skylight rediance and topographic effects were relieved through these image radiation normalization procedures. An images classification procedure also were applied to separating the inharmonious spectral pixels into three main classes to determine which portions of normalized images were normalized reliably.
Based on the reliable normalized classes of there four date images, the NDVI and radiation calculated from each of them could be used to evaluate the vegetation greenness on the entire canopy in Fu-Shan district. These NDVIs of these four multitemporal images were shown to have significant differences from the results of greenness analysis. Compared with the unnormalized images, the average NDVI of each normalized image was on average 0.3 higher than that of the un-normalized, which revealed that the topographic shadow effects had been reduced. According to the NDVI of these four date images, there were some difference among them, and these NDVIs might be affected by the satellite scan attitudes and the seasonal changes of living vegetation condition.

並列關鍵字

Multitemporal images ； geometric correction ； change detection ； greenness ； normalized difference vegetation index