台大實驗林區不同植生型之光譜資訊分析

本研究為配合國土保育計畫之推動，針對不同林業用地常見植生型進行光譜資訊分析，將進行台大實驗林區之造林樹種如柳杉、台灣杉、檜木以及竹林、果園、檳榔園、茶園及高山蔬菜地等山坡地區常見作物類型的光譜資訊分析。利用台大實驗林區1994-2005年的41幅衛星影像的資料進行植生指數分析(NDVI)與雙光譜圖分析。由於影像受大氣效應、光照條件等非地表因子及植物物候的影響導致光譜變異，因此使用影像中的原始DN值、校正後所得輻射量值以及相對性輻射糾正處理三者進行光譜分析，觀察NDVI值變化趨勢之依據，明顯看出孟宗竹林、柳杉、桃李梅果園、檳榔園、茶園及高山蔬菜地皆受季節改變所影響。同時透過光譜變異數分析，發現經直方圖匹配後所得NDVI值較直接以原始DN值或輻射量值計算之NDVI值能區分出不同季節與不同植被。分析植被之變化時，原始DN值之光譜變異須藉絕對校正參數值校正或相對性輻射糾正消除，而以相對性輻射糾正處理的直方圖匹配法效果較佳，能確實瞭解植被之光譜變遷情形。

關鍵字

植生指數；雙光譜圖；直方圖匹配

並列摘要

The aims of this study were to interpret varied spectrum characters of SPOT images by using Normal Difference Vegetation Index (NDVI) and bi-spectrum plot analyses to identify different land use types of agriculture and forestry fields in the NTU Experimental Forest. We chose forty-one SPOT images between 1994 and 2005. Since the spectrum was influenced on some non-surface effects, such as atmosphere, illumination and the change of plant phenology, NDVI value of the spectrum analysis was dependent on the Digital Number (DN), the value of Radiometric Correction and Relative Radiometric Normalization. The results carried out that different land use type were obviously affected by seasonal change. Furthermore, ANOVA in this study indicated that the NDVI value produced by histogram hatch was much better to distinguish different season and vegetation than the NDVI value produced by DN or radiometric value. To eliminate the spectrum variation of DN, using radiometric correction and relative radiometric normalization is suggested when analyzing the changes of vegetations. Thus, the results indicated that the method of histogram patch deserves the best way to clearly identify the variation of vegetation spectrum.

並列關鍵字

Normalized difference vegetation index ； Bi-spectrum plot ； Histogram Match

國際替代計量

台大實驗林區不同植生型之光譜資訊分析

全文下載

主題瀏覽