<紀要論文>
ランドサットTMデータにおける地形による影響の補正方法の比較 : 九州大学宮崎演習林内のスギ林分を対象として

作成者
本文言語
出版者
発行日
収録物名
開始ページ
終了ページ
出版タイプ
アクセス権
JaLC DOI
概要 九州大学宮崎演習林内にある林冠の閉鎖したスギ林分(85.3ha,1050ピクセル)を対象として,ランドサットTMデータにおける地形による影響の補正方法について検討した.補正方法としてDEMの使用を前提とした既存の4手法(cosine法,Minnaert法,C法,Statistic-empirical法)について,その補正効果を比較した.斜面方位別の解析の結果,バンド4,5の輝度に大きな変化がみられ...たため,特にバンド4,5について解析を行った.オリジナルデータと太陽入射角の余弦(cos i)には正の相関がみられたが,Minnaert法,C法,statistic-Mempirical法による補正後,そうした相関はみられなくなった.一方,cosine法ではcos iが小さいときに過補正となり,逆に負の相関がみられた.Scheffe検定の結果,オリジナルデータでみられていた向日および背日の斜面方位クラス間での有意差が,Minnaert法,C法,Statistic-empirical法ではみられなくなった.また,補正後のデータ全体での標準偏差は,cosine法を除く3手法において縮小した.結局,cosine法を除く3手法に有意な補正効果が認められ,それらの3者間の差はほとんどみられなかった.
Spectral variability due to topography can create significant problems in analyzing remotely sensed data for areas of mountainous terrain. The effects of topography on the radiometric properties of Landsat TM data are examined in the context of the remote sensing of forests in mountainous regions. The purpose of this study was to compare four radiometric correction methods for the reduction of slope-aspect effects in a Landsat TM data. Data obtained from canopy closed sugi stands (Cryptomeria japonica D.Don; 85.3 ha, 1050 pixels) were used for this study, which are located on the Kyushu University forest in Miyazaki. The correction methods selected for this study are DEM based model; cosine correction, Minnaert correction, C-correction, statistic-empirical correction. As a result of aspect analysis, there are sharp fluctuations in digital number of band 4 and 5, which sensed reflective near- and mid- infrared spectral. So correction methods were applied on the data of band 4 and 5, then their results were compared among correction methods. There were positive correlation between the original data and cosine of solar incident angle (cos i). The correlation was not found after Minnaert, C and statistic corrections. But when the cosine correction, for the smaller cos i , the stronger overcorrection appeared, then negative correlation was found. The Scheffe test were examined among aspect classes. There is significant difference on original data, but insignificant difference on Minnaert, C and statistic approaches. The standard deviations of three corrected data without cosine correction were reduced. Consequently, cosine correction showed insignificant result, but Minnaert, C and statistic corrections were almost same significant for the reduction of slope-aspect effects.
続きを見る

本文ファイル

p013 pdf 794 KB 864  

詳細

PISSN
NCID
レコードID
査読有無
主題
タイプ
登録日 2009.04.22
更新日 2021.03.03

この資料を見た人はこんな資料も見ています