整合地理資訊系統與遙測技術於林分材積估測之研究

Integration of GIS and Remote Sensing in Estimating Stand Volume

鄭祈全(Chi-Chuan Cheng) ； 許立達(Li-Ta Hsu) ； 陳燕章(Yen-Chang Chen)

《台灣林業科學》 13卷2期 (1998/06) Pp. 155-167

https://doi.org/10.7075/TJFS.199806.0155

摘要

本研究使用林試所六龜試驗林1996年l月及6月SPOT衛星遙測影像，配合數值地形資料及地面樣區調查資料，以多元逐步迴歸及因素分析方法求取材積預測迴歸式，以研究衛星光譜資料與林分材積的關係，並探討應用光譜資料推估林分材積的可行性。其他研究目的尚包括(1)比較不同的衛星資料處理方法（未修正、去除路徑輻射、與地形效應修正）在材積估測上之效用，(2)比較不同季節光譜資料在材積估測上之適用性，以及(3)評估林齡分級與林地分級對於林分材積估測模式建立的影響。
研究結果顯示材積與光譜資料相關性並不高，須佐以地面調查資料及地形資料始能提高其材積估測能力。衛星資料經地形效應修正後，在材積估測上能獲致較佳結果。不同季節的光譜資料差異甚大，因此雖然不同季節光譜資料對於材積預力大致相若，但不同季節必須分別建立適用於該期光譜資料的材積式，依林齡分級或依地形環境分級，雖可能局部提高林分材積估測之準確度、但整體而言，並未能有效改善應用光譜資料估測林分材積之實用性。

關鍵字

遙測 ； 地理資訊系統 ； 林分材積 ； 逐步迴歸 ； 因素分析

並列摘要

The purposes of this study were to explore the relationships between remote sensing data and ground stand volumes, and to evaluate the possibility of estimating stand volumes based on remote sensing data. Stepwise regression and factor analysis were used to derive prediction equations of stand volumes based on 2-season (January and June 1996) SPOT images, topographic data, and ground inventory data of the Liukuei area. Other objectives of this study were (1) to compare the effects of different remote sensing data processing methods (no treatment, removal of path radiation and topographic effects) on estimation of stand volumes; (2) to compare the applicability of using seasonal remote sensing data for estimation of stand volumes; and (3) to evaluate whether the use of classification of both stand age and topographic data would improve the accuracy for estimation of stand volumes.
Results of this study showed that stand volumes and remote sensing images were not closely correlated. Therefore, in addition to remote sensing images, ground inventory data and topographic data should be included to adjust the estimated stand volumes. Removal of topographic effects from remote sensing images produced better estimated stand volumes than did either those without treatment or those with removal of path radiation. Significant seasonal changes appeared in remote sensing images, thus different seasons required their own regression equations when developing estimation models for stand volumes. Classing stands based on their ages or clustering forest land based on topographic conditions improved the accuracy of volume estimation in some classes, however, it did not effectively promote the applicability of using remote sensing images for estimation of stand volumes.

並列關鍵字

remote sensing ； geographic information system ； stand volume ； stepwise regression ； factor analysis