山區森林影像地形校正:MVECA法的嘗試
Abstract
在山區衛星影像的應用上，針對地形起伏所造成的地形效應進行校正為重要的前處理步驟，其足以影響後續影像分類的品質以及量化的精度。臺灣山地占土地面積的三分之二以上，山地地區的森林光譜輻射強度值受地形效應影響甚大，如能對植被影像進行地形校正，則有助於後續研究量化精度的提升。為克服改進前人校正模式的缺點，這篇研究將提出一種新的參考操作模式，基於VECA(Variable Empirical Coefficient Algorithm)以及Minnaert+SCS修正設想的 Modification of Variable Empirical Coefficient Algorithm(MVECA)修正模式。
本研究以簡單統計評估、不同坡度坡向下的NDVI離散係數分析、分類評估等，對MVECA模式進行完整的效能檢驗。結果顯示其在無光照陰暗區以及低光照地區較前人模式具有較好的校正效能，與光照區的視覺差異、離散差異皆有所降低；與相對入射光cos(i)的R值除藍光段外，皆呈現極低度相關，顯示受地形效應影響大幅降低；在方位180°~225°、坡度0°~5°條件下的NDVI離散係數最低校正，效果最好；校正區域135°~180°方位的整體離散係數較其他方位都較低，尤其25°~50°坡度條件下，具有極佳校正效能；校正效能最差的地形位於坡度50°~55°、方位225°~270°條件下，以及方位315°~360°、坡度5°~15°的地形條件區域。研究最後以整合導向為目的，以最大相似分類法對研究區進行植被與裸露地二分類，結果顯示經校正後的影像分類較未校正影像有更高的分類精確度，完全陰暗區經校正後的影像分類較未校正影像有更高的分類精確度，證實本研究所提出的MVECA法具備地形校正的能力。
以MVECA法對山地植被進行地形校正，發現NDVI與分類精度皆較校正前的影像有明顯改善，顯示該法對於森林生態系的監測可提供助益，包括對於陰暗區植被參數的推估與山區森林長時間季節變化的監測提供研究操作的參考。
Topographic correction is a very important approach of pre-process in the remote sensing application in rugged terrain. Without this step, the quantification and classification of the image will be influenced. A new topographic correction approach, the Modification of Variable Empirical Coefficient Algorithm (MVECA), was developed on the theoretical and statistic analysis of the spectral values of remotely sensed data acquired from the rugged terrain and topographic variables. To prove the approach is practical, The approach was explained and analyzed in comparison with that before topographic correction and other approaches. The test site selected for the study is located on the relatively rugged terrain over the southeastern hill of the Yu Shan Mountain in Taiwan, the remotely sensed data utilized for examining the result of the proposed approach MVECA is Formosat2 image acquired over the test site. Visual comparison, statistical analysis, and the classification of corrected image are adopted for feasible evaluation of the proposed algorithm, and the results suggest that the proposed approach MVECA is capable for removing the topographic effects contained in Formosat2 MS image. The result also shows that MVECA has better ability of topographic correction than SCS+C and Minnaert+SCS in the area of the shadowed area and the steep terrain where incidesnt angles may approach 90°.
Topographic correction is a very important approach of pre-process in the remote sensing application in rugged terrain. Without this step, the quantification and classification of the image will be influenced. A new topographic correction approach, the Modification of Variable Empirical Coefficient Algorithm (MVECA), was developed on the theoretical and statistic analysis of the spectral values of remotely sensed data acquired from the rugged terrain and topographic variables. To prove the approach is practical, The approach was explained and analyzed in comparison with that before topographic correction and other approaches. The test site selected for the study is located on the relatively rugged terrain over the southeastern hill of the Yu Shan Mountain in Taiwan, the remotely sensed data utilized for examining the result of the proposed approach MVECA is Formosat2 image acquired over the test site. Visual comparison, statistical analysis, and the classification of corrected image are adopted for feasible evaluation of the proposed algorithm, and the results suggest that the proposed approach MVECA is capable for removing the topographic effects contained in Formosat2 MS image. The result also shows that MVECA has better ability of topographic correction than SCS+C and Minnaert+SCS in the area of the shadowed area and the steep terrain where incidesnt angles may approach 90°.
Description
Keywords
山地地區, 衛星影像, 地形效應, 地形校正, MVECA, Topographic correction, FORMOSAT2 image, mountain forest, MVECA