Image modelling of forest changes associated with acid mine drainage
A planar cellular transition model was investigated as a predictive methodology for modeling tree canopy dynamics along a chemical contamination gradient adjacent to an acid mine. Stochastic raster transitions were developed as a function of both proximity to the tailings and local stem density neighborhood con®guration. Model inputs consisted of scanned and registered archival photographs that were segmented into three forest classes using two dierent apex delineation techniques, one incorporating double-aspect calculations and the other using a high- pass ®lter. The results for each were integrated over selected sample windows. The accuracy of the two techniques was compared using a synthetic data set. The double-aspect technique was better able to identify and delineate tree crowns in synthetic data with signi®cant noise and shadow representative of typical forest conditions. However, the window integration results were more visually interpretable when classes were derived from the high-pass technique. Application of these techniques to the archival photography and subsequent temporal analysis using the Mann± Krendall statistic revealed signi®cant forest degradation over a 40-year period nearer the tailings and within areas of mixed canopy types. 5 1999 Elsevier Science Ltd. All rights reserved.
Walsworth, N.A. and King, D.J., "Image modelling of forest changes associated with acid mine drainage" (1999). Aspen Bibliography. Paper 948.