Aerial photography for post-mine vegetation inventory

Abstract

The objective of this study was to investigate the application of 70 mm aerial photography in the evaluation of revegetated strip mined lands. Large (1:600) and medium (1:10 000) scale 70 mm color and color infrared (CIR) aerial photography was acquired in July 1980 of reclaimed coal-mined lands in north western Colorado. A concurrent field survey of vegetation cover was also carried out. Plant species identification keys were developed from the large scale photography with the aid of intensive ground marking, and were tested by six photo interpreters. Crested wheatgrass (Agropyron cristatum), intermediate wheatgrass (A. intermedium) and smooth bromegrass (Bromus inermis), the dominant grass species present, were identified with varying levels of accuracy (70, 69 and 57% respectively) by the photo interpreters. Alfalfa (Medicago sativa), yellow sweet clover (Meli lotus officinale) and thistles (Cirsium spp.) were also accurately identified (90, 100 and 75% respectively). Several species of shrub seedlings present could not be readily identified. A non-parametric statistical evaluation (Friedman index) of individual photo Interpreter results indicated that, for the skill levels available for this study, there were no significant differences (p >0.05) based on Interpreter experience with the vegetation, study area or photographic interpretation technique. Estimates of vegetation and other categories of ground cover were made using a point line method on the aerial photographs and compared to corresponding point frame cover estimates made on the ground. Regression analysis indicated that for the dominant species present, only alfalfa cover could be reliably estimated (r2 = 0.85; p <0.01) from the aerial photographs. Cover of the various dominant grass species could not be reliably estimated. Comparisons by ground cover category showed that total for (r2 = 0.85), total vegetation (r2 = 0.71) and total ground cover (r2 = 0.66) categories were accurately estimated (p <0.01). Grass (r2 = 0.58) and rock cover (r2 = 0.52) estimates were less accurate (although p <0.01) and litter cover (r2 = 0.01) could not be accurately estimated from the large scale CIR photographs.