Correlation analysis between vegetation fraction and vegetation indices in reclaimed forest: A case study in Pingshuo mining area

摘要

Long-term vegetation monitoring is important for the assessment of the impacts of mining and reclamation activities in an area like the Antaibao, where extensive and rapid opencast mining occur continuously. Accurate information of vegetation fraction is important for assessing how mining activities affect the ecosystem in mining areas. A remote sensing method based on vegetation cover monitoring and assessment by using Landsat data sets with the temporal coverage from 1990 to 2010 was presented and applied to the Antaibao opencast mine, Shanxi Province, China. To reduce the impact of phenology on remote sensing based vegetation monitoring, 7 Landsat data were selected during vegetation growing season (June-September). A total of 32 pixels were selected as the survey fields during vegetation growth season in August 2010. Vegetation fraction (VF) of arbores canopy in the selected plots was computed. We monitored reclaimed vegetation variation through analyzing time series data of 19 vegetation indices, which were usually applied to mine area monitoring, such as NDVI, EVI. Principal component analysis (PCA) of the VIs showed that PC1 and PC2 accumulatively accounted for 91.51% of VIs variance. The Regression analysis demonstrated that PC1 and PC2 were highly correlated with the measured vegetation fraction, with R2 of 0.74. In addition, we concluded that the vegetation system of South Dump, West Dump and West Enlarged Dump had gradually transformed to the natural ecosystem from 1990-2010. The vegetation in Inner Dump and Inner Enlarged Dump should be well restored. Reclamation was a long-time and dynamic project, whose effect is hard to determine in short term, while remote sensing technology is an effective way to monitor the inter-annual variation of vegetation recovery.