Vegetation recovery is an important process following disturbance like harvest or fire. In this research, we looked at how vegetation recovery could be detected and defined using multispectral time series data. We used four commonly computed spectral indices: Normalized Difference Vegetation Index (NDVI), Normalized Burn Ratio (NBR), Tasselled Cap Greenness (TCG), and the shortwave infrared Band 5 (B5) of Landsat 5 and 7. These indices were calculated annually for a time series of Landsat satellite imagery acquired from 1985-2010 for a sample of path-rows across the boreal forest of Canada. Recovery was definde as 80% of the mean of the two pre-disturbance values for each index. The years-to-recover was calculated as the time required for the disturbed pixel to attain the recovery definition.
Nearly all pixels showed some degree of recovery, between 93% (NBR) and 99% (NDVI) of pixels were considered spectrally recovered after 10 years. However, there were still some differences among bioclimatic zones and indices. NDVI and TCG showed the fastest recoery rates in terms of years-to-recover, usually less than 2 years for NDVI, TCG, and B5, and about 4 years for NBR.
Read the article in International Journal of Remote Sensing.