Analysis of trends and changes in the successional trajectories of tropical forest using the Landsat NDVI time series

The vegetative growth of forest canopies changes their spectral response, which can be detected by multispectral sensors and enhanced by utilizing the normalized difference vegetation index (NDVI). The structural variability of canopies in heterogeneous forests can also be related to successional stages. Thereby, a spatiotemporal methodology is presented to associate the 3D photogrammetric information, derived from aerial images, with the NDVI time series extracted from the Landsat imagery, using spatial units connected to ecological succession. The technique is based on time series clustering within superpixels (extracted from the local variance of tree heights) and on trend analysis of the canopy using the breaks for additive season and trend (BFAST) algorithm. The study was conducted in a tropical native forest remnant (Inland Atlantic Forest) in the western region of Sao Paulo State, Brazil. We investigated the relationship between the variability of the forest vertical structure and the NDVI temporal trajectory associated with vegetation vigor in the period from 1984 to 2010. The experiments produced a regularized time series of overlaid superpixels. The cluster statistical analysis was successful in separating the NDVI trajectory into ten classes associated to successional stages, in which the evolution of the vegetation vigor could be observed and compared within degraded, transitional, and preserved areas. Such areas overlapping with the ten trajectory classes allowed us to quantify that 59.9% of the clusters were related to preserved areas, 30.1% to transitional areas, and 10.0% to degraded areas. Additionally, the BFAST algorithm allowed the identification of trend measures and the occurrence of disturbance events within these trajectory classes based on the vegetation response. Therefore, the classification of NDVI trajectory clusters helps to understand how the areas in a heterogeneous forest with different succession stages develop spectrally. Furthermore, monitoring secondary forests is essential for the conservation of biodiversity, ecosystem functioning, and carbon stock, among other values.