LAUSR

Vegetation indexes from remote sensing images are often used for land-cover monitoring and identification of biomass changes. They are also very useful to describe the relationships between the phenological cycle and the carbon sequestration, which are climate change indicators. The Caatinga land-cover is very heterogeneous, making hard the understanding of the land cover processes in different scales (spatial and temporal), due to seasonalities and human actions. The Landsat series products usually can describe spatial land-cover variations, with a low temporal scale, so far. This study aims to improve the temporal representation of the land cover by Landsat images for a Caatinga area. This article presents an evaluation, using a mathematical approach of three-parameter functions to describe the Normalized Difference Vegetation Index (NDVI). Each function performance was evaluated by the reduced chi-square (χ²) and determination coefficient (R 2 ) parameters. The analysis is performed for an annual period and considers land-cover changes by anthropic action. The Cauchy function seems the best option and presented an adjust up to 83% of the total (years and places), with an R² (average) of 0.82. The parameters of this function can be a valuable source for environmental studies in the Caatinga biome, supporting temporal analysis of the vegetation.