Análise comparativa entre métodos de correção atmosférica de imagem Landsat 8 para a extração da Temperatura de Superfície e do Índice de Vegetação por Diferença Normalizada

Favoritar este trabalho

. This paper aims to determine, for a Landsat 8 scene, values of NDVI (Normalized Difference Vegetation Index) and LST (Land Surface Temperature) derived from different products. The NDVI and LST were calculated using the Top-of-Atmosphere reflectance (ToA) and the atmospherically corrected surface reflectance by the DOS1 method. As reference, a surface reflectance value provided by the USGS (United States Geological Survey) was adopted. In order to generate the products, the "Land Surface Temperature" and "Semi-Automatic Classification" plug-ins installed in the software QGIS were used. We used the bands 4 and 5 from OLI sensor for the NDVI and band 10 from sensor TIRS for the LST. Fifty points were chosen randomly throughout the scene to store the values of the pixels in order to be compared with each other. When comparing the ToA and DOS1 results with the surface reflectance values provided by the USGS, the determination coefficient (r²) was equal to: a) 0.9647 for the NDVI and 0.9891 for the LST when estimated with the reflectance at the Top of the Atmosphere (ToA), and; b) 0; 9901 for NDVI and 0.9986 for LST obtained with the product atmosphericly corrected by DOS1. In reason of the influence from the different methods applied, we recommend that future works use the methods described here alongside data containing different atmospheric conditions.