LAND COVER MAPPING IN THE BRAZILIAN PAMPA WITH LANDSAT OLI AND TIRS BANDS

Mapeamento da cobertura de terra no Pampa Brasileiro com LANDSAT OLI e BANDA TIRS

Abstract

When different time periods are considered, detection of past and present changes in land cover are enabled, also for quantifying and qualifying those changes. Land cover/use maps are the primary tools for the management and conservation of natural and man-made areas. For this, remote sensing bands of the reflected spectrum are usually used, leaving aside the thermal data. The objective of this work was to evaluate the inclusion of the thermal band (b10) of the TIRS (Thermal Infrared Sensor) sensor of landsat 8 satellite to increase the land cover maps accuracy in the Pampa biome from object-oriented classification. For the development of the research, 11 scenes of the Landsat 8, OLI sensor and TIRS were used. Thus, 14 cells were selected in the Brazilian Pampa, totaling 5% of its area. The following steps were performed: obtaining land surface temperature (LST) data and vegetation indices; data preparation; object-oriented classification; validation with 1354 reference points and analysis of the results. The results showed that the insertion of thermal bands, especially from different dates, increased the discrimination among classes. The classification presented 86% of global accuracy. Therefore, it is recommended to insert thermal data for mapping and environmental monitoring of the Pampa biome