Aplicação da plataforma Land Viewer para avaliação de área agrícola afetada por enchente do rio Quaraí

Abstract

Information from remote satellite sensors, combined in the form of vegetation indices, is increasingly being used to support agricultural management. The Land Viewer platform offers products derived from the processing of satellite images, such as vegetation indices or color combinations of satellite bands, for up to ten free images per day, in medium resolution, from the Landsat 8, Sentinel 2 and CBERS4 satellites, enabling in this way to obtain significant correlations between spectral data from sensors and information from the land surface obtained from agricultural and environmental processes in specific areas. The same information can be obtained through the geoprocessing of images made available directly by the space agencies, free of charge, without limitations in number, using Geographic Information Systems such as INPE's SPRING. In this sense, the present work aims to: 1) compare the procedure of using the Land Viewer tool and the direct geoprocessing of images, for the case study of an agricultural area in the municipality of Barra do Quaraí, affected by the flood of the Quaraí River, on the Frontier West of Rio Grande do Sul, evaluating the dynamics of floods within the property, through photointerpretation of combinations of images (RGB and NDVI); 2) discuss existing correlations between the NDVI vs data graphs and the agricultural productivity variable, in plots planted with rice, during two harvests, 18/19 and 19/20. For this purpose, the RGB color compositions, the NDVI vegetation index maps, and the average NDVI for the studied plots, from 12 images in each harvest, were extracted from the Land Viewer platform. The vector drawing of the fields was also carried out on the platform. RGB composites and NDVI maps were placed side by side to facilitate photointerpretation and change detection. Mean NDVI were plotted against date, and the resulting curves were analyzed at 6 significant points for comparison. The processing performed in the SPRING software consisted of creating a database, selecting, downloading from the INPE website and inserting bands 3, 4, 5 and 6 from the LANDSAT 8 satellite, for later obtaining BGR456 color compositions and NDVI maps. Within the time available for this research, the results revealed a much higher processing capacity and speed of the Land Viewer platform, to the point where it was not possible to complete the processing of the same number of images in SPRING. As a result, in the image sequences it was possible to identify the entry of the flood in January 2019, and the late and uneven ripening of the affected stand, resulting in a productivity difference of 4,310 kg/ha between the areas affected and not affected by the flood of the Quaraí River. Both NDVI vs Date curves also showed very different behaviour. Therefore, the area affected by the flood, 2018/2019 harvest, had lower productivity than the 2019/2020 harvest, which did not suffer from extreme weather events. Thus, the great potential of the Land Viewer platform for monitoring and analyzing the condition of agricultural vegetation in a fast and efficient way is highlighted.