Estimates of forest losses with remote sensing and multispectral drone in zoonotic landscapes under the effect of deforestation in the Amazon

Remote sensing data is important for deforestation estimates and environmental monitoring of the Amazon forest. In a previous project, data from LANDSAT, SENTINEL and CBERS satellite images were used to produce an analytical protocol for measuring land use and occupation in rural settlements in the Brazilian Amazon. It was shown in articles published from the scientific initiation that the incidence of malaria and yellow fever cases are associated with the variation in composition and configuration of the remaining tropical forest in the Amazon and Atlantic Forest biomes, respectively. However, the major limitation of these studies was the spatial resolution of satellite images, which affects the precision and accuracy of measuring the parameters of the sampled terrain. This could also affect the measure of the effect of deforestation on the risk of malaria to the local population. The present project will use the logistical, operational and intellectual structure of the Main Project (FAPESP-JP2 21/06669-6). The first cross-sectional study of the main project took place and the preliminary results generated were presented in the present proposal, including the theoretical and analytical rationale for the selection of the 40 landscapes under study in the Santa Luzia settlement, municipality of Cruzeiro do Sul, state of Acre, Brazil. Estimates of the composition and configuration of land use and occupation in all these landscapes at three scales (7-3 Km2) were carried out for the period July 2022 (first cross-sectional study of the main project). Here the proposal sets to use and application of an ultra-modern multispectral drone flight plan to capture orthomosaics (images with quality for landscape analysis) of the 40 study landscapes in the second (July 2024) and third (July 2026) studies transversal. These images will be classified with artificial intelligence (machine learning) to produce deforestation estimates with a minimum spatial detection of 0.04 meters between 2024-2026. The project will apply the same classification to images from LANDSAT, SENTINEL and CBERS satellites to compare sensors on deforestation estimates, using the kappa agreement coefficient and linear and logistic regression models. The variation in deforestation estimates will be evaluated in relation to its predictive association with the risk of malaria incidence. This risk will be estimated as an integral part of the Main Project in which there are other researchers and postgraduate students working in national and international collaboration networks. The expected results could contribute to the monitoring of deforestation in the Amazon forest and will provide important information for the current plan to eliminate malaria in Brazil. The results will be disseminated through scientific publications, scientific meetings, conference presentations and in accordance with the dissemination plan of the Amazônia+10 FAPESP program.

Project details

Information Systems
Tools for Elimination

Dec 2023 — Nov 2027

Brazil