2026-09-182026-09-182022-01-31RODRIGUES, Hellem Cristina Teixeira. Detecção e mapeamento de áreas afetadas pelo fogo no estado do Pará. Orientadora: Milena Marília Nogueira de Andrade. 2022. 91 f.: il. color. Dissertação (Mestrado em Gestão de Riscos e Desastres Naturais na Amazônia) - Programa de Pós-Graduação em Gestão de Riscos e Desastres Naturais na Amazônia, Instituto de Geociências, Universidade Federal do Pará, Belém, 2022. Disponível em: . Acesso em:.https://repositorio.ufpa.br/handle/2011/18752Fire is one of the main disruptors to the environment on a global scale. The occurrence of fires and forest fires affect the balance of ecosystems, human health, and consequently, the planet, making the affected environments more vulnerable to climatic and anthropic impacts. For the Amazon, the occurrence of forest fires is related, among other variables, to soil management and non-control of fires. Therefore, there is a need to analyze the Amazon space from new techniques, technologies and ideas that can analyze the landscapes with occurrence of fires and forest fires, being important to achieve resilient environments, inducing an effective response to uncontrolled fire. In this context, the objective of this work was to analyze the areas with occurrence of burning in the state of Pará and the development of an application for the detection and automatic mapping of the areas affected by fire in this state. For this, a script was built in JavaScript language within the Google Earth Engine Platform – GEE with spectral indices. The following were used: the Normalized Burn Ratio (NBR) – normalized burn rate – which is an index designed to highlight the burn area in a large fire zone; and the Difference Index between the NBR (dNBR) that was used to classify fires from high to low severity, also having the classification of areas regenerated after the fire. For the automatic mapping of the areas, the classes of areas affected by fire during the period of analysis were grouped and from the filter methodology with Boolean Kernel in square format, the edges generated in the filter were detected. To generate the data, the Pará mesoregions were used. As a result, it was possible in a short time to obtain on-screen detection of burned areas from the collection of images from the Sentinel 2 satellite or the Landsat 8 satellite, zoning and automatic mapping of fire-affected areas, with the option of exporting the results. from the mapping in raster and vector format to the user's personal drive. For the analyzed section of the BR-163, analyzing the months of July and August, the year with the most active burning was the year 2021 with 7.53% of burned area (4,700.42 km²), followed by 2019 with the percentage of 5.79% (3,616.12 km²) and the last in the ranking for the year 2020, with 2.69% (1,618.78 km²) of area with active fire from the mapping with images from the Sentinel 2 satellite, for Landsat 8, there is a panorama similar to the first place for burned area, with 5.68% of burned area for the year 2021 (3,083.02 km²), followed by the year 2020 with 1.83% (917 .59 km²) and last in the ranking the year 2019, with 1.56% of the burned area (973.96 km²). To further facilitate the use of the application, the entire script was made available in an appendix in this work and access via the repository, allowing the user to obtain the improvements made in the application in real time.Acesso AbertoAcesso Abertohttp://creativecommons.org/licenses/by-nc-nd/4.0/Mapeamento explícitoOperação automáticaGoogle Earth EngineExplicit mappingAutomatic operationDissertaçãoCNPQAMEAÇAS NATURAIS NO AMBIENTE AMAZÔNICOMINIMIZAÇÃO DE RISCOS E MITIGAÇÃO DE DESASTRES NATURAIS NA AMAZÔNIA