Desenvolvimento de índice e ferramenta geoespacial para análise da integridade ecológica em nascentes do cerrado brasileiro

Abstract

The Brazilian Cerrado, a biome of recognized importance for biodiversity conservation and water regulation, is under intense anthropic pressure, especially due to agricultural expansion and environmental fragmentation processes. Given this scenario, this thesis aimed to propose, test, and validate an Ecological Integrity Index (EII) specific to spring areas, using remote sensing data obtained by multispectral drones. In addition, a geospatial plugin for QGIS was developed, capable of operationalizing the entire EII calculation flow, spatial analysis, and automated generation of maps and reports. In Chapter 1, the methodology for constructing the EII was structured, based on the integration of the spectral indices NDVI, NDWI, PSRI, BSI and CHM, synthesized via Principal Component Analysis (PCA). The statistical validation demonstrated strong consistency between the indices, with emphasis on NDVI and NDWI, which presented greater explanatory power in the spatial variability of ecological integrity, according to GeoDetector analyses and Spearman correlation. The application of the EII revealed that the analyzed springs mostly present conditions of low to critical ecological integrity, with emphasis on springs N01, N02 and N06, which exhibited indices below 0.40, reflecting a high degree of degradation. On the other hand, spring N05 maintained moderate integrity (0.504), although with alert zones. Chapter 2 details the development and validation of the GeoFlow plugin, a computational tool that integrates the entire EII processing flow within QGIS. The plugin allows for everything from pre-processing of images to generating final maps, statistical analyses and automatic reports. Performance tests have shown that the plugin significantly reduces processing time, increases scientific reproducibility and allows its use by users without advanced programming knowledge. The functionalities were validated in the same spring areas, confirming the spatial patterns detected in Chapter 1. In general, the results of this thesis confirm that EII is a robust, sensitive and applicable tool to the reality of Cerrado springs, and that its implementation via plugin represents a significant advance in the democratization of geotechnologies applied to environmental monitoring. The proposed approach proves to be scalable and adaptable to other regions and biomes, strengthening environmental management based on high-precision data.