The Use of GeoAI and Multimodal Remote Sensing in Monitoring Amazonian Biodiversity: A Systematic Review

Abstract

The Amazon rainforest plays a fundamental role in maintaining global biodiversity and climate regulation, yet it faces increasing pressures associated with deforestation, forest degradation, and land-use change. In this context, the present study aimed to conduct a systematic review of the scientific literature on applications of GeoAI, multimodal remote sensing, and artificial intelligence for monitoring Amazonian biodiversity. The review followed the PRISMA 2020 guidelines, with searches conducted in the Scopus, Web of Science, ScienceDirect, PubMed, and IEEE Xplore databases. After applying the inclusion and exclusion criteria, 35 studies were selected for analysis. The results revealed a predominance of applications related to deforestation monitoring, biomass estimation, forest degradation, and vegetation structure, mainly using convolutional neural networks, deep learning, and the integration of optical, LiDAR, SAR, and drone/UAS sensors. A significant increase in publications after 2020 was observed, associated with the expansion of multimodal approaches and cloud-based geospatial platforms. The studies demonstrated high analytical performance in detecting complex environmental patterns, although limitations related to methodological heterogeneity, availability of labeled datasets, and external model validation remain important challenges. It is concluded that GeoAI has strong strategic potential to strengthen Amazon environmental monitoring and support conservation, forest management, and territorial planning actions.