The Brazilian Cerrado, one of the most threatened biomes of our planet, illustrates the challenges and opportunities of reconciling economic development with conservation of land and water ecosystems. Here, we assess the state of the art of and present new information on the impacts of agricultural expansion, dams, and water use, and make recommendations for basin management, conservation, and restoration of water‐related Cerrado ecosystems and rivers. The conservation of the Cerrado requires not only the preservation of remnants of its vegetation but also the ability to keep the hydro‐geomorphological and ecological functionality of its rivers, particularly the Araguaia River, the last large well preserved and unregulated system. If business as usual continues, the Cerrado riverine ecosystems may never rebound.

Hydropower dams are touted as one of the cleanest forms of energy production, yet they are associated with severe environmental impacts on both the physical structure and functioning of river ecosystems. The threat is particularly acute in the Brazilian Cerrado—a biodiverse savanna region, spanning over 2 million km², that concentrates the headwaters of several critical South American watersheds. Our study analyzed the current distribution of large and small hydroelectric plants in the Cerrado and focused on understanding their effect on land use changes. We also propose a Dam Saturation Index (DSI) to help spur more integrated planning for this region. Results indicate that the Cerrado river basins contains 116 (30%) of Brazil’s large hydroelectric plants and 352 (36%) of its small hydroelectric plants. Moreover, these plants spurred significant land use changes within a 5-km buffer of the dams, with over 2255 km² of native vegetation cleared by 2000 and an additional 379 km² in the ensuing 20 years, could reach ~1000 km². Based on the historical anthropization process in the Brazilian savannas, we expect new crops, pastures, and urban equipment to be incorporated into this landscape, with different impact loads.

Considered a conservation hotspot of the world biodiversity and a key region for the agriculture production in Brazil, the Cerrado biome has only 7.5% of its native vegetation as fully protected areas. Given this, in 2016 the Brazilian government started an official project to monitoring deforestation in the biome, through the so-called PRODES-Cerrado, responsible for mapping deforested areas from 2000 on, and DETER-Cerrado, responsible to generate deforestation alerts. Seeking to contribute with both context information and confidence levels for the polygons produced by these two monitoring systems, we developed the Deforestation Polygon Assessment Tool. This web-based platform process and presents several analysis for PRODES-Cerrado and DETER-Cerrado polygons using automatic assessments (e.g. BFastMonitor and Weights of Evidence), field validation and spatial analysis with key datasets (e.g. National Land Registry, Land-Use and Land-Cover maps). The platform implements an interactive map which allows a fast and comprehensive visualization of different layers, as well as a Deforestation Report at the polygon level, which gathers all the information about each polygon, providing greater reliability and understanding of the deforestation dynamics in the Cerrado. Future improvements in the platform will consider additional, spatial relations in order to assist government agencies to either prevent or reduce deforestation ocurrences in each municipality in the Cerrado biome.

Selective logging often marks the beginning of forest degradation, but remote sensing alert systems typically detect it long after it starts. This type of disturbance produces a subtle canopy signal not easily recognised in the early stages, causing alerts in some areas of the Brazilian Legal Amazon to lag behind the first visible signs by several months. This study explores whether a geospatial foundation model can help reduce this time gap. For each documented disturbance site, we compile Harmonized Landsat and Sentinel-2 (HLS) image time series spanning October 2024 to May 2026 into a datacube and use it to pretrain, without labels, a compact spatiotemporal masked autoencoder (ST-MAE). This model learns to embed the dynamics of forest degradation into latent-space representations, the so-called "embeddings". Following this, a lightweight downstream model — trained with few labelled samples — interprets ST-MAE's embeddings to evaluate every new HLS observation for recent logging activity. To effectively translate the evaluation scores into production-ready alerts, we apply a simple change-detection approach: an edge filter for sudden logging-level changes, paired with a persistence filter for sustained detections. We conducted a leave-one-out cross-validation assessment across 20 well-documented logging sites: yielding no false alarms prior to the first disturbance evidence, the change detector confirmed the onset of the 20 logging activities with a median delay of 15 days after the first post-disturbance image. The downstream model's scores remain low during the pre-disturbance period but increase once logging begins, confirming that it responds to logging-related events rather than to other landscape dynamics or calendar artefacts. This label-free pretraining enables effective performance with minimal annotations, providing a practical solution to reduce the temporal gap in monitoring forest degradation.