The increasing availability of products generating burned area (BA) maps in recent years necessitates the creation of more accurate reference perimeters to validate these products and provide users with information about their accuracy. For this purpose, reference perimeters were created using Sentinel-2 images in Latin America and the Caribbean (LAC) for the year 2019. The sampling was adapted to the peculiarities of the Sentinel-2 tiling grid system, and statistically representative sample units were selected for biomes and fire activity through stratified random sampling. Fire perimeters were extracted using a Random Forest supervised classification and results were manually supervised and refined. Efforts were made to maximize the temporal length covered by the reference perimeters for each sample, aiming to minimize temporal errors when using the perimeters for validation. The dataset covers 569,214.2 ${\mathrm{k}\mathrm{m}}^2$ (3.5% burned, 88.7% unburned, and 7.8% unobserved). These perimeters were compared with higher spatial resolution PlanetScope-derived perimeters, resulting in 8.4% commission errors and 3.8% omission errors. As a validation exercise, MCD64A1 and FireCCI51 global burned area products were validated using the Sentinel-2 reference dataset created, confirming that the temporal extent of the reference perimeters significantly affects the validation of such products. The reference fire perimeters are publicly available in the Burned Area Reference Database (BARD).

Land cover is a key determinant of landscape structure and ecological processes across spatial and temporal scales. In freshwater ecosystems, where riparian and catchment land cover regulate hydrology, nutrient inputs, and habitat quality, most ecological assessments rely on systematically produced land cover maps whose reliability varies with scale and context. Despite their widespread use, the extent to which these products accurately represent landscape patterns and long-term dynamics at ecologically relevant scales remains poorly understood. Here, we evaluate the reliability of systematic land cover maps and assess whether Landsat-based supervised classification (SC) provides more accurate and temporally consistent representations of land cover patterns in draining catchments and riparian areas. We analysed five headwater catchments (< 70 km ² ) within Natura 2000 sites in northern Spain and reconstructed land cover dynamics over four decades (1984–2023) using Random Forest classification implemented in Google Earth Engine. SC results were compared with CORINE Land Cover, SIOSE, and the Spanish National Forest Inventory across three spatial scales, relevant for freshwater ecological processes: catchment, riparian corridor, and reach. SC achieved consistently high accuracy across all periods (mean overall accuracy = 87.8%), outperforming CORINE and SIOSE and matching NFI only for forest classes. Discrepancies between SC and systematic maps increased at finer spatial scales and varied with landscape context, particularly in heterogeneous riparian environments. Systematic maps also showed limited temporal consistency, frequently misrepresenting land cover trends relative to SC. These scale-dependent and context-specific biases can alter estimates of riparian vegetation and landscape composition and their temporal dynamics, potentially leading to misleading assessments of freshwater ecosystem condition. Our findings demonstrate that reliance on coarse-scale land cover products can distort ecological inference, particularly for freshwater ecosystems where ecological condition depends strongly on catchment and riparian landscape structure. Integrating reproducible, remote sensing–based classifications offers a practical pathway to improving the accuracy and interpretability of land cover indicators in landscape ecological research.