National 10 m Forest Tree-Genus and Stand-Composition Map of Romania

This dataset provides the first national 10 m map of forest tree genera and stand composition across Romania. It distinguishes pure and dominant genus-composition classes and provides a spatially continuous representation of forest composition at the national scale. The map was developed by integrating Sentinel-2 imagery, topographic and climatic predictors, and forest-management reference data within an autoencoder–multilayer perceptron framework. The dataset contains 22 forest genus-composition classes representing Abies, Carpinus, Fagus, Picea, Pinus, Populus, Quercus, Robinia, Salix, Tilia, and grouped deciduous and evergreen classes. Pure classes represent stands in which one genus accounts for at least 80% of stand composition, whereas dominant classes represent stands in which the leading genus accounts for more than 50% but less than 80%. The map was evaluated using spatially independent model testing, external field observations, comparison with a pan-European forest product, and regional comparisons with Romanian National Forest Inventory estimates. It is intended to support forest monitoring, biodiversity assessment, conservation planning, landscape analysis, and sustainable forest management. An interactive visualization is available through Google Earth Engine: Link