A new map of South Manchurian mixed forests

We publish the results of mapping the South Manchurian mixed forests (SMMFs) for 2021, with comparative analysis against 2001. SMMFs represent one of the most biodiverse temperate forest ecosystems globally, spanning northeastern China, southern Primorsky Krai (Russia), and the Korean Peninsula. These forests face significant degradation, with large areas replaced by secondary growth, agricultural land, and anthropogenic vegetation. Precise quantification of remaining SMMFs is critical due to their high conservation value. Our methodology integrated an original SMMFs distribution dataset (vegetation relevés, literature records, and regional maps) to train satellite data classifiers. We utilized Global Land Analysis and Discovery (GLAD) analysis-ready Landsat time-series at 30m resolution. Mapping employed decision tree ensembles calibrated with manually collected training data, with stratified sampling for accuracy assessment. Potential distribution modeling incorporated community-level presence/absence datasets using topographic (elevation, slope, aspect) and bioclimatic predictors. Training Data.csv – Training dataset: Vegetation relevés and verified occurrence points for SMMFs forest2001mask.tif and forest2021mask.tif – Binary forest cover masks (1: forest, 0: non-forest) for 2001 and 2021 smmf_2001_binary.tif and smmf_2021_binary.tif – SMMFs distribution maps (binary: 1 = SMMFs, 0 = other land cover) Sampling.csv – Stratified sampling points for accuracy assessment sampling_sum.tif – Stratified aggregation layer for error quantification smmf_combined.tif – Composite SMMFs distribution raster (2021) Map_SMMFs_area.shp – SMMF area distribution per 10×10 km grid cell Map_SMMFs_patches.shp – Spatial boundaries of SMMFs forest patches with size attributes proj_Current_SMMF.tb[...].tif – Potential SMMFs distribution models (target-background sampling ensemble) proj_Current_SMMF.sample[...].tif – Potential distribution models (stratified sampling ensemble) Results_by_algo_SMMF_tb.csv – Model evaluation metrics (target-background approach) Results_by_algo_SMMF_sample.csv – Model evaluation metrics (stratified sampling approach)