Upscaling Tower-Based Net Ecosystem Productivity to global 250m using the Data Augmentation Method by Considering their Spatial Distribution

Terrestrial ecosystems have emerged as critical carbon sinks, holding a crucial role in the carbon cycle. Net ecosystem productivity (NEP) is a highly significant parameter in terrestrial ecosystems, representing the net ecosystem exchange (NEE) between ecosystems and the atmosphere, without considering other carbon fluxes from disturbances. In this NEP product, we harmonized various sets of tower-based NEP from flux sites as target variable, remote sensing product and meteorological data as traning variables. We further optimizied these smaple sets to address the problems in spatial distribution, culminating in a global NEP product spanning the years 2001-2020, achieved through the application of the random forest method. This dataset contains NEP data for global terrestrial ecosystems for the period 2001-2020 in PgC with a temporal resolution of 1 year. The spatial resolution of the product is 250m and the data format is TIFF. For detailed instructions on how to use the dataset, see User Guides.doc!