GLOBMAP FTC: a global annual fractional tree cover dataset since 2000

Fractional tree cover facilitates the depiction of forest density and its variations. However, it remains challenging to estimate fractional tree cover from satellite data, leading to substantial uncertainties in forest cover changes analysis. We generated a global annual fractional tree cover dataset (GLOBMAP FTC) from 2000 to 2021 with a resolution of 250 m using the MODIS land surface reflectance products. MODIS annual observations were realigned at the pixel level to a common phenology with the midpoint date of growing season centered in the calendar year. Twelve features that can differentiate between trees and herbaceous vegetation were extracted from the realigned MODIS observations. A massive training data, consisting of 465.88 million sample points, was collected across the globe from four high-resolution forest cover products, including ESA WorldCover, GlobeLand30, PALSAR FNF, and ESRI Land Cover. A feedforward neural network model was trained to predict tree cover. Global and regional analysis demonstrated that the dataset could characterize the forest cover changes, including both abrupt forest loss and gradual forest gain. This dataset can be applied for assessment of forest cover dynamics and monitoring the progress of forest restoration projects and policies.

The dataset is provided by 296 to 323 1200 km × 1200 km tiles for each year, aligning with the MODIS standard tiling system. The data are stored in the Sinusoidal projection with Geotiff format. The files are designated with the naming structure "GLOBMAPFTC.AYYYY001.hHHvVV.V01.tif", where “YYYY” represents the year of the data, while “HH” and “VV” indicate the horizontal and vertical tile indices respectively, aligning with the MODIS standard tile. The valid range of tree cover value is 0-100, with a scale factor of 1.0 and the unit of percentage (%).