Evolution of the representation of global vegetation by vegetation continuous fields

Two decades of global annual fractional vegetation cover products have been derived using daily surface reflectance and land temperature data from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument aboard NASA's Terra satellite. These MODIS Vegetation Continuous Fields (VCF) are unique a distinct advance over discrete (ordinal) characterizations. VCF consists three continuous layers at 250 m resolution summing to 100%: tree cover, non-tree non-vegetated plus water. This paper describes fully-automated machine learning algorithm used create details recent improvements. Comparisons latest in MOD44B collection 6, with field-based other remote sensing demonstrate improved agreement: root mean square (RMS) deviations range 9 23%. A comparison canopy values those Global Ecosystem Dynamics Investigation (GEDI) gives an R2 value 0.53. Automation provides flexibility for readily porting method derive additional instruments similar radiometric characteristics such as Visible Infrared Radiometer (VIIRS) payload Suomi NPP JPSS platforms, successors.