Pastures and rangelands are the single most extensive land use on Earth and we need to understand the productivity of these areas and how they are changing over time to meet global climate goals. Yet these areas rarely have dedicated monitoring efforts and there is currently no good understanding of their location, area, and condition.

A new Land & Carbon Lab research consortium, Global Pasture Watch, aims to produce recurrent and high-resolution (30-meter) maps for pasture areas and productivity from 2000 onwards. The project uses ensemble machine learning algorithms, visual interpretations, crowdsourcing, and earth observation data fusion to produce per-pixel probabilities of land use and land cover, livestock density, short vegetation height, and gross primary productivity (GPP).