Open-Earth-Monitor Global Workshop 2026

While forest monitoring has reached high levels of maturity, grassland ecosystems remain a critical "blind spot" in global conservation. To address this, Global Pasture Watch (GPW) has established a comprehensive baseline using 30m multi-decadal datasets (2000–2022) covering grassland extent, vegetation height, and livestock density. However, the inherent heterogeneity and rapid seasonality of these landscapes present significant current challenges for traditional pixel-based classification. To overcome these barriers, our next steps involve transitioning to next-generation machine learning models that utilize Sentinel-2 spatial-temporal embeddings. By moving beyond simple spectral signatures to rich, high-dimensional latent representations, we can better capture the nuances of managed vs. natural grasslands and monitor Gross Primary Productivity (GPP) with unprecedented precision. This evolution in our workflow aims to deliver near-real-time, actionable insights, transforming how we track land-use conversion and guide sustainable restoration across the world’s most vulnerable non-forest biomes.