Operational sentinel-2 system for monthly near‑real‑time irrigated area mapping in the Limpopo river basin

Monitoring irrigated agriculture is critical in the water-scarce Limpopo River Basin (LRB). However, existing approaches are often coarse, retrospective, or season-aggregated, which limits their ability to capture smallholder irrigation and the month-to-month dynamics required for operational management. This study addresses this gap by developing and validating a scalable, semi-supervised framework to produce monthly dry-season (May–September) 10 m irrigated-area maps and associated water-use estimates across the LRB for 2019–2024. The workflow integrates Sentinel-2 imagery, a Random Forest classifier, time-lagged precipitation–vegetation analysis, and slope masking in Google Earth Engine, and links mapped irrigated area to FAO’s WaPOR (Water Productivity through Open access of Remotely sensed derived data) evapotranspiration to estimate water use. Validation against independent field observations (n = 190) achieved 80% overall accuracy (κ = 0.60). Dry-season irrigated area declined from ~ 211,000 ha (2019) to ~ 185,000 ha (2024), while mean dry-season water use increased from ~ 103–134 × 106 m3, indicating rising irrigation intensity. Irrigation hotspots were concentrated in key sub-basins including the Middle Olifants, Crocodile, and Letaba. The resulting open-access, basin-scale product provides operational irrigation intelligence to support transboundary water allocation and drought response. It also offers a replicable model for other water-stressed basins.