Zimbabwe’s national EO-assisted crop monitoring system

Zimbabwe is one of sub-Saharan Africa’s most agriculturally diverse countries, yet for years its national institutions faced a stubborn problem: how to produce timely, accurate, and affordable crop statistics at national scale — and how to monitor crops under intensifying climate stress. Field surveys were costly, slow, and difficult to sustain annually at scale.

Since 2024, FAO’s EOSTAT programme — with UCLouvain (Belgium) and the European Space Agency — has worked with Zimbabwe’s Ministry of Lands, Agriculture, Fisheries, Water and Rural Development (MLAWRR) and the Zimbabwe National Statistics Agency (ZIMSTAT) to change that. Using Sentinel-2 satellite imagery and the open-source Sen4Stat toolbox, the project produced Zimbabwe’s first national-scale crop type maps, statistically-grounded maize acreage estimates, and a field-level drought monitoring prototype.

By 2025, crop mapping accuracy rose from 77% to over 91%, ground data collection grew six-fold to over 17,000 georeferenced observations, and national partners began operating Sen4Stat on Zimbabwe’s own computing infrastructure. The project is now transitioning into a permanently institutionalised, nationally-owned monitoring system.

Zimbabwe’s agricultural sector supports the livelihoods of roughly two-thirds of its population, with maize as the dominant staple crop. Yet tracking what is grown — and how much — has been chronically difficult: traditional surveys require hundreds of enumerators across the country, producing statistics that are often delayed, imprecise, or too costly to sustain annually.

The challenge is compounded by Zimbabwe’s acute exposure to climate variability. The 2023-2024 El Niño drought — one of the most severe in recent memory — triggered emergency food assistance for millions. In the southern district of Chiredzi, sorghum fields were severely stressed while nearby irrigated sugarcane remained healthy. Without field-level monitoring tools, these contrasts remained invisible to national decision-makers until it was too late to act.

ZIMSTAT and MLAWRR needed annual crop maps, improved survey statistics, real-time drought signals, and a system they could own and operate. FAO’s EOSTAT programme, working with ten Zimbabwean institutions including ZCHPC, ZINGSA, the Meteorological Department, WFP, and Digital Earth Africa, was built to deliver all of this.

The centrepiece of the system is the national-scale crop type map produced annually using Sen4Stat. In 2024, Sentinel-2 time series was processed using a Random Forest classifier trained on thousands of georeferenced field observations, identifying 15 crop and land cover classes at 10-metre resolution across the entire country — Zimbabwe’s first national map at this level of detail. Overall accuracy reached 77%.

For 2025, the windshield survey grew nearly six-fold from 2,910 to 17,241 georeferenced points. For the first time, the country was stratified into agro-ecological zones (AEZ), with a separate Random Forest model per zone. This had a tangible consequence: a single national model had been misclassifying tea plantations in Zimbabwe’s cooler eastern highlands as sugarcane — two crops with similar spectral signatures but very different ecological profiles.

Zone-specific models correctly reclassified these as tea, confirmed by field verification, lifting overall accuracy to 91.1% and the cropland mask to 93%.

These maps were combined with field survey data through a regression estimator — a model that fuses satellite spatial coverage with statistical rigour to produce national maize acreage estimates with 95% confidence intervals. For the 2024 season, the national maize area was estimated at approximately two million hectares.

On the early warning side, a high-resolution Vegetation Health Index (VHI) was developed for Chiredzi district and scaled into a fully automated tool in 2025 — operating at 10-metre resolution, roughly 100 times finer than FAO’s standard 1-km global VHI product.

A central innovation is phenological alignment: rather than comparing vegetation health at fixed calendar dates, the tool aligns each season to the same stage of the crop growth cycle. Because planting dates in Zimbabwe can shift by several weeks between years, a calendar-day comparison is biologically misleading. The tool anchors each season where the NDVI crosses 20% of its amplitude, ensuring year-on-year comparisons reflect equivalent phenological moments.

The practical impact is striking. For a tracked pixel in Chiredzi during 2021 — a year with favourable rainfall — the unshifted VCI scored only 0.55, suggesting stress. This was an artefact: the crop was developing later than average. Once peak-aligned, the same pixel scores 1.00 — a complete reversal of the diagnostic conclusion from the same satellite data.

The system integrates freely available global satellite missions with nationally collected ground data:

The most immediate beneficiaries are ZIMSTAT and MLAWRR officials who can, for the first time, examine a 10-metre map showing precisely where maize, sorghum, tobacco, sugarcane, tea, and more than a dozen other crops are growing nationwide — making survey-based estimates more accurate and less costly. For smallholder farmers, better data leads to better-calibrated food security responses and more timely government interventions.

The 2023-2024 El Niño drought illustrated the system’s value vividly: the high-resolution VHI showed that within the same Chiredzi district, irrigated sugarcane remained healthy while rain-fed sorghum declined steadily from January through April 2024 — crop-specific, field-level evidence that is impossible to obtain from coarse 1-km products.

Capacity development has been central: three national webinars, a three-day workshop in Kadoma (November 2024), and continuous UCLouvain support have built the skills needed. By 2025, Zimbabwean technicians at ZCHPC were operating Sen4Stat independently.

Most significantly, what began as an FAO project is becoming a national institution: a Steering Committee, co-led by the Agricultural Marketing Authority and MLAWRR, and four Technical Working Groups are driving a three-year roadmap for full national ownership.