3.2.2 — Food Security Systems — maturity: live

Crop Yield Forecasting

Using multi-spectral and SAR satellite time-series to generate sub-national crop yield estimates weeks before harvest, independent of ground-survey networks.

Auto-drafted reference page — content reviewed for shape, individual references not yet link-checked.

Governments that rely on farm-level surveys and trader reports to forecast national harvests are always behind the curve. By the time the data is compiled, prices have moved, import tenders are late, and the window for emergency procurement has narrowed. A sovereign satellite stack changes the timeline: dense revisit multi-spectral imagery feeds vegetation indices (NDVI, EVI, LAI) and crop growth models that produce county-level yield estimates six to eight weeks before combine harvesters roll.

The satellite stack works in layers. Optical constellations at 3-10 m resolution map crop type, phenological stage and canopy health across the entire agricultural calendar. SAR adds a weather-independent layer that sees through the cloud cover that routinely blankets tropical and monsoonal growing regions at exactly the wrong time. Fusing both streams into a calibrated crop model — anchored against historical yield statistics and soil-moisture data — produces probabilistic yield forecasts with province-level uncertainty bounds that decision-makers can actually use.

The operational payoff is direct budget sovereignty. A ministry of agriculture holding a credible, early yield forecast can time grain reserve purchases, negotiate import contracts from a position of knowledge rather than rumour, and pre-position food assistance before a deficit becomes a crisis. Nations that outsource this intelligence to commercial data vendors or donor-funded monitoring programmes hand the same information to commodity traders and foreign governments simultaneously, eliminating any pricing advantage.

What matters

A six-to-eight-week forecast lead time over harvest is the difference between competitive import procurement and panic buying at peak prices.
SAR coherence change detection catches crop lodging and early senescence that optical indices miss during cloudy monsoon and tropical wet seasons.
Sub-national (district or county) granularity is operationally required — national averages mask localised deficits that trigger internal food insecurity.
Vendor-supplied yield forecasts are routinely shared with commodity desks and hedge funds; sovereign data stays inside the government's decision cycle.

Sovereignty score: 8/10 — Yield forecasts are strategic economic intelligence; a nation that sources them from a commercial vendor or foreign programme surrenders its information advantage to the very markets it must negotiate with.

Commercial satellite data providers sell the same imagery and derived products to commodity trading desks, hedge funds and foreign governments, eliminating any procurement or policy advantage a ministry might otherwise hold.
Donor-funded monitoring programmes (FEWS NET, GEOGLAM) prioritise humanitarian early warning over national procurement strategy and may withhold, delay or re-frame data to serve their own mandates.
Domestic crop statistics combined with satellite-derived phenology constitute sensitive national economic data; routing this through foreign cloud platforms exposes it to foreign intelligence collection and legal-process compulsion.
Export-control regimes and vendor licensing terms can restrict access to high-resolution SAR or optical data precisely during geopolitical tensions — the same periods when harvest intelligence is most operationally critical.

Reference architecture

Payload: Push-broom multi-spectral imager, 5 m GSD, bands: Blue/Green/Red/Red-edge/NIR/SWIR-1/SWIR-2; secondary X-band SAR payload, 5 m stripmap, 80 km swath, dual-polarisation (VV+VH) for soil-moisture and crop-structure retrieval
Bus class: ESPA-class microsat, 120-160 kg, 600 W payload power; dual-payload configuration requires a larger bus than a pure optical cubesat approach
Orbit: Sun-synchronous LEO at 520-560 km; 18-satellite walker constellation (12 optical, 6 SAR-primary); 3-day full-coverage revisit at 5 m, daily at 20 m through cross-track agility; local overpass timed 10:00-10:30 LT for consistent solar illumination
Ground segment: 4-station national network covering major agricultural zones (S-band TT&C, X-band high-rate downlink at 300 Mbps); co-located calibration/validation sites with reference flux towers and automatic weather stations; SatNOGS-compatible UHF beacon for telemetry redundancy
Software & data
Latency
Cost band
Lead time

References

FAO Crop Monitoring for Food Security — Methodology and Global Applications — FAO's Global Information and Early Warning System integrates satellite-derived vegetation indices and crop model outputs to produce sub-national yield forecasts for food-insecure countries. The system demonstrates that timely remote-sensing data materially improves the accuracy of pre-harvest production estimates.
ESA WorldCereal: Global Cropland and Crop Type Mapping — ESA's WorldCereal project used Sentinel-1 SAR and Sentinel-2 multi-spectral data at 10 m resolution to produce consistent global cropland and crop-type maps, demonstrating the feasibility of SAR-optical fusion for large-area agricultural monitoring.
USGS Landsat-Based Agricultural Monitoring — GLAM System — The USGS Earth Resources Observation and Science Center provides multi-decadal Landsat archives used to calibrate crop growth models and validate in-season yield forecasts. Long time-series consistency is essential for bias-correcting model outputs against historical production statistics.
GEOGLAM Crop Monitor — Standards for Satellite-Based Crop Assessment — GEOGLAM coordinates national and international satellite monitoring efforts for agricultural production, publishing monthly crop condition assessments used by G20 governments. The programme explicitly notes that countries with autonomous satellite capacity contribute more reliable and timely national-level inputs.
World Bank — Satellite Data for Agricultural Yield Forecasting in Developing Countries — World Bank research confirms that satellite-derived crop yield models in low- and middle-income countries reduce pre-harvest forecast errors by 20-40% compared to ground-survey-only methods, with the greatest gains in regions with sparse extension-worker networks.