Built from satellite climate models and national geospatial infrastructure — not from individual farmer activity, except where explicitly noted below. This page exists so you don't have to take that on trust.
Regional coverage today: Zambia only (all 10 provinces). The underlying satellite archive covers the continent — a second country is a real data-loading project, not a toggle, so we scope it per engagement rather than claiming coverage we haven't built.
Data sources
Satellite-based rainfall estimates and soil moisture modeling, refreshed on a 5-day (pentadal) cycle. Every province has a 10-year historical baseline (720 pentadal readings) used to compute anomalies and drought-risk probability — not a single-year snapshot.
Zambia's national spatial data infrastructure — water points, boreholes, aquifer and catchment coverage, and 12,461 surveyed farm parcels. Independently maintained government geospatial data, not modeled or estimated by Netagrow.
Crop mix and field data reported by farmers using the Netagrow app — the one data source here that isn't a satellite or government re-serve. Aggregated to province level only; see the privacy section below for exactly how.
Known limitations
Drought-risk trajectory data includes an ensemble_size for every reading — the number of historical years on record that could be used for that specific forecast horizon. Readings with ensemble_size below 5 should be treated as lower-confidence; longer horizons (120-160 days out) naturally have fewer complete historical years to draw on than near-term ones (30 days).
Anomaly is computed against a historical baseline average. For a pentad-of-year that's historically almost rainless, a small baseline denominator can produce an extreme percentage that doesn't reflect a proportionally extreme real-world change. We flag this as a known limitation rather than hide it — if you see an anomaly beyond roughly ±500%, look at the underlying rainfall_mm and baseline_avg_mm values directly rather than the percentage. This is on our list to fix at the computation layer.
Our climate-risk tables retain a history of past computation runs. Every dashboard view and API response is scoped to the most recent run by default — we don't mix historical snapshots into a "current" figure. (The institutional API's drought-risk endpoint offers an explicit history=true option for anyone who wants the accumulated series on purpose.)
Farmer data privacy
The one dataset that comes from our own farmer network (crop mix by province) is aggregated with a strict floor: any province with fewer than 20 distinct reporting farmers returns a suppressed result rather than a thin, potentially re-identifiable figure. Raw coordinates are never returned. No farm-level or farmer-level data is ever exposed via the dashboard or the API — only province-level aggregates.
Every other dataset (rainfall, drought risk, soil moisture, water infrastructure) is satellite- or government-sourced and was never farmer data to begin with.
Two ways in
The Corporate dashboard — scorecard, map, AI-generated brief — is self-serve at K1,200 per user a month.
Register / Sign in →The institutional API (rainfall, drought risk, water, crop distribution, 40-year satellite backtest data) is part of the Enterprise tier — a manually-provisioned key, not self-serve.
Read the API documentation →Talk to us before you build on it — we'd rather explain a limitation up front than have you find it in production.