Geospatial intelligence · Central Asia
GEOBOX turns satellite imagery into field-level evidence for the people who manage Central Asia's water, farmland and cities. Built on more than a million observations collected in the field — not inferred from open data.
GEOBOX delivers to regional water authorities, agricultural agencies and city administrations across Central Asia — from district-scale irrigation audits to an urban heat platform built for a single city.
The problem
Irrigation across the region is still steered by allocation records — how much water a district was assigned, not how much its fields consumed. Some parcels are drowned, others starved, and until the season closes nobody has the evidence to say which.
Satellite data only closes that gap if somebody has checked it against reality. Ours has been.
Irrigated land across Central Asia, most of it managed on allocation figures rather than consumption data.
Of applied water lost before it reaches the farm gate in the weakest-performing districts.
Year-to-year swing in yield — largely predictable once the inputs are field-validated.
Of ground campaigns behind our training data. That baseline is not something a general-purpose model can borrow.
What we do
Each one is a deliverable, not a dashboard subscription: the analysis, the data layer underneath it, and a tool your team can operate without a GIS department.
Field-level water productivity from actual evapotranspiration set against applied water. Every parcel scored, every district benchmarked, the underperformers named and located.
For water authorities, basin agencies and irrigation districts
Multi-temporal optical and radar fusion resolves crop type at parcel level and separates irrigated land from rainfed — a base layer everything else can sit on.
For agriculture ministries, agri-lenders, insurers and agribusiness
Habitat suitability crossed with climate stress anomaly, trained on multi-year field records, to flag where an outbreak is building while the cheap window is still open.
For plant protection services and emergency planners
Land surface temperature and emissivity resolved to the individual building, with an intervention calculator that prices the cooling before anything is planted.
For city administrations, urban planners and architects
How we work
Three steps, in this order. The first one is the one most vendors skip.
Step 01
Multi-sensor imagery — Landsat, Sentinel, Planet, UAV — fused with more than a million labelled field observations from our own campaigns across the region. Sensor-agnostic by design, so the method survives a change of satellite.
Step 02
Purpose-built classifiers trained on our own dataset, not general models nudged in a plausible direction. Accuracy is reported at field level, against parcels our teams have physically stood in.
Step 03
Most clients have no in-house geospatial team and should not need one. We hand over an interface built for the person who acts on the answer: the water manager, the agronomist, the city planner.
Selected work
Four projects across water, agriculture, pest risk and urban climate — each validated against data collected on the ground.
Water · Qashqadaryo, UZ & Zhambyl, KZ
Water productivity scored for every field across 788,000 hectares, and the underperformers put on a map.
Pest risk · Southern Kazakhstan
Two independent outbreak pathways, separated — and flagged by district before the season starts.
Agriculture · South Kazakhstan
92,293 parcels classified into ten crop types, with irrigated land separated from rainfed.
Urban heat · Almaty, KZ
A city classified from Oasis to Inferno, with a per-building index and an intervention calculator behind it.
Why it holds up
A model is only as honest as the ground it was checked against.
Our classification and risk models are trained against more than a million field-labelled observations gathered across Central Asia over a decade of campaigns.
That baseline is the difference between a map that looks plausible and a number a regulator will sign. It is also the part that cannot be bought, scraped, or prompted into existence.
Work with us
Auditing an irrigation network, underwriting a season, forecasting an outbreak, cooling a city — we start from the decision in front of you, not from a platform demo.
