Agriculture & Environment
Water Availability Intelligence API
Water risk is becoming a financing and siting question rather than an agronomic one. The data to answer it is mostly public and almost never assembled.
- Agricultural lenders and insurers
- Irrigation and agritech platforms
- Industrial site selection teams
- Water utilities
- Sustainability and disclosure teams
What your customers would see
example interfaceMockupIllustrative mockup of the interface we would build. Figures shown are examples, not results from a live deployment.
The problem this solves
Drought exposure used to be something a farmer managed and a reinsurer priced. It has become a question asked by lenders assessing agricultural security, by manufacturers choosing where to put a plant that needs process water, and by disclosure frameworks that now require firms to report water risk across their operations and supply chains.
The answers available are usually either too coarse or too specific. Global water stress indices rate whole catchments on a five-point scale, which is useful for a map and not for a decision about a particular holding. A local hydrological study answers one question properly at a cost and timescale that rules out portfolio-scale assessment. Between the two there is very little.
What fits that gap is a composite built from open data: precipitation anomalies and their persistence, soil moisture from satellite, river gauge levels relative to their own historical distribution, reservoir storage where published, groundwater trends where monitored, and abstraction licensing constraints where they are on a register. None of that is a hydrological model of an aquifer. It is a defensible, repeatable indicator with its components visible, which for most of these decisions is the right level of precision.
Who buys this
Agricultural lenders and insurers, Irrigation and agritech platforms, Industrial site selection teams, Water utilities, Sustainability and disclosure teams
What the API returns
What comes back
- Water stress and drought risk scores by location and catchment
- Precipitation anomaly and deficit persistence against local climatology
- Soil moisture status from satellite observation
- River flow and reservoir storage percentiles where data is published
- Groundwater level trends where monitoring networks exist
- Irrigation demand estimates from crop, stage and evapotranspiration
- Abstraction constraint indicators from published licensing registers
- Seasonal outlook with explicit skill reporting by region and lead time
Endpoints
| Method | Path | Returns |
|---|---|---|
| POST | /v1/locations/water-risk | Water stress and drought risk for locations or parcels |
| GET | /v1/catchments/{id}/status | Current hydrological status for a catchment |
| GET | /v1/locations/{id}/outlook | Seasonal water availability outlook |
| POST | /v1/irrigation/demand | Irrigation requirement estimates for cropped areas |
| GET | /v1/catchments/{id}/constraints | Published abstraction and licensing constraints |
| WS | /v1/stream | Push drought threshold breaches and status changes |
Request and response
illustrativeRequest
{
"locations": [
{ "location_id": "farm_1182", "lat": 52.61, "lon": 0.42, "use": "irrigated_arable" },
{ "location_id": "site_ebro_3", "lat": 41.64, "lon": -0.88, "use": "process_water" }
],
"horizon": "seasonal",
"include": ["components", "constraints", "outlook_skill"]
}Response
{
"issued_at": "2026-10-22T00:00:00Z",
"locations": [
{
"location_id": "site_ebro_3",
"catchment": "ebro_middle",
"water_stress_score": 74,
"band": "high",
"components": {
"precipitation_anomaly_12m_pct": -28.4,
"deficit_persistence_months": 14,
"soil_moisture_percentile": 0.11,
"river_flow_percentile": 0.08,
"reservoir_storage_pct_of_capacity": 31.2,
"reservoir_vs_10y_median": -22.0,
"groundwater_trend": "declining"
},
"constraints": {
"register_status": "restrictions_in_force",
"detail": "Basin authority drought decree limiting new abstraction, published 2026-08-14",
"source_as_of": "2026-10-01"
},
"outlook": {
"horizon": "2026-11/2027-01",
"direction": "continued_below_normal",
"confidence": "low",
"skill_note": "Seasonal precipitation forecast skill in this region is weak; treat as a scenario, not a forecast"
}
}
]
}Where the data comes from
| Precipitation observations and reanalysis | National gauge networks and open reanalysis such as ERA5 for the long climatology every anomaly is measured against. Mostly open; some national archives are licensed |
| Satellite soil moisture | Open missions including SMAP and Sentinel-1 derived products. Coarse resolution relative to a single field, which constrains how locally these scores can be read |
| River gauge and reservoir data | National hydrological agency publications. Open in many countries with attribution, patchy or absent in others, and occasionally restricted for commercial reuse |
| Groundwater monitoring networks | Borehole level records where a monitoring network exists and publishes. Coverage is the limitation — many important aquifers are barely monitored |
| Abstraction and licensing registers | Published allocations and drought restrictions. Authoritative where they exist, frequently lagging the operational reality by weeks |
| Seasonal forecast ensembles | Open seasonal prediction systems, used for outlook with their genuinely limited skill reported rather than concealed |
The intelligence layer
The difference between a demo and a service is mostly in this section. These are the parts that keep the answer trustworthy once real traffic is hitting it.
- A composite index with every component exposed, so a user can see whether a high score is driven by reservoir storage or by soil moisture and judge it themselves
- All anomalies computed against each location's own historical distribution rather than against absolute thresholds, since a dry year in East Anglia and a dry year in Aragon are not the same quantity of water
- Deficit persistence treated as a separate dimension from deficit magnitude, because fourteen months of mild shortfall depletes storage differently from one severe dry season
- Catchment-scale and point-scale indicators kept distinct, as reservoir and river data describe a catchment while soil moisture describes a pixel, and conflating them produces a score that means nothing at either scale
- Irrigation demand estimated from reference evapotranspiration, crop coefficients by growth stage and soil water-holding capacity, using established published methods
- Seasonal outlook delivered with measured skill scores by region and lead time, and marked as a scenario where skill is weak — which in most mid-latitude regions it is
- Validation of drought indicators against recorded drought declarations and documented impacts rather than against each other
- Monitoring for gauge outages and register publication lag, both of which make conditions look better than they are by leaving the latest deterioration out
What it changes for the business
Credit and security assessment
Consistent water risk scoring across an agricultural loan book replaces case-by-case judgement.
Siting decisions
Process water availability and abstraction constraints affect plant location before capital is committed.
Insurance pricing and reserving
Catchment-level drought indicators support exposure management across a portfolio.
Disclosure reporting
Component-level water risk with named sources produces the auditable basis a disclosure framework asks for.
How it fits into your product
How it embeds
- Locations as coordinates, parcel geometry or catchment identifiers
- Batch endpoints for portfolio-scale scoring across thousands of locations
- Map layers for catchment status and stress bands in your existing mapping interface
- Webhooks on threshold breaches, restriction announcements and status changes
- Report generation for lending, insurance and disclosure files with sources cited
- White-label delivery for agritech, lending and sustainability platforms
Included in every build
- Authentication, scoped API keys and per-customer rate limiting
- Input validation with predictable, documented error responses
- Versioned endpoints so your integration never breaks without notice
- Request and response logging for audit and dispute resolution
- Latency, error-rate and usage monitoring with alerting
- OpenAPI specification, sandbox environment and client examples
- Backups, failover design and a documented recovery procedure
- An agreed support arrangement once the service is live
How we would build it
about 20 weeksDiscovery and feasibility
Weeks 1–3
This one depends heavily on data access, so we start by proving the sources exist, can be licensed for your use, and actually carry the signal the model needs. If they do not, you find out here.
Architecture and baseline model
Weeks 4–6
A written architecture and a baseline model measured against a holdout period, so you know what accuracy you are buying before the full build starts.
Production build
Weeks 7–16
Data pipelines, model training and serving, historical archive, validation harness, API, security, monitoring and documentation.
Validation and integration
Weeks 17–20
Backtesting against held-out history, accuracy reporting your team can audit, integration support and an agreed retraining cadence.
Commercial models
There is no fixed price list, because the cost is driven by data complexity, integration count, model requirements, request volume and the support level you need. Scope is agreed in writing before development starts.
Fixed-scope prototype
A defined slice of the service, built and priced as one piece of work. The usual first step when the business case needs proving internally.
Fixed-price build
An agreed scope against a written architecture, delivered for an agreed price. Best when the requirements are already clear.
Build then retainer
We build the service, then operate and improve it under a monthly agreement covering monitoring, updates and model maintenance.
White-label product
The service runs under your brand, your domain and your pricing, with MLPivot invisible to your customers.
Questions buyers ask
How reliable is the seasonal outlook?
Weaker than people want it to be, and we report it rather than dressing it up. Seasonal precipitation forecast skill is genuinely useful in some tropical regions with strong teleconnections and close to negligible across much of the mid-latitudes. The API returns measured skill by region and lead time, and labels low-skill outlooks as scenarios. A confident six-month rainfall forecast for northern Europe would be a fabrication.
Is this a hydrological model?
No. It is a composite indicator built from observations and published registers, with its components visible. A proper groundwater or catchment model requires aquifer geometry, abstraction records and calibration that are not publicly available, and building one is a multi-year specialist exercise. For portfolio screening, siting shortlists and disclosure the indicator is the appropriate tool. For a decision that turns on one aquifer, a hydrogeologist is.
What resolution does this work at?
Catchment components such as river flow and reservoir storage describe the catchment, not your field. Satellite soil moisture is kilometres-scale. So the output is honest at catchment to sub-catchment level and should not be read as a field-level water balance. We return each component with its native scale attached so nobody mistakes one for the other.
Can you use national hydrological data commercially?
Often but not always, and it varies by country more than people expect. Several agencies publish river and reservoir data openly with attribution; some require a licence for use in a commercial service; a few prohibit redistribution of the underlying series while permitting derived indicators. We check each jurisdiction during discovery and design the response schema so that what you can serve to your customers matches what you are licensed to serve.
How current are abstraction restrictions?
As current as the authority publishes, which can lag the operational position by weeks. A basin authority may be managing a developing situation well before a decree is gazetted. We return the register status with its as-of date and treat it as a lagging indicator rather than a live constraint feed, because presenting it as live would be wrong in exactly the situations that matter most.
Related blueprints
Soil Suitability Scoring API
Soil properties, crop suitability scores and land productivity estimates for land valuation, lending and agronomic planning.
Crop Yield Forecasting API
Expected yield by crop, location and season, with honest uncertainty, for agritech platforms, insurers, lenders and grain buyers.
Crop Disease Risk API
Weather-driven disease and pest risk scoring with outbreak alerts and scouting recommendations, for agronomy and input platforms.
Want this built for your platform?
Tell us what your product does today and what you want another system to be able to ask it. We will come back with an approach, a scope and a price.