AGRICULTURE · HOW THE API WORKS

A coordinate on a field.A crop-viability answer.

For insurers, lenders, agri-input companies and food supply teams: replace spot-checks, satellite-tile pipelines and subjective field reports with one objective 0–100 score per coordinate — computed from live Sentinel-2 imagery, terrain and climate reanalysis.

WHAT IT CAN DO

What the API does for agriculture.

Score any field on Earth

No shapefile, no polygon, no farm registry. A latitude/longitude pair is enough to get a viability read on a parcel you have never visited.

Detect moisture and heat stress

NDMI moisture reserve plus climate reanalysis expose dry-down and heat-degree build-up before it shows up in yield.

Measure real vegetative vigour

NDVI is sampled on the actual Sentinel-2 pixel at 10 m, with cloud and shadow pixels rejected rather than smoothed over.

Rank portfolios consistently

Because the model is deterministic, thousands of parcels can be ranked on the same axis and re-scored on the same axis next season.

THE FLOW

Coordinate to decision, step by step.

Every agriculture request runs through the same five stages. Nothing is cached from a neighbouring tile and nothing is invented — if a pixel is unusable, the response says so.

  1. 01
    Input
    lat / lng + mode: agriculture
    • Single HTTP POST
    • No shapefile or polygon
    • No upload, no GIS stack
  2. 02
    Observe
    Live pixel + climate sampling
    • Sentinel-2 red / NIR / SWIR at 10–20 m
    • SCL cloud & shadow gate
    • Climate reanalysis at the point
    • DEM five-point stencil
  3. 03
    Derive
    Indices and stress signals
    • NDVI vigour · NDMI moisture
    • NDWI standing water penalty
    • NDBI sealed-surface penalty
    • Heat-degree and drought build-up
  4. 04
    Fuse
    Weighted, deterministic model
    • Surface weighted highest
    • Climate second
    • Terrain and georisk as modifiers
    • No random seeds
  5. 05
    Output
    0–100 score + band + components
    • Machine-readable JSON
    • Component sub-scores exposed
    • Same coordinate → same answer
THE DETERMINISTIC PROMISE
Same coordinate → same score. Every request. Every week. Every quarter.

No random seeds. No hidden drift. If you got a 74 last Tuesday, you'll get a 74 next Tuesday — unless the ground itself has changed. That is what makes the score defensible in an underwriting memo, a planning submission, or a board pack.

INPUTS

Three fields in. That is the whole input.

No shapefiles. No polygons. No uploads. No GIS stack to maintain.

latnumber

Latitude in decimal degrees, −90 to 90. The exact point of the field, plot centroid or sample location.

lngnumber

Longitude in decimal degrees, −180 to 180.

mode"agriculture"

Tells the model which decision you are making. Agriculture mode weights surface vigour and climate most heavily.

REQUEST
POST /v1/land-intel/assess
{
  "lat": 12.9716,
  "lng": 77.5946,
  "mode": "agriculture"
}
RESPONSE
200 OK
{
  "score": 74,
  "classification": "HEALTHY",
  "band": "Productive",
  "mode": "agriculture",
  "components": {
    "surface": 78,
    "climate": 71,
    "terrain": 82,
    "georisk": 90
  },
  "observations": {
    "ndvi": 0.62, "ndmi": 0.21,
    "ndwi": -0.18, "ndbi": -0.09,
    "elevation_m": 912, "slope_pct": 1.8
  },
  "generated_at": "2026-07-11T12:04:22Z"
}
DATA SOURCES

Enterprise-grade data. Open licences. Zero cost.

Every agriculture score is built on the same public Earth-observation feeds that power commercial GIS stacks sold for tens of lakhs to crores per annum. We expose them directly through one simple HTTP call.

Sentinel-2 L2A optical imagery

10–20 m multispectral surface reflectance from ESA's Copernicus Sentinel-2 constellation, served as cloud-optimised GeoTIFFs via the AWS Earth Search STAC catalogue. Every request samples the actual pixel at your coordinate for red, green, NIR and SWIR bands.

Source: ESA / Copernicus · AWS Earth Search STAC · 10–20 m · keyless
NDVI — vegetation vigour
NDWI — open water / wetness
NDBI — built-up surface
NDMI — canopy / soil moisture
Copernicus DEM global elevation

GLO-90 digital elevation model from the Copernicus DEM programme, sampled through a five-point stencil so slope and local relief are measured, not assumed. The centre point plus four surrounding points gives a true local gradient.

Source: Copernicus DEM GLO-90 · Open-Meteo elevation API · 90 m · keyless
Elevation (m)
Slope gradient (%)
Local relief (m)
Scene Classification Layer quality gate

Every sampled pixel is checked against the Sentinel-2 SCL. Cloud, cloud shadow, thin cirrus, saturated and no-data pixels are rejected outright — not smoothed over — so the score only sees ground truth.

Source: Sentinel-2 SCL band · QA check per pixel · no cloud interpolation
Cloud rejection
Shadow rejection
Saturated/no-data rejection
Climate reanalysis & hazard catalogue

Open-Meteo historical climate reanalysis provides temperature, precipitation and humidity baselines at the coordinate. The 5-year seismic catalogue supplies catalogued ground-motion frequency and peak magnitude within 200 km.

Source: Open-Meteo climate API · 5-year global seismic catalogue
Temperature / precipitation
Drought / stress signals
Seismic events 5 yr
THE DETERMINISTIC PROMISE
Same coordinate → same score. Every request. Every week. Every quarter.

No random seeds. No hidden drift. If you got a 74 last Tuesday, you'll get a 74 next Tuesday — unless the ground itself has changed. That's what makes the score defensible in an underwriting memo, a planning submission, or a board pack.

Same coordinate → same score → same decision, every time.
THE OUTPUTS

What you get back.

A plain-English report anyone can read. No remote-sensing degree, no GIS software, no manual report.

1
Overall score
0–100

One simple number that tells you how suitable this point is for crops.

High score = healthy, productive land. Low score = stressed or risky.
2
Land condition
Live satellite read

What the ground actually looks like right now — green cover, moisture, and surface type.

No site visit needed. No shapefile. No manual report.
3
What it means
Plain label

A simple label like Healthy, Productive, or Stressed, plus a recommended action.

A loan officer or insurer can act on it instantly without reading a technical report.
4
Why it scored that way
Component breakdown

See the four drivers behind the score — vegetation, climate, terrain, and georisk.

Explain a low score to a farmer: the canopy, not the weather, is the problem.
5
Decision-ready
Use in your system

The result comes as clean JSON that any software can read.

Plug it into credit rules, insurance payouts, farm-advice apps, or supply-chain dashboards.

The parameters behind the score.

Every response is one score plus the measured parameters that produced it. Which parameters come back depends on the vertical you query.

Vegetation vigour
NDVI

How much live, photosynthesising canopy is on the ground.

High NDVI means a dense, healthy crop stand. A drop mid-season flags stress, pest damage or failed sowing.
Plant & soil moisture
NDMI

Water held in the canopy and topsoil.

Falling NDMI while NDVI is still high is the earliest warning of drought stress, before the crop visibly browns.
Standing water
NDWI

Open or ponded water on the parcel.

Persistent NDWI means waterlogging or flooding — the difference between a dry-spell claim and a flood claim.
Built-up interference
NDBI

How much of the parcel is roads, roofs or bare impervious surface.

Rising NDBI on farmland means conversion or encroachment — the arable area is shrinking.
Thermal & rainfall regime
NASA POWER climate

Temperature envelope and precipitation over recent seasons.

Separates a bad year from bad land, and sets which cultivars are realistic.
Terrain
Copernicus GLO-90 DEM

Slope and relief across the parcel.

Steep ground limits mechanisation and drives irrigation runoff losses.

What those outputs do in real life.

Parametric payout without a loss adjuster

Re-score affected coordinates after a weather event. When the score crosses the agreed band, the payout fires — no site visit, no dispute over what the field looked like.

Credit decisions in the branch

A loan officer enters a plot location and gets a viability band on screen. Strong land clears instantly; marginal land is referred with the exact component that caused it.

Supplier land verification

Score supplier farms quarterly and surface degrading or converted land before it becomes a compliance failure in a sourcing audit.

Input and advisory targeting

Rank dealer-network parcels by potential so premium seed, fertiliser and agronomy hours go where they can actually pay back.

LIVE DEMO

Pick a coordinate. See the real output.

These are live calls to the production agriculture engine — real Sentinel-2 imagery, real elevation, real climate and seismic records. Nothing here is mocked.

POST /v1/land-intel/assess · mode: agriculture
fetching…
Rootfifteen score
/ 100

Intensively farmed alluvial plain

Why it scored that way
Climaten/a
Terrainn/a
Geo-riskn/a
Surfacen/a
What the satellite measured here
Vegetation vigour (NDVI)

How green and actively growing the crop stand is

Crop & soil moisture (NDMI)

Water held in the canopy and topsoil

Standing water (NDWI)

Ponding, flooding or open water on the parcel

Built-up interference (NDBI)

Roads, roofs and bare impervious surface eating arable area

Annual rainfall

Water the parcel receives in a year

Mean temperature

Growing-season heat load

Slope

Steepness — affects erosion and machinery

Imagery date

When the satellite last saw this parcel

Fused live from Sentinel-2 L2A optical imagery, Copernicus GLO-90 elevation, NASA POWER climate reanalysis and the USGS seismic catalogue — returned in a single JSON response, typically in about a second.

TRADITIONAL VS ROOTFIFTEEN

How this was done before — and what changes.

Every row is a task agriculture teams already do today. The left column is the traditional method. The right column is the same task with one API call.

Getting a viability read
Send an agronomist to the plot and wait for a visit slot.
One API call returns a viability score in under a second.
Coverage
Only farms your field team can physically reach.
Any coordinate on Earth, including plots you will never visit.
Crop-stress detection
Reported after yield loss is already visible.
NDVI and NDMI expose stress while the canopy is still standing.
Same coordinate → same score → same decision, every time.
WHAT THE NUMBER MEANS

Ten bands. One axis. Zero ambiguity.

The 0–100 axis is partitioned into ten calibrated bands, each mapped to a defensible agriculture decision — identical across every coordinate on Earth.

90–100Prime
Optimal soil moisture, balanced thermal regime, strong vegetative vigour (NDVI > 0.75).
Greenlight for high-value cropping. Eligible for the cleanest parametric insurance tiers.
80–89Strong
Healthy canopy with minor heat or moisture excursions across the season.
Plant with standard agronomy. No mitigation required.
70–79Productive
Mostly favourable, with isolated dry-down or thermal stress windows.
Suitable for staple crops. Schedule one or two extra irrigation passes.
WHERE TEAMS USE IT

The operational workflows Rootfifteen replaces.

Parametric crop insurance
Lending & farm credit
Input & advisory targeting
Sourcing & supply traceability
Commodity trading & origin risk
Government food security & land-use policy

Parametric insurance payouts

Re-score affected coordinates after weather events and trigger payouts when parcels drop below agreed bands.

Farm credit underwriting

Auto-approve high-quality parcels, flag marginal applications, and decline weak land with a data-backed explanation.

Supply chain land verification

Score supplier farms quarterly and surface degraded or protected-land risk before it becomes a compliance problem.

Precision input targeting

Rank dealer-network parcels by potential so premium products and advisory time go where they can actually perform.

WHERE TO IMPLEMENT

Drop the API into the systems that already run agriculture decisions.

Rootfifteen is one HTTP endpoint. That means it fits inside any software that already touches agriculture workflows — no new dashboard, no new map stack, no internal GIS team required.

Crop insurance platforms
Underwriting & claims

Bind policies, score parcels at quote time, and trigger parametric payouts after weather events without loss adjusters.

Farm credit apps
Loan origination

Auto-approve strong land, flag marginal applications, and explain declines with component sub-scores.

Agronomy dashboards
Advisory targeting

Rank parcels by vigour and moisture so agronomists and input dealers spend time where it pays back.

Supply-chain traceability tools
Land verification

Score supplier farms quarterly and catch degraded or converted land before audits do.

Government food-security systems
Policy & monitoring

Monitor crop health across districts and prioritise drought relief, irrigation, and extension spend.

Reinsurance pricing engines
Portfolio pricing

Build a reproducible, auditable land-quality factor into treaty pricing and risk selection.

INTEGRATION NOTE
One endpoint, one coordinate, one decision-ready score.

Every agriculture call returns a 0–100 score, a calibrated band, a machine-readable classification, and the component sub-scores that explain it. Wire it into your underwriting engine, your site-screening workflow, or your command map exactly like any other REST API.

NEXT STEP

Now try it in your own stack.

Tell us what you are building and we will send you endpoint docs, a key, and limits that fit your use case.