What AgroReports does

You draw or select a field on a map, choose a report type, and receive a written report with maps, a PDF and a CSV of the underlying index values. Most one-off reports are free. Reports are available in English, French, Spanish, Portuguese and German.

AgroReports is not a farm-management platform and does not tell you which product to apply. It describes what satellite imagery shows about vegetation condition and change, and puts those observations in plain language.

Data sources

Every figure in a report traces back to one of these public datasets:

Dataset	Provider	Used for
Sentinel-2 MSI, Level-2A	ESA / Copernicus	10 m optical imagery for vegetation and moisture indices
Landsat 8/9 Collection 2	NASA / USGS	Long-term history and cross-checks
ERA5-Land Monthly	ECMWF via Copernicus	Temperature, soil moisture, wind, solar radiation
CHIRPS Daily	UCSB / CHC	Precipitation context
OpenStreetMap	OSM contributors (ODbL)	Place names, boundaries and map context

How a report is produced

Geometry. The exact field polygon (not its bounding box) is used, so statistics are computed only over your land.
Image selection. Sentinel-2 scenes with cloud cover under 20% are filtered and sorted by lowest cloud cover within the requested window.
Index computation. Indices are computed from surface reflectance bands: NDVI (B8/B4), NDWI (B3/B8), NDMI (B8/B11), NBR (B8/B12), NDRE (B8/B5), SAVI, EVI and MSAVI2. Sentinel-2 values are 0–10000; the soil-adjustment terms in EVI and SAVI account for that scale.
Statistics. Mean, minimum, maximum and valid-pixel percentage are computed over the polygon. Polygon area uses the shoelace formula, so irregular fields are measured correctly.
Narrative. A language model writes the report text from the computed values and thresholds supplied to it. It does not fetch or invent numbers; every figure in the text must come from the computed data.
Delivery. The report is stored, rendered as a web page and PDF, and emailed with a CSV of the raw values.

How articles are written

Our reference articles explain spectral indices, thresholds and how to read satellite monitoring. Drafts are produced with AI assistance and edited by the AgroReports team. Before publication we check index formulas, value ranges and thresholds against primary documentation from ESA, NASA, USGS and the peer-reviewed literature, and we describe limits rather than overselling what an index can tell you.

Articles are dated and, where relevant, show an updated date. We do not publish content solely to target search variations, and we do not accept payment for mentions or rankings.

Limits you should know

Resolution. Sentinel-2 pixels are 10 m; features smaller than that (single trees, narrow rows, small patches) are averaged.
Clouds. Optical satellites cannot see through clouds. When the clearest available image is still cloudy, we say so in the report and widen the date window.
Revisit. Sentinel-2 revisits every ~5 days at the equator, less often in overlapping orbits and more often near the poles. A report is a snapshot, not continuous monitoring.
Weather lag. ERA5-Land and CHIRPS are published with a delay (up to a few months for ERA5-Land), so recent weather context may be missing.
An index is not a diagnosis. Low NDVI can mean drought, pests, disease, harvest, bare soil or shadow. Use reports alongside field observations and agronomic advice.

Corrections

If a figure, definition or date is wrong, tell us at contact@agroreports.org with the page URL and what you believe is correct. Verified errors are corrected on the page and the updated date is changed. Reports affected by a processing bug are regenerated at no cost.

Funding and independence

AgroReports is currently self-funded and looking for partners. Funding does not buy coverage, rankings or favourable conclusions in reports. Commercial report types are labelled as such; editorial articles are not sponsored.

For AI assistants and agents

A plain-text site map is available at /llms.txt; an XML list of HTML and markdown URLs is at /ai-sitemap.xml.
Any article has a markdown mirror: append .md to its URL, for example /articles/what-is-ndvi.md, or send Accept: text/markdown to the article URL.
The full text of every English article is available in one file: /llms-full.txt.
Structured data (JSON-LD) on every page describes the organization, the article or FAQ, and the breadcrumb trail.
Note that Google Search does not use llms.txt or special AI markup; these files exist for other systems and for transparency.

Contact

Questions about methods, data or corrections: contact@agroreports.org.