CropRadar — Public data sources

Wheat & barley datasets found for the CropRadar project (UM6P / OCP): what each one is, which crops it covers, how to get it, and what it's for. License flags marked ⚠️ need verification before production use.

Guide written Oct 2026 · 37 datasets · Morocco-first

Current scope: no satellite crop-mapping for now — ground truth, tables, weather, soil & field photos

💡 Tap the blue chips to filter by what a dataset is for

Sources by data type

Where the data lives, grouped by what kind of data each source hosts.

Field / plant photos

Figshare (Moroccan parcel photos) · Kaggle (wheat disease + weed sets, login) · Mendeley Data (wheat leaf disease) · Zenodo (wheat-head counting) · Weed-AI (weed boxes) · IEEE DataPort (barley UAV, login)

Soil

SoilGrids (ISRIC) — global rasters + point API

Yield tables

FAOSTAT — Morocco wheat + barley CSVs, 1961–2024

Weather & climate

NASA POWER (free API) · Open-Meteo (free API) · CHIRPS (rainfall rasters)

Trial data (CSV / XLSX)

Dryad (aphid loads) · MELDATA / ICARDA (aphid trials) · PLOS ONE (irrigation trial files)

Mixed (code + data repos)

GitHub · Kaggle · Zenodo

Paper-only, no data

Springer barley hyperspectral (request-only) · Ethiopian barley study · Saïs tillage trial · Helsinki weed UAV

Parked for now

WorldCereal · BreizhCrops · Kazakhstan UAV · CropHarvest

Dataset guide — one paragraph each

Every dataset, in plain words. Crop chips show coverage; blue chips show what each one is for.

Ground truth — Moroccan parcels (Figshare)

14 cropsWheat (2,854 parcels)Ground truthIrrigation & water

10,383 field polygons from six Moroccan zones (Doukkala, Gharb, El Haouz, Tadla, Souss, Settat), visited between December 2023 and January 2025. Crops: wheat is the star with **2,854 parcels**, plus corn, alfalfa, potatoes, beets, peas, peanut, onions, tomatoes, sugarcane, pumpkins, fava beans, melons, watermelon, tree crops (argan), bare soil and weeds. Each field carries its irrigation type, visit date, zone and the filename of a field photo (photos live in a 16 GB companion archive we haven't downloaded yet). CC BY 4.0, fully open. This is our ground truth — the answer key everything else gets validated against.

Open

Wheat disease images

LWDCD2020

WheatDisease classification

about 8,926 field photos of wheat with leaf rust, Septoria, leaf smut, wheat streak, powdery mildew, and healthy plants. Wheat only. Shared via a Google Drive link from a third-party GitHub README; license ⚠️ unverified — don't treat as production-safe until confirmed. Good training material for a wheat disease classifier.

Verify license

Wheat disease images

Wheat Leaf Dataset (Mendeley Data)

WheatDisease classification

407 leaf photos labeled by pathologists: stripe rust (208), healthy (102), Septoria (97), taken on a wheat farm in Ethiopia in uncontrolled field conditions. Wheat only. Publicly viewable on Mendeley Data (a free login may be needed to download); license ⚠️ undeclared. Rusts and Septoria hit Moroccan wheat too, so it's a usable benchmark.

Verify license

Wheat disease images

Wheat nitrogen deficiency & leaf rust (Kaggle)

WheatDisease classification

1,159 wheat leaf photos at booting stage covering nitrogen deficiency, leaf rust, and healthy plants, with a 70/15/15 train/val/test split. Wheat only. Needs a Kaggle login; license ⚠️ unverified. Useful for telling nutrient-deficiency stress apart from disease stress.

Verify license

Wheat disease images

yilikal wheat leaf disease (Kaggle)

WheatDisease classification

wheat leaf photos in 5 classes: healthy, leaf rust, Septoria, stripe rust, Fusarium head blight, organized into train/val/test folders. Wheat only. Needs a Kaggle login; license ⚠️ unverified. The only set here covering Fusarium head blight.

Verify license

Wheat disease images

Wheat_Coccinellid (Mendeley Data)

WheatPest modeling

wheat field photos annotated for coccinellids (ladybirds), the natural enemies of aphids; built to evaluate a wheat-aphid detection study. Wheat only, plus insects. Publicly listed on Mendeley; license ⚠️ undeclared. Niche: aphid scouting imagery.

Verify license

Barley disease images

MaBaKI

BarleyPhenotyping

scanned barley kernel photos: 6 two-row malting barley varieties grown in controlled plots in Poland, scanned at 400 dpi (each raw image holds 450–480 kernels), giving over 10,000 segmented single-kernel images per variety, plus a healthy-vs-defective collection. Barley only. Publicly published; license ⚠️ statement not surfaced — verify before ingesting. Use case: variety identification and grain quality, not field disease.

Verify license

Barley disease images

Barley hyperspectral disease set (Lundstedt et al.)

BarleyDisease classification

hyperspectral signatures of barley infected with net blotch, spot blotch, stem rust, powdery mildew, scald, head blight, loose smut, common root rot, and Fusarium head blight. Barley only. Request-only from the authors — not public. Listed because net blotch, rusts and powdery mildew matter for Moroccan barley; skip for now.

Not public

Barley disease images

Ethiopian barley disease study

BarleyDisease classification

farm photos in 3 classes (healthy leaf, net blotch, leaf scald), labeled with agronomist help; their EfficientNetB0 reached 97% in the experiment. Malting barley only. Paper-only: no dataset download found. A candidate for a polite author email later.

Not public

Wheat weed images

MMIDDWF

WheatWeedsWeed detection

2,489 wheat-field photos of wheat, broadleaf weeds and grass weeds in RGB *and* depth, captured from nine viewpoints for 3D weed detection. Wheat + weeds. The paper is CC BY; the dataset/repo license ⚠️ needs verification. Origin: China.

Verify license

Wheat weed images

smallSSD

WheatWeedsWeed detection

960 wheat/weed photos with bounding boxes plus 100,032 unlabeled photos, from 8 experimental fields with varying drill rates, fertilizer and herbicide treatments. Wheat + weeds. CC BY-NC 4.0, fully open via a GitHub wrapper that downloads from Zenodo. Note the NC clause blocks commercial reuse.

Open

Wheat weed images

NarrabriWheat

WheatWeedsWeed detection

184 wheat photos with bounding boxes for annual ryegrass, turnipweed and common sowthistle. Wheat + weeds. License ⚠️ unconfirmed on the Weed-AI page — verify. Origin: Australia.

Verify license

Wheat weed images

PMDNet

WheatWeedsWeed detection

5,967 wheat-field weed photos used in a wheat weed-detection study. Wheat + weeds. Access and license ⚠️ unconfirmed; paper on MDPI.

Verify license

Barley weed images

RoboWeedMap

BarleyWeedsWeed detection

1,147 photos of barley with grass weeds (Poaceae) and mustard-family weeds (Brassicaceae), with whole-plant bounding boxes. Barley + weeds. License ⚠️ unconfirmed on the Weed-AI page — verify. Origin: Denmark.

Verify license

Barley weed images

BAWSeg

BarleyWeedsWeed detection

a large-scale UAV benchmark for barley weed semantic segmentation: repeated RTK drone surveys over 2 commercial rainfed barley paddocks in West Australia across 4 growing seasons (2020–2023), flown with a DJI Phantom 4 Multispectral (blue/green/red/red-edge/NIR plus RGB), with radiometric calibration panels. Barley + weeds. Needs a free IEEE DataPort account; license ⚠️ not surfaced. Closest agronomic match to Moroccan barley (rainfed Mediterranean-type system).

Verify license

Barley weed images

Helsinki perennial weed study

BarleyWeedsWeed detection

custom UAV with a MicaSense RedEdge 3 plus a thermal camera over barley at the 4-leaf stage; 544 photos with multispectral + NDVI + thermal + canopy-height channels. Barley + weeds. Paper-only — the dataset was never released.

Not public

Soil

SoilGrids (ISRIC)

All cropsSoil properties

global 250 m predictions of soil pH, organic carbon, clay/sand/silt, bulk density, CEC, nitrogen, carbon stocks and water variables at 6 depth intervals. All crops (wheat and barley included). CC BY 4.0, fully open: GeoTIFF/COG download, WMS, and a point-query REST API. Full Morocco coverage — our soil backbone.

Open

Yield tables

FAOSTAT (QCL)

WheatBarley+170 cropsYield prediction

country × crop × year tables of production (tonnes), area harvested (ha) and yield (kg/ha), 1961–2024, covering wheat and barley among 170+ crops and 244 countries. Fully open CSV; license CC BY-NC-SA 3.0. Our single most important Morocco series: national wheat and barley yield history.

Open

Yield tables

Crop-yield prediction dataset (Kaggle)

WheatYield prediction

28,242 rows × 7 columns derived from FAO/World Bank data: country, crop, year (1990–2013), yield (hg/ha), annual rainfall, pesticide tonnes, average temperature; 101 countries, 10 crops including wheat (barley not among the 10). Needs a Kaggle login; license ⚠️ unverified. Quick regression baseline; prefer FAOSTAT direct.

Verify license

Yield tables

GWHD 2021

WheatYield predictionPhenotyping

field photos with wheat-head annotations: 275,468 heads across train/val/test splits of 3,657 / 1,476 / 1,382 photos, for counting and detection — a yield proxy. Wheat only. Zenodo download cited in the project; verify the record's license field. Global collection.

Verify license

Yield tables

CropHarvest

Multi-cropCrop mapping

90,480 georeferenced crop labels, 65,690 of them paired with Sentinel-1, Sentinel-2, ERA5 and SRTM time series. Multi-crop (verify wheat/barley and North Africa samples on the record). CC BY-SA 4.0, fully open on Zenodo. Good ML-ready benchmark for crop-type classification from satellite. *(Parked 2026-10-05: satellite data out of scope for now.)*

OpenVerify license

Fertilizer tables

Fertilizer recommendation (GitHub)

WheatBarleyFertilizer recommendation

table of temperature, humidity, moisture, soil type, crop (including wheat and barley), N/P/K → fertilizer class. Wheat & barley. Public CSV in a repo; license ⚠️ undeclared. India-oriented classes — prototyping only, not Moroccan agronomy.

Verify license

Fertilizer tables

Kaggle S5E6 fertilizer prediction

WheatBarleyFertilizer recommendation

temperature, humidity, moisture, soil type, crop type (including wheat and barley), N/P/K → fertilizer name (Urea, DAP, NPK 10-26-26, NPK 20-20-20, ammonium sulphate, KCl, SSP). Wheat & barley. Needs a Kaggle login; competition terms; partly synthetic. Prototyping only.

Fertilizer tables

Crop water requirement (Kaggle)

Multi-cropIrrigation & water

768 records: crop type, soil texture, growth stage, temperature, humidity, water requirement. Multi-crop. Needs a Kaggle login; license ⚠️ unverified. Small — prototyping only.

Verify license

Weather & climate

NASA POWER

All cropsWeather data

daily agro-climatology for any coordinate: precipitation, temperature (mean/max/min), humidity, solar radiation, evapotranspiration, 1981–present. All crops. Fully open REST API, no key needed. Morocco-wide usable.

Open

Weather & climate

Open-Meteo

All cropsWeather data

hourly temperature, humidity, radiation, precipitation and reference ET for any coordinates back to 1940, plus a 16-day forecast endpoint. All crops. Free for non-commercial use (CC BY 4.0 attribution required); no API key. High Morocco relevance.

Weather & climate

CHIRPS

All cropsWeather data

daily precipitation at 0.05° (~5 km), 1981–near present, a satellite + station blend designed for drought monitoring in data-sparse regions. All crops. Free and open download (also on Google Earth Engine). The standard rainfall product for African ag work.

Open

Satellite crop maps

Parked for now — satellite crop-mapping is out of current scope. Kept here for reference; revisit later.

Satellite crop maps

WorldCereal (ESA)

WheatBarley (+ rye)Crop mapping

global 10 m crop-type maps from Sentinel-1/2; 2021 products map wheat separately and group wheat + barley + rye as cereals; Phase II plans separate wheat and barley products. Fully open via the project's Zenodo repository and Google Earth Engine. Full Morocco coverage — directly usable for winter-cereal season maps.

Open

Satellite crop maps

BreizhCrops

WheatBarleyCrop mapping

Sentinel-2 time series for crop-type mapping in Brittany, France, with explicit wheat and barley labels (~608,000 labeled observations; benchmark subset 13,850 wheat, 4,982 barley). Verify license and download on the repo. Method-transfer value (temporal CNN baselines), not Moroccan fields.

Verify license

Satellite crop maps

Kazakhstan UAV multispectral

WheatBarleyCrop mappingPhenotyping

5 drone flight sessions (7,200–8,300 raw aerial photos each, 15–17.7 GB per session) plus processed orthomosaics per crop: barley 12.7 GB (59 plots), wheat 11.8 GB (60 plots), with a plot metadata table (variety, sowing rate, sowing date). Wheat & barley. About 80 GB total; license and download mechanism ⚠️ need verification. A genuine paired wheat/barley UAV resource for pre-training.

Verify license

Phenotyping & stress

Wheat chlorophyll-fluorescence stress (Mendeley)

WheatPhenotyping

wheat water-stress classification data based on chlorophyll fluorescence measurements. Wheat only. CC BY 4.0 per the secondary inventory; confirm on the record. Drought-stress phenotyping is core to Moroccan wheat.

Phenotyping & stress

HyperLeaf

BarleyPhenotyping

hyperspectral barley-leaf imagery across 4 cultivars × 3 fertilizer levels. Barley only. Only a dataset-loader repo surfaced; original source, license and download ⚠️ all need verification.

Verify license

Pests

Wheat aphid loads (Dryad)

WheatPest modeling

tabular data from 120 studies (China + Europe, 1970–2017): wheat aphid loads with temperature, pesticide use, fertilization, land use and natural-enemy covariates. Wheat only. Direct Dryad download, usually no account; Dryad's default is typically CC0 — ⚠️ verify on the record. Evidence base for pest-pressure modeling.

Verify license

Pests

ICARDA RWA trials (MELDATA)

WheatBarley (host)Pest modeling

experimental CSVs: bread-wheat evaluation for Russian wheat aphid resistance, 2002 & 2003 experiments, with data dictionaries. Wheat (and the Russian wheat aphid also attacks barley — host overlap). CC-BY-NC-4.0; some files need a MELDATA login, some are request-access. High Morocco relevance: RWA is a major North African wheat/barley pest and ICARDA works in Morocco.

Irrigation & water-use trials

PLOS ONE winter-wheat irrigation trial

WheatIrrigation & water

full supporting XLSX data: precipitation, tiller number, water consumption, water-use efficiency, yield components and grain yield under varied irrigation frequency × planting pattern. Wheat only. CC BY, fully open from the article page. Water-use-efficiency evidence; China conditions.

Open

Irrigation & water-use trials

Seasonal agriculture dataset (GitHub)

Wheat7 more cropsYield predictionIrrigation & waterFertilizer recommendation

4,000 farm records (India): rainfall, temperature, humidity, sunlight, soil pH/moisture, NPK, fertilizer, pesticide, irrigation method, water used, yield t/ha, economics; 8 crops including wheat (barley not listed). Treat as synthetic — prototyping only.

Morocco-specific extras (context, mostly not data)

ICARDA genebank (Rabat)

WheatBarleyChickpeaLentilVarieties & breeding

~95,000 wheat/barley/chickpea/lentil accessions relocated to Rabat; breeding trials run at Marchouch/Settat. Open data flows through MELDATA. Context for where Moroccan trial data lives.

Morocco-specific extras (context, mostly not data)

Saïs wheat tillage × nitrogen experiment

WheatYield predictionFertilizer recommendation

0/50/100/150 kg N/ha wheat trial in the Saïs region. Paper-only; underlying trial data not downloadable.

Not public

Morocco-specific extras (context, mostly not data)

Al Moutmir / UM6P / OCP

CerealsYield prediction

876 cereal demonstration platforms reported ~20% yield gains, but publish no raw dataset.

No datasets match your search.