Grains production: area, yield and output by country (USDA FAS PSD)

Commercial use OKFresh

Download sample CSV Download sample JSONSample rows only (up to 20) — not the complete dataset.

Area harvested (thousand ha), yield (MT/ha) and production (thousand MT) for 9 grain commodities (wheat, rice milled, corn, sorghum, barley, millet, oats, rye, mixed grain) across ~160 countries plus aggregates, marketing years 1960/61 to the present, from USDA FAS PSD Online (https://apps.fas.usda.gov/psdonline/downloads/psd_grains_pulses_csv.zip). Each row is one (commodity, country, marketing year, attribute): area harvested in thousand hectares, yield in metric tons per hectare, production in thousand metric tons (rice also carries rough production and milling rate). Method: USDA's official PSD forecasts, reviewed monthly by an interagency committee chaired by the World Agricultural Outlook Board; historical rows carry the vintage (vintage_year, vintage_month) of the PSD release that last revised them. Units: thousand ha / MT per ha / thousand MT. Caveats: marketing years start in different calendar months per country/commodity — never compare market_year labels across countries as calendar years; EU member states appear only as EU aggregates (EU15 before 1999, EU from 1999); the latest 1-2 marketing years are forecasts subject to revision; country codes are FIPS 10-4 mapped to ISO alpha-3 (see connector docs for historical entities). Coverage: worldwide, 1960/61 to present, yearly (marketing-year) granularity. Provenance: USDA FAS PSD Online keyless bulk CSV, U.S. public domain — please cite the U.S. Department of Agriculture, Foreign Agricultural Service. Primary key: (commodity_code, country_code, market_year, attribute). Join keys: country_code (ISO 3166-1 alpha-3), market_year. Cadence: refreshed monthly (WASDE cycle); new marketing-year data typically appears within days of the monthly release.

Source: USDA Foreign Agricultural Service
Rows: 131,539
Columns: 10
Source cadence: Monthly
Last refreshed: Sep 23, 2026
Theme: economy

Column	Type	Description
commodity_code	string	FAS 7-digit commodity code, zero-padded (e.g. 0410000 = Wheat, 0440000 = Corn).
commodity	string	Commodity description as published by FAS.
country_code	string	ISO 3166-1 alpha-3 country code, resolved from the provider's FIPS 10-4 code via the country name (FIPS codes collide with ISO alpha-2, so the code itself is never used directly). Historical entities and aggregates keep stable documented codes: EU (European Union), EU15 (EU-15), YUG (former Yugoslavia), YUG_FRY ('Yugoslavia (>05/92)' file label), SUN (USSR), YMD (South Yemen), CSK (Czechoslovakia), SCG (Serbia and Montenegro).
country_name	string	Country/aggregate name as published by FAS.
market_year	integer	Local marketing year of the observation (e.g. 2026 = the 2026/27 marketing year). Marketing years start in different calendar months per country and commodity, so market_year labels must not be compared across countries as calendar years.
vintage_year	integer	Calendar year of the PSD release in which this row's estimate was last revised (provider's Calendar_Year column; observed to track the monthly WASDE release cycle).
vintage_month	integer	Calendar month (1-12) of the PSD release in which this row's estimate was last revised (provider's Month column).
attribute	string	Canonical PSD attribute (area_harvested, yield, production, rough_production, milling_rate, beginning_stocks, imports, ty_imports, ty_imports_from_us, total_supply, exports, ty_exports, domestic_consumption, feed_consumption, fsi_consumption, ending_stocks, total_distribution). TY- prefixed attributes are on the October-September trade year; the rest follow the local marketing year.
unit	string	Canonical unit: 1000_ha = thousand hectares, 1000_mt = thousand metric tons, mt_per_ha = metric tons per hectare.
value	float	Attribute value in the stated unit. The latest 1-2 marketing years are forecasts subject to revision.

First 10 sample rows — a preview, not the complete dataset.

commodity_code	commodity	country_code	country_name	market_year	vintage_year	vintage_month	attribute	unit	value
0410000	Wheat	AFG	Afghanistan	1,960	2,006	7	area_harvested	1000_ha	2,230
0410000	Wheat	AFG	Afghanistan	1,960	2,006	7	production	1000_mt	2,279
0410000	Wheat	AFG	Afghanistan	1,960	2,006	7	yield	mt_per_ha	1.02
0410000	Wheat	AFG	Afghanistan	1,961	2,006	7	area_harvested	1000_ha	2,230
0410000	Wheat	AFG	Afghanistan	1,961	2,006	7	production	1000_mt	2,279
0410000	Wheat	AFG	Afghanistan	1,961	2,006	7	yield	mt_per_ha	1.02
0410000	Wheat	AFG	Afghanistan	1,962	2,006	7	area_harvested	1000_ha	2,341
0410000	Wheat	AFG	Afghanistan	1,962	2,006	7	production	1000_mt	2,279
0410000	Wheat	AFG	Afghanistan	1,962	2,006	7	yield	mt_per_ha	0.97
0410000	Wheat	AFG	Afghanistan	1,963	2,006	7	area_harvested	1000_ha	2,341

Point any LLM at the metadata endpoint — the documentation above is machine-readable too (JSON-LD + Croissant).

cURL

curl "https://datazimuts.com/v1/datasets/usda_fas/usda_fas_grains_production" | jq '{title, rows, columns_count, license}'

Python

import requests

ds = requests.get("https://datazimuts.com/v1/datasets/usda_fas/usda_fas_grains_production").json()
print(ds["title"], ds["rows"], "rows")

# Sample rows for an LLM context window
for row in ds.get("sample_rows", [])[:5]:
    print(row)

API endpoint: https://datazimuts.com/v1/datasets/usda_fas/usda_fas_grains_production

Tip: fetch /llms.txt for the full machine-readable catalog.