Farmland Atlas: land value, rent and risk by county
Every county in the fifty states on the public record: land value, rent, yield, the insurance record, drought, night heat, water and who farms it. Pick a layer, then tap a county.
Dimmed layers are not yet measured. They fill in after the first data run.
Sources and vintages
The national picture
Every county in the Atlas, added up. Open a section for the table behind it.
Farm loans: what ag bankers in each Fed district report on repayment and renewals, on the by-state page.
The national figures, layer by layer
Across the Atlas · built 2026-10-19 · 3141 counties in 50 states.
July nights: 2025 of 3135 counties are warming since 1976 with the 95% interval clear of zero, 0 cooling, 1110 with no clear direction; the median trend is +0.45 F per decade. In 2178 of 3140 counties the 2010s were the warmest complete decade since the 1900s (the 2020s are not complete and are not counted).
Each county has at most 51 Julys since 1976, and a single county's trend has to be large to clear its own 95% interval. Fitted on every July since 1895 instead, 2615 of 3135 counties are warming with the interval clear of zero.
Water: 608 of 2161 counties irrigate a quarter or more of their harvested cropland and 412 irrigate half or more; the median county irrigates 4%. In 886 of 2663 counties nine-tenths or more of the irrigation water is pumped from the ground (USGS 2015: crop irrigation where USGS splits it out, 1672 counties; all irrigation including golf courses in the other 991).
Loss record: $210.6 billion of crop insurance indemnities in closed crop years since 1989 across 2840 counties (nominal dollars; the crop year still being paid is left out; pasture and rangeland rainfall-index policies excluded). Heat and drought share by period: 1989-1999 27% · 2000-2009 42% · 2010-2019 42% · 2020-2025 46%. Paid for failure of irrigation supply or equipment: $3.4 billion.
Irrigated against dry cash rent: 475 counties publish both with enough paired years and enough dry rent to test (97 more are not tested: dry rent under $25). Measured as irrigated rent over dry rent, so general rent inflation cancels out, the multiple is rising in 77 and falling in 54, with the 95% interval clear of zero; by chance alone about 12 of 475 would be called each way. Falling fastest: Box Butte County, NE: 8.10 times dry rent in 2013 to 5.34 in 2026; Thurston County, WA: 6.12 times dry rent in 2009 to 2.20 in 2026; Lawrence County, AR: 3.31 times dry rent in 2012 to 1.67 in 2026. Rising fastest: Sutter County, CA: 4.00 times dry rent in 2010 to 12.11 in 2023; Yolo County, CA: 5.09 times dry rent in 2009 to 10.23 in 2026; Independence County, AR: 1.84 times dry rent in 2009 to 4.21 in 2023.
Loss ratio: across 2890 counties RMA has paid $210.6 billion of indemnity on $246.4 billion of premium in closed crop years since 1989, a loss ratio of 0.85; in 826 of the 2475 counties with $1 million or more of premium, indemnities have exceeded premium over the whole record. By period: 1989-1999 1.02 · 2000-2009 0.79 · 2010-2019 0.86 · 2020-2025 0.86. The crop year still being paid is left out; rainfall-index pasture policies are excluded.
Land value, 2022 census: the median county reports $4,382 an acre for land and buildings (range $325 to $2,551,600, 3067 counties). Non-irrigated cash rent as a share of that value is published for 933 counties, median 2.48%; under 2% in 263, over 4% in 12.
CRP: 25,753,630 acres enrolled across 3110 counties in 2025; 2,144,307 acres expire in the three fiscal years from 2027. In 382 of 1923 counties CRP is a tenth or more of the cropland base.
Drought: summed over 3132 counties, the year with the most county-weeks with at least half the county in severe drought or worse since 2000 was 2012 (41,841 county-weeks). In 2025, the latest full year, 121 counties spent half the year or more in that condition.
Solar and wind, Form 860 2025: 154,264 MW of solar in 1326 counties and 159,732 MW of wind in 518; 43,643 MW of batteries. Proposed to EIA: 140,162 MW solar and 28,250 MW wind in 730 counties (825 counties counting proposed batteries).
Wells and rights. Nebraska: 267,850 registered wells, 105,339 of them irrigation wells not decommissioned, median county median depth 197 ft. Kansas: 111,420 points of diversion, 51,917 active irrigation, median county median priority year 1977.
Rented vs owner-farmed, 1997 to 2022
Foreign-held farmland, 2010 to 2024
USDA counts 46.3 million acres of U.S. agricultural land held by foreign investors at the end of 2024, 3.6% of privately held agricultural land, and nearly double 2010, spread across 2,559 counties. The biggest holder is Canada, not China, and nearly half of it is forest, not corn ground. Every number below comes from USDA’s disclosure files.
China: 247,659 acres, 0.5% of foreign-held acres (USDA AFIDA, Dec 31, 2024). Who is behind it: 2.9 million of the 47.2 million acres in USDA’s detail files (6.2%) name no foreign country (USDA codes 998 and 999). The files also say which holdings are owned and which are leases of 10 years or more. We have not added up that split yet, so there is no owned-outright number here; farmdoc daily (Sep 23, 2026) has one.
46.3 million acres of U.S. agricultural land held by foreign investors as of Dec 31, 2024 (USDA AFIDA report) · the detail file’s rows sum to 47,241,820, and the county and country figures here use the rows · 2,559 counties · up from 24,841,588 row-summed acres in 2010 · largest holder country: Canada · 47% of the acreage is forest land · source: USDA FSA AFIDA disclosure filings · page data refreshed 2026-07-18
The county map is the Atlas map above: show foreign-held acres on it, then pick a county for its record here. AFIDA is a self-reporting law, so a county with none reported may hold none or may not have filed.
Pick a county on the map above to see its foreign-held acres by year, the holder countries and what kind of land it is.
Gaps at 2015 and 2021: USDA’s downloadable detail files skip those years, so the line breaks there instead of guessing across the gap. In some counties 2022 is a one-year jump that falls back in 2023: Autauga County, AL reports 61,371 acres in 2020, 104,884 in 2022 and 58,741 in 2023. The files do not say why.
USDA Farm Service Agency, Agricultural Foreign Investment Disclosure Act (AFIDA) detailed holdings files, 2010–2024, via fsa.usda.gov. Our sums of the rows run above USDA’s published headline (47.24 million vs 46.3 million for 2024). USDA also hosts its own ArcGIS dashboard. Data years 2015 and 2021 are not in USDA’s downloadable file set. Page data refreshed 2026-07-18; AFIDA updates annually.
Where July nights are warming, by state
Each county has one July a year since 1976, so its own trend has to be large before the 95% interval clears zero. The whole-record fit, every July since 1895, has more years; in some states it finds warming in more counties and in others in fewer, because the early 1900s were warm in some regions. Both columns count counties whose own interval clears zero.
| State | Counties | Warming since 1976 | Warming since 1895 | Median trend since 1976, °F a decade |
|---|
What ground rents for against night heat and water
Non-irrigated cash rent for every county against one exposure at a time. Most of the pattern in the chart is north to south: warm nights and cheap ground both sit in the south. The county you have open is marked.
| Irrigated | Counties | Median dry rent |
|---|
A table, not a chart: the median county irrigates 4% of its cropland, so a chart piles up against zero.
Do warmer nights show up in the yield record?
-
| Night-heat measure | Counties | r | Chance of an r this large with no real link | Yield trend, coolest → hottest fifth, bu/ac a year |
|---|
The yield trend runs from 2008, so it measures genetics and management as much as weather. Eight tests are run and all eight are shown; the chance column is multiplied by eight. The rows within a state are the ones that count, because across the country the south is both warmer and different in every other way. How the test works.
What the Atlas is
What this puts on one page for each county: how warm its July nights are and which way they are going; how much of its cropland only works with a pump and where that water comes from; what has destroyed crops there since 1989 and whether the insurance book has paid out more than it took in; what the census says the land is worth and what rent buys of that; how much sits in CRP and when those contracts end; how many weeks of each year since 2000 it spent in drought; how much solar and wind is on the ground and proposed; and, where a state publishes the register, its wells and water rights. Each is its own layer. I never add them together, because a combined score hides which problem the ground has.
How it is measured
Every number comes from a public federal or state file, named with its year. A county with nothing published shows a dash and the reason. The crop year still being settled is left out of every insurance figure and shown apart.
Questions
Why are heat and water graded separately instead of combined into one score?
Because a combined score hides which problem a county has. A county with cool nights and a falling aquifer is a different purchase from a county with hot nights and rain-fed ground, and one number would call them the same.
What does the July night temperature measure?
The monthly mean of each day's minimum temperature for the county, in degrees Fahrenheit, from NOAA NCEI's nClimDiv county dataset, which runs from 1895. July is the silking month for corn across most of these states. Extension research puts the harm line for corn at nightly lows above 70 F, so a July whose average low was at or above 70 F is counted as a hot July.
Where does the crop insurance loss record come from?
USDA Risk Management Agency's Summary of Business cause-of-loss files, one per crop year since 1989. Each indemnity is grouped by the peril that caused it: heat and drought, excess moisture and flood, hail, wind, freeze and frost, irrigation failure, price decline, and other. Dollar totals and shares come from these files. The loss ratio comes from RMA's county Summary of Business instead, which carries premium on every policy; the cause-of-loss files carry premium only on policies that had a loss.
How does irrigated rent compare with dry rent?
Irrigated cropland cash rent divided by non-irrigated cash rent, same county, same year, from the USDA NASS Cash Rents Survey. A multiple rather than a dollar gap, because rents rose almost everywhere and a dollar gap widens even when nothing about the water changed. Irrigated rent also pays for the well, pump and pivot, so the multiple is not the price of water alone. Its direction is called only where the 95% interval of the fitted trend excludes zero and dry rent is at least $25 an acre.
What is a crop insurance loss ratio and why does it matter for land?
Indemnities paid divided by total premium, from USDA RMA's Summary of Business county files, every policy since 1989 except the rainfall and vegetation index pasture plans (PRF), which are counted separately. The open crop year is left out. At 1.00 the county's insurance book paid out what it took in; above it, claims exceeded the whole premium, the farmer's share and the federal subsidy together. Insurers, lenders and anyone pricing a lease look at it because it is the longest continuous record of how often crops in that county have failed against what was expected, with the dollars attached.
Where does the land value come from, and is it an appraisal?
The USDA Census of Agriculture, which asks every operator the market value of their land and buildings and publishes the average per acre by county for 2012, 2017 and 2022. It is not an appraisal and not a sale price; it is what operators reported, buildings included. This page shows it as published and divides the county's non-irrigated cash rent by it.
Why are irrigation wells shown for some states and not others?
Because only some states publish an open well register with an irrigation code that can be read and checked. The Atlas reads Nebraska's registered wells and Kansas's WIMAS water rights, and from September 2026 the registers of Minnesota, Michigan, Ohio, Wisconsin, Oklahoma, California, Arizona, Idaho, Oregon and Nevada, each where that month's pull succeeds. Registers differ in what they count and how far back they go, so compare a county with its own state. A county in any other state reads "no open register read yet".
How much American farmland does China own?
Not much: 247,659 acres at the end of 2024, about 0.5% of the foreign-held acres in USDA's AFIDA detail files (our sum of the rows; Hong Kong is listed on its own, 144,090 acres). Canada holds 16.1 million. Chinese-connected holdings are politically sensitive and several states have restricted them, but by acreage they are a small slice. Since 2023 the filings carry specific flags for secondary interests from China, Iran, Russia, and North Korea.
Is 46 million acres a lot?
USDA puts it at 3.6% of privately held agricultural land. In USDA’s detail files it has nearly doubled since 2010, from 24.8 million row-summed acres to 47.2 million, and that growth is the more important number. About 22 million of the 47 million acres are forest.
Why is so much of it forest?
Because global timber companies and investment funds own large tracts of U.S. timberland, and forest counts as agricultural land under AFIDA. Roughly 22 of the 47 million acres are forest, the single biggest reason raw AFIDA totals get misread as "foreign corn ground."
My county shows nothing. Does that mean no foreign ownership?
It means no disclosed holdings. AFIDA relies on self-reporting with civil penalties for failure to file; federal auditors have found compliance gaps. Absence of a filing is evidence, not proof.
Where can I get the raw data?
USDA FSA publishes the detailed holdings files annually, linked below. The Atlas adds them up without changing any figure.
Sources
NOAA NCEI · USDA NASS · USDA RMA · USGS · U.S. Drought Monitor · EIA · USDA FSA · Nebraska DNR · Kansas DWR. Sources and methods
Made by Sigurd Lindquist, CCA, Chetek, Wisconsin.