United States

Description of cropping systems, climate, and soils

Annual crop production area in the United States of America (USA) occupies 97 million ha. Major crops are maize, soybean, wheat, and rice, which account for 87% of total crop area, and the country is one of the largest producers and exporters of these crops. During the last decade (2005-2014), the USA accounted for 38 and 35% of global annual maize and soybean productions, respectively.

Table 1. Average (2005-2014) total production, harvested area, and average yield of maize, soybean, and wheat in USA. Source: FAOSTAT


Total production (Mt)

Harvested area (Mha)

Average yield



















Maize-soybean crop systems

More than 85% of the maize and soybean is produced in the north-central region known as the ‘Corn Belt', where continuous maize (˜35%) and 2-year maize–soybean rotation (˜65%) are the dominant cropping systems. The central and eastern regions of the Corn Belt have favorable climate for rainfed crop production, with total annual rainfall that ranges from 800 to 1100 mm. The western edge of the Corn Belt has less rainfall and includes the eastern Great Plains states of North Dakota, South Dakota, Nebraska, and Kansas, where irrigated agriculture accounts for more than 50% of the total maize and soybean production. Soils are generally deep, fertile, rich in organic matter, and have large water holding capacity in the rootable soil depth (i.e., > 200 mm of plant available water). Dominant soils correspond to the Mollisols and Alfisols orders. A detailed description of weather, soil and cropping systems can be found in Connor et al (2011) and Grassini et al. (2014). In wet years, transient early-season waterlogging is likely in soils with moderate-to-low infiltration rates in the central and eastern regions of the Corn Belt, where subsurface tile drainage is currently used on about one-third of total cropland to mitigate this problem (Sugg, 2007). In the past 60 years, there has been a shift from continuous maize under conventional tillage to a 2-year maize–soybean rotation under some form of conservation tillage (defined as any tillage method that leaves =30% of soil surface covered by plant residues). Of total area sown with maize and soybean, a respective 52% and 75% is under conservation tillage (Horowitz et al., 2010). Almost all soybean area in the Corn Belt is sown with transgenic cultivars and the vast majority of those cultivars possess a transgene that provides tolerance to the glyphosate. The use of glyphosate-resistant cultivars has not only provided near-total weed control (though glyphosate-resistant weeds are now increasingly appearing), but it has also accelerated producer adoption of conservation tillage- or no-till practices as well as narrowing of row spacing. Use of transgenic Bt maize with resistance to a number of damaging insect species is also widespread.

Rice crop systems

US rice production can be split into three ecological zones, the Upper Sacramento Valley (CA), the Gulf Coast (TX, LA, and parts of MS), and the Mississippi River Valley (AR, MO, and parts of MS) (Livezey and Foreman, 2004). Although each of these regions has unique climate characteristics, the climate in the two Southern regions is generally humid with a small range of diurnal temperature variation compared to the climate in CA, which is generally arid/semi to arid with large diurnal temperature variation. CA growers use primarily medium-grain temperate japonica rice varieties, whereas growers in the Southern regions primarily use long-grain tropical japonica and hybrid rice varieties. The use of hybrids and herbicide resistant varieties is increasing in the Southern US since roughly 2001, while CA still uses primarily conventional inbred varieties. US rice production systems are direct seeded; CA rice is primarily water-seeded (i.e. – pre-soaked seed is applied to flooded fields via airplane), whereas Southern US rice is primarily drill-seeded. Furthermore, in the southern US, rice systems are often rotated with other crops (i.e. soybean and maize) while in CA, due to poorly drained heavy clay soils not suitable for other crops, most rice systems are continuous rice.

Data Sources and Assumptions (following GYGA protocols: http://www.yieldgap.org/web/guest/methods-overview)

Harvested area and actual yields

SPAM2005 harvested area maps were used for selecting the reference weather stations for each crop and water regime case. County-level data on crop harvested area and average yields for each crop were retrieved from the USDA-NASS (http://www.nass.usda.gov/Quick_Stats/). Statistics from the most recent 10 cropping seasons (2005-2014) were used to calculate average yields for maize and from the 5 most recent cropping seasons (2010-2014) for rice.

Weather data and reference weather stations

For rainfed and irrigated maize, historical (last 20+ years) measured daily weather data, including solar radiation, maximum and minimum temperature, precipitation, relative humidity, wind speed and precipitation were obtained from the High Plains Regional Climate Center (HPRCC), the Illinois Water and Atmospheric Resources Monitoring Program (WARM), Ohio Agricultural Research and Development Center (OARDC) Weather Service, the University of Wisconsin Extension Ag Weather, the Indiana Purdue Automated Agricultural Weather Station Network (PAAWS), North Dakota Agricultural Weather Network (NDAWN), Michigan State University Enviro-Weather, Oklahoma Mesonet, State Climate Office of North Carolina (SCO), West Texas Mesonet, the Southern Research and Outreach Center (SROC) and the Southwest Research and Outreach Center (SWROC) form the University of Minnesota, the South Dakota Climate and Weather, and the Missouri Mesonet (AgEBB). Meteorological stations selected from these networks are located in agricultural areas, rather than in urban areas, which helps to ensure the representativeness of the weather data for estimating yield potential. Due to lack of high quality measured weather data, we used gridded daily weather data from the NASA Prediction of Worldwide Energy Resource (POWER; http://power.larc.nasa.gov/) for one location in TX (Draughon Miller).

For irrigated rice, historical (last 13 to 15 years) measured daily weather data included maximum and minimum temperatures, precipitation, relative humidity, and wind speed. Data was retrieved from the California Irrigation Management Information System (CIMIS), the National Weather Service (NWS) station network, the Louisiana Agriclimatic Information System, the Delta Agricultural Weather Center, and the Texas A&M Integrated Agricultural Information and Management System (iAIMS) Climatic Database. Similar to maize, these stations are located in primarily agricultural rather than urban areas.

Quality control and filling/correction of the weather data was performed based on GYGA weather protocols and comparison against weather records from adjacent National Weather Service-Cooperative Observer Network (NWS-COOP) meteorological stations. NASA-POWER was used as source of incident solar radiation data to fill missing solar radiation data. Hence, complete long-term weather records were available for simulating yield potential (Yp) for irrigated crops and water-limited yield potential (Yw) for rainfed crops.

Based on crop harvested area distribution and the agro-climatic zones defined for USA (Van Wart et al., 2013a), a total of 21 (irrigated maize) and 45 (rainfed maize) reference weather stations (RWS) were selected following van Bussel et al (2015). Additional RWS were selected to account for regions where local maize production is important or where current maize area has expanded substantially during recent years. A total of 14 RWS were selected for rice. RWS buffer zones accounted for 59 and 62% of total harvested area for irrigated and rainfed maize and 87% for irrigated rice, while the agro-climatic zones where these locations were located accounted for 77, 87 and 92% of national area for these three crop-water regimes combinations.

Soil data

For rainfed crops, the dominant soil series were identified for each RWS buffer based on gSSURGO soil database (NRCS, Soil Survey Staff). Briefly, maize harvested area within each soil type within each RWS buffer was calculated and we included as many soil types as needed to cover =50% maize area within each buffer. Rooting depth was set at 1.5 m for all soil types due to lack of physical and chemical constrains to root growth and based on observed soil water extraction patterns for maize (Payero et al., 2006)..

Crop system and management information for crop simulations

Maize simulations

Management practices were obtained for each RWS buffer zone. Average sowing date for the last 10-y period (2005-2014) was retrieved from USDA-RMA while data on planting data and cultivar maturity were provided by seed companies and local agronomists. Requested information include: dominant crop rotations and proportion of each of them to the total harvested area, average planting dates, dominant cultivar name and maturity, and actual and optimal plant population density. The provided data were subsequently corroborated by other local and national experts. Further details on management input data can be found in Morell et al (2016).

Simulations using maturity of the most widely used cultivar or hybrid for the region surrounding each RWS were performed using Hybrid-Maize model (Yang et al., 2004). The model was previously validated for its ability to simulate maize yield potential across a wide range of environments in the US Corn Belt (Grassini et al., 2009; Liu et al., 2005) and found to be robust at portraying key G x E x M interactions across a wide range of yields, from 0.5 to 18 Mg ha-1. The model does not require site-specific parameter calibration for simulating yield potential, that is, all model parameters governing photosynthesis, respiration, leaf area expansion, light interception, biomass partitioning, and grain filling rate and duration are generic (Yang et al., 2004).

To account for differences in initial soil water at planting time among years, simulations of rainfed crops were initiated at immediately after harvest of the prior crop assuming 50% of plant available soil water based on Grassini et al. (2009). Soil surface residue coverage at harvest of the prior crop was set at 50% to reflect the large amount of non-grain aboveground biomass left by previous crop and the large proportion of maize production area under no-till or conservation tillage practices (Horowitz et al., 2010). Surface runoff was assumed negligible to reflect producer preference for avoiding highly sloping land for maize production, and wide adoption of conservation tillage and no-till practices, which greatly reduce runoff. Because dominant crop rotations are 2-y soybean-maize and continuous maize, maize yield potential was simulated for all the years for which weather data were available. Because there is little (if any) difference between actual and optimal plant population, the simulations were based on the range of optimal plant population reported for each buffer zone by the local agronomists. Simulations assumed no limitations to crop growth by nutrients.

A factor that is not accounted for by the crop simulation model, with a likely positive impact on rainfed yields (especially in drought years), is the water supply from perched water tables in the central and eastern regions of the Corn Belt. Five independent sources of information were used to identify RWS buffers with likely influence of perched water tables on water-limited yields: (i) data on depth to water table from gSSURGO (NRCS Soil Survey Staff), (ii) maps on subsurface tile-drained area distribution in USA (Sugg, 2007), (iii) consultation with local agronomists, (iv) data from shallow groundwater wells networks (e.g., http://isws.illinois.edu/warm/groundwater/) and (iv) trends in actual yield at these locations (i.e., high yields with little year-to-year variation despite contrasting differences in rainfall). Following this approach, we identified 19 RWS buffer zones in the central and eastern regions of the US Corn Belt where sub-surface irrigation is very likely to have a positive impact on rainfed fields. We assumed that water was non-limiting for those fields located in areas with presence of water table within the rootable soil depth during the growing season. At these 19 RWS, separate simulations were performed for water-limited (no influence of water table) and non-water limited environments (soils with water table within rootable soil depth during the maize growing season) and results were weighted based on their relative share of maize harvested area. In the case of irrigated crops in the southern Great Plains (OK, TX and southeastern KS), maximum temperature during the peak of the growing season was reduced by 0.5ºC to account for the ‘cooling' effect associated with frequent irrigations in this region. Finally, simulated yield potential and water-limited yield potential for northern locations at ND, SD, MN and WI were relatively low and the resulting yield gap small (ca. 10% of simulated yield potential).  We suspect that potential yields are higher than the current Hybrid-Maize model predicts, possible because the current trigger for a killing frost is too conservative.

Simulations utilized weather, soil, and management data obtained for each RWS. For each RWS, all selected soil types were simulated and then weighted by their relative proportion to retrieve an average Yw at the level of the RWS buffer zone. Quality control of results was performed following the guidelines provided by Grassini et al. (2015). A weighted average actual yield was calculated based on the actual yield reported, separately for rainfed and irrigated crops, for the counties that coincide with the buffer zone and the relative contribution of each county to the total crop harvested area in the buffer zone. In those states in central and eastern Corn Belt where actual yields are not disaggregated by water regime but irrigation may still be important in some specific areas (e.g. central WI, central-east IL, southern MI, and northern IN), counties with large irrigated area were not included for the calculation of rainfed actual yield based on irrigated harvested area maps (http://www.agcensus.usda.gov/Publications/2012/Online_Resources/Ag_Atlas_Maps/Crops_and_Plants/). There were, at least, 5 counties within each RWS buffer (average of 10 counties per RWS). Reported Yw and Yp in the Atlas are long-term (20+ years) averages.

Yield gap (Yg) was calculated as the difference between long-term average Yw or Yp and average (2005-2014) actual farmers' yields. Including more years before 2005 in the calculation of average actual yield would bias the estimate of average actual yield due to a strong, positive technology trend in USA. In a few years (2005 and 2012), a severe drought affected summer crops and an important fraction of the planted maize area was not harvested in the western part of the Corn Belt. Hence, the actual maize grain yield reported for those years represents the yield from the most favorable environments for maize production within each RWS buffer. To correct this bias, we adjusted the actual maize yields based on the proportion of planted area that was not harvested within each RWS buffer.

Rice simulations

Yield potential was estimated by simulation using the ORYZA (v3) crop model (Bouman et al., 2001). This model was chosen due to its widespread use and existing body of work validating it for various rice cropping systems. Further calibration and validation of this model was required to adequately simulate US rice yield potential for representative high-yielding varieties typical of the types most widely planted in major U.S. rice-producing regions as described in Espe et al. (2016). In order to minimize the effect of variation between simulations, yield potential was simulated for each buffer zone over a  15-year time span at all locations except those in LA, which only had 13 years of weather data. Long-term mean yield potential in RWS buffer zones and as aggregated for larger spatial scales followed GYGA procedures.

For each buffer zone, the average emergence date as reported by state agronomists was used to initiate simulations. Sensitivity analyses were performed to evaluate the impact of using the long-term average emergence date rather than a distribution of emergence dates centered at the average, with only slight yield differences observed (absolute difference of less than 0.25 Mg ha-1; results not presented). Therefore, the average emergence date was deemed suitable for simulating yield potential. Annual simulated yield potentials were then averaged by buffer zone to determine the buffer zone yield potential. Individual simulation results were quality controlled for unrealistic results or failed simulations prior to averaging.

Data from the USDA-NASS database (USDA - National Agricultural Statistics Service, 2016) were used to determine actual yields within each buffer zone. Since buffer zones were constructed without regard for state or county boundaries, county-level data were retrieved and aggregated to obtain estimates for each buffer zone. In order to do this, the buffer zone average yield was calculated as a weighted average of county estimates, where weights were determined by the proportion of buffer zone harvested area within each county.

Where Yk is the average yield for buffer zone k, µj is the average reported yield for county j, n is the number of counties with harvested acres in buffer zone k, aj is the harvested area of county j in buffer zone k, and ak is the total harvested area in buffer zone k. To minimize potential confounding effects of yield trends over time, only the most recent reported data (2010 to 2014) was used. Buffer zone estimates were calculated by year and then averaged across years to get the average buffer zone yield.

† We thank Dennis Todey (South Dakota State University), Ken Scheeringa (Purdue University), Rick Wayne (University of Wisconsin-Madison), Mark Seeley (University of Minnesota), Jenny Atkins (WARM Program, University of Illinois at Urbana-Champaign), Daryl Herzmann (Iowa State University), and Bill Sorensen and Natalie Umphlett (UNL) for their help to access historical daily weather data.


Bouman, B., Kropff, M., Tuong, T., Wopereis, M., ten Berge, H., van Laar, H., 2001. Oryza2000: modeling lowland rice, ed. International Rice Research Institute.

Connor, D.J., Loomis, R.S., Cassman, K.G., 2011. Crop Ecology. Productivity and Management in Agricultural Systems. Cambridge Univ. Press, Cambridge, UK.

Espe, M.B., Yang, H., Cassman, K.G., Guilpart, N., Sharifi, H., Linquist, B.A., 2016. Estimating yield potential in temperate high-yielding, direct-seeded US rice production systems. F. Crop. Res. In press. doi:10.1016/j.fcr.2016.04.003

Food and Agriculture Organization of the United Nations, FAO Statistical Databases. 747 Available at http://faostat.fao.org/

Grassini, P., Specht, J.E., Tollenaar, M., Ciampitti, I., Cassman, K.G., 2014. High-yield maize–soybean cropping systems in the US Corn Belt. In: Crop Physiology (2nd Edition), Sadras, V.O., Calderini, D.F. (Eds). Academic Press, San Diego.

Grassini, P., van Bussel, L.G.J, Van Wart, J., Wolf, J., Claessens, L., Yang, H., Boogaard, H., de Groot, H., van Ittersum, M.K., Cassman, K.G., 2015. How good is good enough? Data requirements for reliable crop yield simulations and yield-gap analysis. Field Crops Res. 177, 49-63.

Grassini, P., Yang, H., Cassman, K.G., 2009. Limits to maize productivity in Western Corn-Belt: A simulation analysis for fully irrigated and rainfed conditions. Agric. Forest 761 Meteoro. 149, 1254-1265.

Horowitz, J., Ebel, R., Ueda, K., 2010. No-till farming is a growing practice. USDA-ERS Economic Information Bulletin No. 70.

Fischer, R.A., 2015. Definitions and determination of crop yield, yield gaps, and of rates of change. F. Crop. Res. 182, 9–18. doi:10.1016/j.fcr.2014.12.006

Liu, X., Andresen, J., Yang, H., and Niyogi, D., 2015. Calibration and Validation of the Hybrid-Maize crop model for regional analysis and application over the U.S. Corn Belt. 802 Earth Interactions. 19, 1-16.

Livezey, J. and L. Foreman. 2004. Characteristics and production costs of U.S. rice farms. US Dept. of Ag. Statistical Bulletin. 974-7.

Lobell, D.B., Cassman, K.G., Field, C.B., 2009. Crop Yield Gaps: Their Importance, Magnitudes, and Causes. Annu. Rev. Environ. Resour. 34, 179–204. doi:10.1146/annurev.environ.041008.093740

Morell FJ, Yang HS, Cassman KG, Van Wart J, Elmore RW, Licht M, Coulter JA, Ciampitti IA, Pittelkow CM, Brouder SM, Thomison P, Lauer J, Graham C, Massey R, Grassini P (2016) Can crop simulation models be used to predict local to regional maize yields and total production in the U.S. Corn Belt? Field Crops Research (In Press) doi 10.1016/j.fcr.2016.04.004

Payero, J.O., Klocke, N.L., Schneekloth, J.P., Davison, D.R., 2006. Comparison of irrigation strategies for surface-irrigated corn in West Central Nebraska. Irrig. Sci. 24, 817 257-265.

Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Web Soil Survey. Available online at http://websoilsurvey.nrcs.usda.gov/

Sugg, Z., 2007. Assessing U.S. farm drainage: can GIS lead to better estimates of subsurface drainage extent? World Resources Institute, Washington D.C [online WWW]. Available URL:http://pdf.wri.org/assessing_farm_drainage.pdf

USDA–National Agricultural Statistics Service (NASS), 2000-2014. Quick stats 2.0. 843 Available at: http://quickstats.nass.usda.gov/

van Bussel, L.G.J., Grassini, P., Van Wart, J., Wolf, J., Claessens, L., Yang, H., Boogaard, H., de Groot, H., Saito, K., Cassman, K.G., Van Ittersum, M.K., 2015. From field to atlas: Upscaling of location-specific yield gap estimates. Field Crops Res, 117, 853 98-108.

Water and Atmospheric Resources Monitoring Program. Shallow Groundwater Network. (2015). Illinois State Water Survey, 2204 Griffith Drive, Champaign, IL 61820-7495. http://dx.doi.org/10.13012/J8CC0XMK

Van Wart, J., Van Bussel, L.G.J., Wolf, J., Licker, R., Grassini, P., Nelson, A., Boogaard, H., Gerber, J., Mueller, N.D., Claessens, L., Cassman, K.G., Van Ittersum, M.K., 2013a. Reviewing the use of agro-climatic zones to upscale simulated crop yield potential. Field Crops Res. 143, 44-55.

Yang, H.S., Dobermann, A., Lindquist, J.L., Walters, D.T., Arkebauer, T.J., Cassman, K.G., 2004. Hybrid-maize - A maize simulation model that combines two crop modeling 882 approaches. Field Crops Res. 87, 131-154.

Go to the Atlas

This year the mapviewer will be updated.
For now please enable Flash in your browser.

Get access to the Atlas for advanced users


Download GYGA results

All downloads are available under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Data from the Global Yield Gap Atlas is available for commercial use. We have different commercial license types available, either for a single use of the data within a project context or multiple use and commercialization, suiting your needs. Please email Patricio Grassini (UNL) or Martin van Ittersum (WUR) to discuss options.

Country agronomists Country agronomists

Maize team:

Chris Graham

South Dakota State University

Cameron Pittelkow

University of Illinois

Peter Thomison

The Ohio State university

Sylvie Brouder

Purdue University

Raymond Massey

Missouri University

Jeff Coulter

University of Minnesota

Joe Lauer

University of Wisconsin

Mark Licht

Iowa State University

Ignacio Ciampitti

Kansas State University


Rice team:

Bruce A. Linquist

Dept. of Plant Sciences, University of California - Davis, Davis, CA, USA

Matthew B. Espe

Dept. of Plant Sciences, University of California - Davis, Davis, CA, USA

Dustin Harrell

LSU AgCenter, College of Agriculture, Louisiana State University, Baton Rouge, LA, USA

Steve Linscombe

LSU AgCenter, College of Agriculture, Louisiana State University, Baton Rouge, LA, USA

Kent McKenzie

Rice Experiment Station, California Cooperative Rice Research Foundation, Biggs, CA, USA





Merle Anders

Net-Profit Crop Consultancy PLLC, AR, USA


Donn Beighley

Dept. of Agriculture, Southeast Missouri State University, Malden, MO, USA


Randall Mutters

University of California Cooperative Extension, Oroville, CA, USA


Ted Wilson

Texas A&M AgriLife Research Center, Beaumont, TX, USA