Comparing 214 field-warming observations with seven crop models, Peking University researchers estimate that 1°C of global warming could cut rice yields by about 8 percent, roughly double earlier projections.
Rice feeds more than half of the world’s population. In much of Asia, it is the foundation of the daily diet, and even modest changes in harvests can ripple through food prices and food security. How climate change will affect rice yields is therefore one of the most important questions in agricultural science.
A new study suggests that many projections have been too optimistic. Researchers at Peking University and international collaborators report in Science Advances that widely used crop models underestimate rice’s vulnerability to extreme heat, and that once that vulnerability is accounted for, projected yield losses roughly double.
Averages miss the extremes
Earlier assessments of climate impacts on rice have focused largely on rising average temperatures. But according to the release, they have tended to overlook extreme heat events, especially during the most sensitive stages of the crop’s growth.
Those stages are flowering and grain filling. When temperatures spike during flowering, pollen can fail and flowers may not set grain. During grain filling, heat disrupts the development of the kernels. The study found that extreme heat during these periods lowers grain-setting rates and disrupts grain development, causing significant yield declines.
Agronomists have long known that rice is particularly sensitive at flowering. Temperatures above roughly 35 degrees Celsius at the time the flowers open can sharply increase the share of sterile spikelets, the small units that would otherwise become grains.
Rice flowers are open only briefly, usually for a few hours in the late morning, which makes the timing of a hot spell especially important. A heatwave that strikes during that window can do far more damage than the same temperatures a few weeks earlier or later.
Testing models against the field
The research was led by Professors Zhou Feng and Wang Xuhui of Peking University’s College of Urban and Environmental Sciences. The team analyzed 214 observations from field-warming experiments around the world, in which researchers artificially raise temperatures over growing crops, and compared the results with the output of seven crop models.
Field-warming experiments typically use devices such as infrared heaters or open-top chambers to raise the temperature around plants while leaving them in otherwise realistic conditions. They offer a direct test of how crops respond to heat, against which computer models can be checked.
The researchers found that temperatures above 30 degrees Celsius were a key control on yield losses. At more than 70 percent of sites, heat exposure explained more than half of the observed changes in yield.
Twice the projected loss
When the team corrected the models using the field observations, the projected damage grew substantially. Under a scenario of 1 degree Celsius of global warming, the original models projected a global rice yield decline of 3.8 percent, plus or minus 1.7 percent. After correction, the estimate rose to 8.1 percent, plus or minus 2.3 percent.
The largest revisions were in South and Southeast Asia, where much of the world’s rice is grown and eaten. For Pakistan, India and Bangladesh, projected losses rose from 2.1 percent to 6.6 percent.
The authors conclude that extreme heat, and not just changes in average temperature, is critical to understanding future risks to agriculture.
That distinction matters because extreme heat is expected to increase faster than average temperatures in many regions. A modest rise in the mean can bring a much larger increase in the number of very hot days, and the study points to heat of that kind as a major driver of losses in rice.
The stakes extend well beyond individual farms. India is the world’s largest rice exporter, and when it restricted exports of non-basmati white rice in 2023, citing concerns about domestic supplies and prices, global rice prices rose sharply. Lower yields in the region would affect markets and food security far beyond its borders.
Better models, better adaptation
The study offers a way to improve crop models by incorporating field-based evidence of heat vulnerability. More realistic models could help governments, researchers and farmers plan for climate change more effectively.
Crop models are computer programs that simulate how plants grow in response to weather, soil and management. They are widely used to project the effects of climate change and to inform policy, but they differ in how they represent processes such as heat stress, which is one reason their projections have varied so widely.
It also points to adaptation measures that could reduce the damage. These include developing and planting heat-tolerant rice varieties, adjusting planting schedules so that flowering avoids the hottest periods, improving irrigation and managing soils more effectively. The findings, the authors suggest, are relevant to global food security planning.
The analysis has limits. It relies on a set of field experiments that may not represent every growing region, and it applies corrections to a particular group of seven models. Future warming may also bring other changes, such as shifts in rainfall and higher carbon dioxide levels, that affect yields in different ways.
The release does not include comments from the researchers, the full author list or details of funding. But the core message is clear: if crop models miss the effects of extreme heat, they may be underestimating how much climate change will cost one of the world’s most important crops.
Notes
- “Vulnerability to high temperature shapes global warming impacts on rice yield.” Science Advances, 2026. DOI: 10.1126/sciadv.aed9226.
- Peking University. “Global crop models underestimate rice vulnerability to extreme heat.” EurekAlert!, 29 July 2026. https://www.eurekalert.org/news-releases/1139775.
Photo by Pixabay on Pexels.com