New Model Shows Climate Risk to Wheat Yields Is Worse Than Thought
15 October 2026, Beijing: A new global crop modelling study has found that widely used climate models significantly underestimate how much combined heatwaves and drought already cut wheat yields, with implications for how governments, insurers and breeders plan for future harvests. The study, led by researchers at Southwest University and Beijing Normal University in China, working with the Chinese Academy of Sciences and NASA’s Goddard Institute for Space Studies, was published in the journal Nature Food on October 8.
The team built an improved crop model and tested it against 1981-2015 weather and yield data covering global wheat production, then compared its performance with a standard multimodel ensemble widely used in climate-impact research. The improved model raised predictive skill by roughly 30% for drought-driven yield losses, 20% for heatwave-driven losses, and 7% for compound events, where both stresses hit in the same season. Using the new model, the researchers calculated that droughts have cut average global wheat yields by 9.7%, heatwaves by 5.8%, and compound heatwave-drought events by 13.3%, a toll higher than what ensemble models, which capture the interaction between the two stresses less precisely, had estimated.
Irrigation’s protective effect is weakening
The study also examined irrigation as an adaptation tool, and found it remains effective at offsetting yield losses from drought and heat, but that its protective effect has “markedly weakened” over the study period as climate disasters have intensified. In practice, that means irrigated wheat is still better protected than rain-fed wheat, but the gap is narrowing as more frequent and more severe heat and drought events outpace what current irrigation practices can buffer against.
The researchers built their model using a dense set of inputs, including high-resolution climate reanalysis data, global and subnational wheat yield statistics, agro-meteorological station records, satellite-derived leaf area index data, and global datasets of irrigated versus rainfed cropland, benchmarked against a widely used multimodel climate-impact ensemble known as GGCMI. Running the model at 10-kilometre resolution is what let the team isolate the compound-event signal that coarser multimodel approaches tend to blur together or understate, the authors argue.
Part of a wider reassessment of wheat’s climate risk
The Nature Food paper adds to a growing body of research this year on how climate change is reshaping the balance of risks to global wheat, much of it converging on a similar warning: that heat and drought risk, taken separately or together, is rising faster than earlier climate-impact assessments assumed, and that current adaptation tools, from irrigation to earlier sowing to breeding for heat tolerance at flowering, are not keeping pace evenly across regions. What this new paper adds specifically is a demonstrated jump in model accuracy for compound events, the stress combination hardest to capture and, this study suggests, the one most understated until now.
Wheat is one of the most widely traded staple grains, grown across climates from Punjab and the Pampas to the North China Plain and the US Great Plains, so a systematic underestimate of compound climate risk has implications well beyond any single country’s farmers. Insurers, commodity traders and national food-security planners all lean on climate-impact models of the kind this study critiques, so an improvement of this scale in capturing compound-event risk could feed into more cautious assumptions about future supply volatility, independent of any change in farming practice on the ground.
Also Read: Bayer to Distribute INERA’s Biological Crop Nutrition Products in India
Global Agriculture is an independent international media platform covering agri-business, policy, technology, and sustainability. For editorial collaborations, thought leadership, and strategic communications, write to pr@global-agriculture.com
