Toshichika Iizumi

Global gridded crop models (GGCMs) are important tools for assessing climate impacts on agriculture, yet significant divergence in their projections limits interpretability, and impact studies often treat GGCMs as black boxes. Targeted ensemble sensitivity analyses are demanding and not transferable to different ensembles. Here, we comprehensively evaluate climatic and soil drivers of crop yield anomalies in a state‐of‐the‐art GGCM ensemble, using maize as a representative crop. Gradient boosting classifiers detect anomalies, SHapley Additive exPlanations (SHAP) values quantify feature importance, and methods are applied to a recent GGCM experiment driven by reanalysis climate data. We find broadly similar climatic drivers across the ensemble, though feature importance distributions differ. Low precipitation dominates under rainfed conditions, while solar radiation typically ranks second, highlighting that drought impacts depend on atmospheric water demand often omitted from sensitivity analyses. In some GGCMs, excess rather than insufficient water drives anomalies. The comparison with climate sensitivities derived from a global yield reference data set indicates that no single GGCM performs best across all climate regions. Individual GGCMs may respond well to one climatic stressor but not others, underscoring that bulk benchmarking of GGCMs against observations can obfuscate processes requiring targeted improvement. Our findings demonstrate that evaluating opportunistic data, that is, experiments produced for other purposes, yields vital insights into GGCM divergence in impact studies. Code is publicly available on GitHub to support future attribution analyses and inform broad audiences about drivers of observed results.

Stable crop production is the basis for global food security. Climate change has increased extreme climate events, often leading to low yields (bad harvests). However, favorable climatic conditions and elevated carbon dioxide concentrations may also enhance the yields of good years (good harvests) in some locations and seasons. Here, we detect the historical impacts of climate change on good and bad harvests of maize and wheat in 63 and 53 countries, respectively. Good and bad harvests were defined as the three highest and lowest yield years during 2000–2019. Using a random forest-based yield anomaly dataset, we analyzed yield anomalies under historical and non-warming (NW) climate conditions. The NW simulations represent yields without anthropogenic climate change. The impacts of human-induced climate change were quantified as the difference between the two. Negative yield impacts were more widespread than positive impacts. Climate change has worsened bad harvests in 76% of the maize-producing countries and 55% of the wheat-producing countries, both of which are mainly located in the mid-latitudes. In contrast, enhancements in good harvests occurred in 32% and 21% of the countries primarily located at high-latitudes. As a result, the amplitude of interannual yield variability increased in 13% and 6% of maize- and wheat-producing countries, respectively, owing to both enhanced good harvests and worsened bad harvests. During the study period, growing-season mean temperatures in most producing countries remained below the crop-specific optimum levels; therefore, warming improved the thermal conditions for crop growth. Human-induced changes in precipitation play a more important role in bad harvests than temperature, as growing-season precipitation is often below the crop water needs. The findings of this study suggest that the adverse effects of climate change on maize and wheat production outweigh the favorable effects, underscoring the need for climate change mitigation to sustain global food security.

Global gridded crop models (GGCMs) are important tools for assessing climate impacts on agriculture, yet significant divergence in their projections limits interpretability, and impact studies often treat GGCMs as black boxes. Targeted ensemble sensitivity analyses are demanding and not transferable to different ensembles. Here, we comprehensively evaluate climatic and soil drivers of crop yield anomalies in a state-of-the-art GGCM ensemble, using maize as a representative crop. Gradient boosting classifiers detect anomalies, SHapley Additive exPlanations (SHAP) values quantify feature importance, and methods are applied to a recent GGCM experiment driven by reanalysis climate data. We find broadly similar climatic drivers across the ensemble, though feature importance distributions differ. Low precipitation dominates under rainfed conditions, while solar radiation typically ranks second, highlighting that drought impacts depend on atmospheric water demand often omitted from sensitivity analyses. In some GGCMs, excess rather than insufficient water drives anomalies. The comparison with climate sensitivities derived from a global yield reference dataset indicates that no single GGCM performs best across all climate regions. Individual GGCMs may respond well to one climatic stressor but not others, underscoring that bulk benchmarking of GGCMs against observations can obfuscate processes requiring targeted improvement. Our findings demonstrate that evaluating opportunistic data, i.e., experiments produced for other purposes, yields vital insights into GGCM divergence in impact studies. Code is publicly available on GitHub to support future attribution analyses and inform broad audiences about drivers of observed results.

Irrigation has enabled higher and more stable food production than when farmers relied only on rainfed cropping systems in the past. However, its future benefits may diminish under climate change due to increasing crop water demand and constraints on water availability. Here, we compare crop yields between irrigated and rainfed conditions on currently irrigated cropland. We use an integrated crop and water model under multiple climate scenarios up to 2100 and quantify their differences as an indicator of the irrigation effect. We find that the irrigation effect on yields remains close to the current level until the middle of this century. This corresponds to a global decadal mean temperature increase of up to +2°C relative to preindustrial levels. Under a high emission scenario, with approximately +5°C warming by the end of this century, the irrigation effect is projected to decrease to approximately 64% of the current level. Absolute crop production is projected to decrease by approximately 19%, while irrigation‐induced production gains decline to approximately 53% of the current levels under the present distribution of irrigated cropland. These results suggest that the effectiveness of irrigation as a buffer against crop water stress diminishes with warming. This highlights the importance of understanding how climate‐driven changes in crop water demand and supply constrain irrigation effectiveness.

Virtual water trade (VWT) redistributes water embodied in agricultural commodities across borders and thereby shapes global interdependence between water resources and food security. Recent studies have increasingly used integrated assessment models (IAMs)—including GCAM, a partial-equilibrium IAM—to project future agricultural production and trade balances under future climate and socio-economic change and to infer virtual water transfer flows(e.g., Graham et al., 2020). However, such approaches assume that commodities are traded in a single global markets, making it difficult to explicitly quantify bilateral exporter–importer dependency structures.In this study, we develop a scenario-based framework to estimate bilateral virtual water trade of rice and wheat toward 2100 by combining projections of harvested area (land-use), climate-driven yield changes, and population dynamics with an extrapolation of current trade structures. Using baseline bilateral trade matrices from FAOSTAT, we assume that (i) exporter-specific allocation to destination countries and (ii) national export-to-production ratios remain fixed, and we scale bilateral trade volumes in accordance with scenario-driven changes in production and demand. We then compute bilateral VWT by linking projected crop flows with crop- and location-specific water-use coefficients. The analysis focuses on SSP2 as the primary scenario, with additional SSP comparison(SSP126 and SSP585). This framework enables assessment of how future VWT magnitude and bilateral dependency patterns may evolve differently between rice—characterized by relatively thin international markets—and wheat, which is traded in thicker global markets, providing insights for water–food security assessment under future climate and socio-economic change.

The El Niño-Southern Oscillation (ENSO) will remain the major mode of natural climate variability in a warmer climate. However, how its impacts on crop yields may change under future climate change remains uncertain. Here, we present projected ENSO impacts on yields of maize, wheat, rice and soybean in the middle (2035-2064) and end (2065-2094) of the 21st century under low (Shared Socioeconomic Pathway [SSP] 1-2.6) and high (SSP5-8.5) warming scenarios. The recent climate-crop model ensemble we used (consisting of 12 global gridded crop models [GGCMs], provided by the Agricultural Model Intercomparison and Improvement Project [AgMIP]’s Global Gridded Crop Model Intercomparison [GGCMI] and the Intersectoral Impact Model Intercomparison [ISIMIP] project phase 3) successfully captured the observed major characteristics in the geographical spatial patterns of ENSO-induced yield changes, especially for wheat in La Niña years and for maize and soybean in both El Niño and La Niña years. However, its ability was low for rice. The relatively localized distribution of rice cultivation may limit the model ensemble’s ability to capture ENSO-driven yield variability. Overall, the results indicate that ENSO will continue to be a major driver of global yield variability. The negative impacts of El Niño on maize in eastern Brazil and the positive impacts in Argentina are projected to persist across all warming levels, whereas the negative impact on soybean yields in India may vary depending on the warming level. Yet, substantial uncertainties remain in the projected ENSO impacts, emphasizing the need to improve both climate and crop models to better prepare cropping systems for ENSO-related risks in a warmer climate.

Global gridded crop models (GGCMs) are crucial for projecting the impacts of climate change, yet their performance under climatic extremes remains poorly understood. Using historical subnational yield data, we evaluate 13 GGCMs from the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) across major crops. Models largely capture the signal of heat and drought extremes but consistently underestimate the magnitude of losses. For extreme wet conditions, they often miss both the signal and magnitude (losses underestimated in 83–95\% of cases, best to worst model). However, underestimation is reduced for compound heat–drought extremes (38–77\%). Notably, the model ensemble median, widely used to reduce uncertainty, performs poorly under extremes, systematically underestimating risks. We identify model characteristics that influence underestimation under extremes, such as the representation of radiation or soil nutrient dynamics. Improving key process representations and input data quality is essential to strengthen climate risk assessments for global agriculture and food security.

Maize cultivation faces increasing risks from climate change, particularly because of increasing temperatures and extreme weather events. This study investigates whether the ‘hot’ Earth system models (ESMs) of the Coupled Model Intercomparison Project Phase 6, whose past warming trends are typically greater than the observations, tend to overestimate future temperature and precipitation changes across the global maize harvesting areas. By applying an emergent constraint (EC) approach to 30 ESMs, we assess the future mean and extreme temperature (Δ T _ave and Δ T _max ) and precipitation (Δ P _ave and Δ P _max ) changes during maize growing seasons. We find that Δ T _ave , Δ T _max , and Δ P _max averaged over the global harvesting regions are significantly correlated with the past global mean temperature trends, indicating that hot ESMs tend to overestimate the future changes in these variables. ECs reduce the inter-ESM variances in these projections by 43%, 39%, and 18%, respectively. Notably, the regions with the highest maize production, such as the USA and China, are projected to experience the greatest increases in Δ T _ave and Δ T _max . The fraction of the global maize production exposed to historically rare high temperatures increases substantially in the raw projections but is moderated when ECs are applied. These findings suggest that the use of hot ESMs may lead to overestimated impacts of climate change on maize and that EC methods offer a robust pathway for refining impact assessments.

Extreme summer heatwaves in 2023 and 2024 severely disrupted the vegetable supply chain to the Tokyo metropolitan area. This study analyzes the impacts of these extreme climate events on urban food supply and market stability, using governmental statistics on wholesale arrival volumes and prices for fifteen major vegetables. Compared with the 2010–2022 baseline, the arrival volumes of carrot, Japanese radish, and tomato fell by 19–27%, while prices rose by 46–82%. The findings highlight how consecutive heatwaves amplified risks to food security and underscore the urgent need for climate-resilient agri-food systems. The study contributes to multidisciplinary research integrating agricultural meteorology, climate-impact assessment, and sustainable food-system management.

Spatio-temporal patterns of extreme climate events have been extensively studied, yet two questions remain underexplored: Do such events occur regularly, and how do regularity patterns change under global warming? We address these questions by investigating dominant periods in crop failure, heatwave, and wildfire data. Here, we show that under pre-industrial conditions dominant periods emerge in 28% of cropland exposed to crop failure and 10% of wildfire-affected areas, likely related to climatic oscillations such as the El Niño-Southern Oscillation, while heatwaves occur irregularly. The number of dominant periods increases by 2–13% during the transition from the pre-industrial era to the anthropocene. In the anthropocene, the occurrence of extreme events shifts towards monotonic growth, replacing previous natural regularity patterns. Linearly de-trended projections reveal an additional shift towards smaller dominant periods due to climate change. These shifts in regularity are crucial for adaptation planning, and our method offers an additional approach for studying extreme events.