Extreme weather events like heatwaves, droughts, and extreme precipitation negatively impact crop production and food security. Global warming is increasing the frequency and intensity of these extremes and the likelihood of their simultaneous occurrence globally. Synchronized occurrences lead to amplified societal impacts, often exceeding the sum of individual events. Synchronized crop failures across multiple breadbasket regions pose a significant risk to global food security and food supply chains, particularly impacting import-dependent regions. While previous research assessed synchronized breadbasket failures statistically or in relation to climate variability on annual and seasonal timescales, the role of the jet stream in driving concurrent mid-latitude weather extremes has not been fully quantified. The jet stream's summertime circulation regimes act as circumglobal teleconnections, promoting simultaneous heat and rainfall extremes with adverse effects on agriculture. High-amplitude wave patterns (wave-5 and wave-7), characterized by the number of ridges and troughs in the mid-latitudes, have been observed during major Northern Hemisphere summer weather extremes. While the importance of atmospheric wave patterns for local or coinciding extreme weather events has been established, their impact on yield co-variability between regions remains unquantified in observations and model experiments. Accurate risk assessments require crop models capable of reproducing observed extreme weather-yield responses and climate models that accurately reproduce the relationship between wave patterns and extreme weather. This study addresses this gap by examining the observed concurrence patterns of low yields in major breadbasket regions associated with wave-5 and wave-7 patterns and evaluating the ability of models to reproduce these relationships.

The literature highlights the detrimental effects of extreme weather events on global crop production and food security. Studies have shown the increasing frequency and intensity of such events due to climate change, and the potential for simultaneous occurrences leading to cascading impacts. Previous research has assessed the risks of synchronized breadbasket failures using statistical methods or by considering dominant modes of climate variability. However, a comprehensive understanding of the role of the jet stream in driving concurrent weather extremes and their effects on yield co-variability across regions has been lacking. Existing studies have shown the importance of atmospheric wave patterns for local or coinciding extreme weather, but their impact on yield anomalies has primarily been quantified regionally, not across multiple breadbasket regions. The limitations of current climate and crop models in accurately simulating such high-impact events necessitate further investigation. This study builds upon prior research by focusing on the interaction between jet stream patterns, surface weather anomalies, and the resultant impacts on crop yields across multiple major crop-producing regions.

This study uses data from the Global Gridded Crop Model Intercomparison (GGCMI) project, focusing on five key crop regions (accounting for a large portion of global maize and wheat production). The analysis involves comparing historical model experiments to observed crop data. The research uses the most recent experiments from GGCMI, driven by observations and bias-adjusted CMIP6 climate models, covering the historical period (1960-2014). Three key research questions were addressed: 1) How accurately do bias-adjusted CMIP6 simulations reproduce upper tropospheric jet patterns and associated surface weather anomalies? 2) Can crop models driven by reanalysis data and bias-adjusted CMIP6 simulations reliably reproduce observed regional crop yield anomalies related to specific wave patterns? 3) How do wave events modulate the concurrence of low yields in major crop-producing regions, and how do results from the crop model driven by reanalysis and climate models compare to observed signals of low yield concurrence? The study utilized ERA5 reanalysis data (1960-2014) for 2 m temperature, precipitation, and meridional winds. W5E5 data (1979-2014), a bias-adjusted ERA5 dataset, was also used. CMIP6 daily data for 2-meter temperature, precipitation, and meridional winds (1960-2014 and 2045-2099 under SSP5-8.5) were included. Data was detrended and processed to calculate weekly anomalies. Wave events were identified using a method based on Fourier decomposition of meridional wind fields. A compositing method was used to estimate regional composite yield anomalies for years with and without wave events. The Likelihood Multiplication Factor (LMF) was introduced to quantify the effect of wave events on concurrent low yields in pairs of crop-producing regions. The LPJmL crop model was driven by W5E5 reanalysis data and four bias-adjusted CMIP6 climate models to estimate the effects of high-amplitude waves on crop yield anomalies. Crop-producing regions were defined based on harvested area thresholds.

The study found that while upper tropospheric wave patterns are well-reproduced in bias-adjusted CMIP6 multi-model means, associated temperature and precipitation anomalies are largely underestimated. The underestimation is particularly pronounced in regions where crop-producing areas spatially align with wave-induced temperature anomalies. Comparisons between reanalysis data and model outputs show discrepancies in surface anomalies during wave events, which translate to underestimation of crop productivity risks in key regions. The occurrence of multiple wave events negatively impacts combined maize and wheat yields at regional and global levels, as observed in FAOSTAT data. However, this impact is not accurately reproduced by crop model experiments, especially when driven by bias-adjusted climate models. While the overall yield anomalies averaged across all regions show good agreement for wave-5, regional anomalies in models differ from the observed values. The study found a significant reduction in disagreements between observation-based and model-based assessments when the crop model is driven by bias-adjusted ERA5 reanalysis data (W5E5) as compared to GCM simulations. Future projections, based on the bias-adjusted models, show an increase in negative impacts on wave-7 for crop yields in certain regions, primarily due to amplified heat response. However, these projections have considerable uncertainty, particularly regarding the concurrence of poor yields. The occurrence of wave events increases the likelihood of concurrent low yields in pairs of major crop-producing regions, a finding mostly reproduced by historical model experiments, particularly for wave-5. However, regionally, yield losses are significantly underestimated in crop models driven by climate model outputs, while those driven by reanalysis data show more accurate responses. The underestimation of surface extremes and their impacts on yields, observed in CMIP5 models, persists in recent climate and crop simulations. The study identifies high-impact blind spots in current climate risk assessments, highlighting the need for further research to improve model accuracy.

The study's findings highlight significant biases in current climate and crop models' ability to accurately simulate the impacts of extreme weather events driven by meandering jet streams. The underestimation of surface temperature and precipitation anomalies during high-amplitude wave events translates to an underestimation of the risks to crop productivity. This has considerable implications for assessing and mitigating risks to global food security. The discrepancy between observed and modeled yield responses is particularly striking, emphasizing the need for improved model representation of complex interactions between atmospheric circulation, surface weather, and crop responses. While some improvement in model agreement is observed when using reanalysis data as input, limitations persist, especially concerning regional variations in model accuracy and the prediction of future changes in concurrent yield losses. The results stress the uncertainty surrounding future projections of synchronized crop failures and the need for further model development and validation. The inclusion of physically constrained machine learning methods might offer avenues for improvement. These findings underscore the importance of considering model limitations when assessing climate risks, emphasizing the need for a precautionary approach given the significant societal and economic implications of widespread crop failures.

This study reveals critical biases in climate and crop models' representation of synchronized crop failures caused by meandering jet stream patterns. The underestimation of surface weather anomalies and their impacts on yields suggests that current models may provide a conservative estimate of future risks. The findings highlight the urgent need for improved model development, incorporating physical constraints and process-based understanding, to better quantify the complex risks of concurrent extreme weather events and their implications for global food security. Further research should focus on resolving the identified model limitations and exploring the potential for physically constrained machine learning techniques to enhance prediction accuracy. The results emphasize the pressing need for rapid emission reductions to mitigate the potential for increasingly unmanageable climate extremes.

The study's findings are based on a specific set of climate and crop models, and the results may not be generalizable to all models. The analysis focuses on a limited number of crop-producing regions and crops (maize and wheat), which might not fully capture the global picture. The study uses bias-adjusted climate models, and any biases remaining in the adjusted data could still influence the results. The regional definition of breadbasket regions and the use of national yield statistics might have influenced the results. The sensitivity of the crop model to the timing of wave events within the growing season warrants further investigation. The analysis relies on historical data and may not fully capture the complexities of future climate scenarios.