Risks of synchronized low yields are underestimated in climate and crop model projections
Published July 4 in Nature Communications Vol. 14:3528
Risks of synchronized low yields are underestimated in climate and crop model projections
Kai Kornhuber, Corey Lesk, Carl F. Schleussner, Jonas Jägermeyr, Peter Pfleiderer & Radley M. Horton.
Nature Communications volume 14, Article number: 3528 (2023)
Open Access: This article is licensed under a Creative Commons Attribution 4.0 International License
Below is the full text of Title, Authors, Publication, Abstract and Discussion (conclusions). Last sentence is very important. Please visit the Open Access study at the DOI or nature.com sites (each visit records another “access”), there are many interesting graphics and an incredible amount of data-wrangling. This study reveals an existential threat to world-wide food production. First, we run population numbers up to the planet's maximum carrying capacity, then find that crops fail from our incessant burning. -KM
Title: Risks of synchronized low yields are underestimated in climate and crop model projections
Authors: Kai Kornhuber, Corey Lesk, Carl F. Schleussner, Jonas Jägermeyr, Peter Pfleiderer & Radley M. Horton
Nature Communications volume 14, Article number: 3528 (2023)
DOI: 10.1038/s41467-023-38906-7 https://www.nature.com/articles/s41467-023-38906-7
Published Jul 4 2023
Abstract
Simultaneous harvest failures across major crop-producing regions are a threat to global food security. Concurrent weather extremes driven by a strongly meandering jet stream could trigger such events, but so far this has not been quantified. Specifically, the ability of state-of-the art crop and climate models to adequately reproduce such high impact events is a crucial component for estimating risks to global food security. Here we find an increased likelihood of concurrent low yields during summers featuring meandering jets in observations and models. While climate models accurately simulate atmospheric patterns, associated surface weather anomalies and negative effects on crop responses are mostly underestimated in bias-adjusted simulations. Given the identified model biases, future assessments of regional and concurrent crop losses from meandering jet states remain highly uncertain. Our results suggest that model-blind spots for such high-impact but deeply-uncertain hazards have to be anticipated and accounted for in meaningful climate risk assessments.
Discussion
Concurrent crop failures in major crop-producing regions constitute a systemic risk as associated spikes in food prices can lead to conflict and undernutrition in countries that rely on imports1,3. Thus, understanding the likelihood of concurrent crop failures and the degree to which models are able to reproduce observed relationships is important for increasing the resilience of the global food system15 and mitigating climate risks.
While circulation patterns associated with high amplitude Rossby waves are accurately reproduced in climate models (Fig.1, Table S1), the magnitude of surface temperature and precipitation anomalies is largely underestimated, including in important crop producing regions (Figs. 1, 2). Similar results have been reported for other climate models mostly following the CMIP5 protocol31, however it is notable, that bias-adjusted output from CMIP6 experiments do not exhibit considerably improved spatial correlation and an accurate magnitude in surface anomaly fields (Fig. 1, Tables S1, S2), possibly because bias adjustment optimizes fields for different subsamples rather than high amplitude wave events.
Investigating future projections under a high emission scenario, we find no consistent global increase in wave amplitudes in models (Figs. S1, S2). This might be due to the fact that our wave diagnostic is applied to the mid-latitudes (37.5–57.5°N) and is therefore not sensitive to suggested37 changes in sinuosity in higher latitudes where the increased temperature gradient from increased land warming increases zonal winds and improves waveguidability37,38. Instead, lower wave amplitudes might be a consequence of the projected weakening of summertime stormtracks, associated with an increase in weather persistence39,40 in the mid-latitudes39,41, which might lower the magnitude of meridional winds on which the wave diagnostic is based on (see methods). We identify a regional amplification of troughs and ridges, in particular over the NA West coast and Eurasia (Fig. 1d, h, Fig. 2b, i, f) for wave-7 and wave-5 respectively. Amplified land-atmosphere feedbacks which are acting on top of regional circulation changes in a warmer climate42,43,44 are other potential factors for an increased regional temperature response. With the impacts of recent extreme heat events45,46 and associated wildfires47,48, such as the severe Pacific northwestern heatwave of 202145 and the extraordinary Siberian heatwave of 202049, these regions potentially emerge as high risk areas.
We find that simultaneous extremes linked to a meandering jet stream from amplified Rossby waves19 lead to regional yield losses (Fig. 3) and to concurrent low harvests across the mid-latitudes (Fig. 4). This increased likelihood of concurrent low yields in major breadbaskets, is mostly reproduced by historical model experiments, whether driven by reanalysis data or climate models in particular for wave-5 (Fig. 4). Regionally, however, we find that yield losses are mostly underestimated in crop models driven by climate model output (Fig. 3), while crop models driven by reanalysis data show more accurate responses.
The mostly adequate model representation of concurrent low yields combined with predominant underestimation of local impacts on yields, parallels the reasonable representation of modeled wave patterns and an underestimation of associated surface anomalies in bias-adjusted model output.
The biases in modelled crop yield response to wave event draw the reliability of future projections into question, but the societal importance of yield projections and the absence of a better approach argue for discussion under consideration of identified caveats. Increased local impacts on yields are identified in regions where future surface anomalies are projected to increase (e.g. NA for wave-7 and EAS for wave-5 (Fig. 2b, I, Fig. 3). However, the projected concurrence of poor yields, is found to be less conclusive (Fig. 4d, h), as models show divergent responses to future warming scenarios (Figs. S15, S16).
With positive trends observed in magnitude and frequency of extreme weather events, in particular for extreme heat, concurrent weather extremes causing interconnected and potentially disruptive impacts have become more common and will increase further if greenhouse gas emissions remain unmitigated8,9,50. Informed adaptation measures depend on models that simulate not just mean changes but also the changes in complex low probability but high-risk scenarios such as the concurrent and persistent extreme heat and rainfall extremes as observed e.g. in the extreme summers of 201825,51 in Europe and Russia in 201028,52, both with severe agricultural impacts53,54.
Assessing these complex risks depends upon an adequate representation of the location, magnitude, frequency and sub-seasonal distribution of extreme weather events, which may change under future emission scenarios. Our results highlight how the evolution of risks of multiple breadbasket failures under climate change are characterized by deep uncertainty in part due to the insufficient representation of the underlying climate impact drivers in models55. Other major observational and modelling uncertainties regarding climate change risks to global crop production include the magnitude of the CO2 fertilization effect and general changes in the hydrologic cycle. Our results point to an additional modelling uncertainty with import specifically for inter-regional crop yield covariability, which have unique consequences for the global food system. Future work should examine potential interactions among these key uncertainties, particularly the potential modulation of jet-related hazards by mean hydrologic and thermodynamic change.
While climate models have been excellent in projecting the mean response to continued anthropogenic greenhouse gas emissions56, our analysis suggests that they might provide a conservative estimate of how concurrent extreme weather events driven by specific circulation regimes might evolve in future and how they might affect regional crop yield and covariability across regions. Further we highlight that the underestimation of surface extremes identified in CMIP5 models31 and impacts on yields from their bias-adjusted output still persist in most recent climate and crop simulations. Physically constrained machine learning methods designed to maintain patterns and coherence across variables might offer an effective tool for an improved bias adjustment for more accurate impact assessments57.
Our study points towards potential high-impact blind spots in current climate risk assessments, highlighting the urgent need for more empirical and process-based research to support model improvements in the climate and agriculture domains, supplemented by expert elicitation, qualitative storylines58, and decision-centric approaches59. Evidence for high-risk blind spots such as an underestimation of synchronized harvest failures as identified here, manifests the urgency of rapid emission reductions, lest climate extremes and their complex interactions might increasingly become unmanageable.