Future changes in spatially compounding hot, wet or dry events and their implications for the world’s breadbasket regions

Recent years were characterized by an increase in spatially co-occurring hot, wet or dry extreme events around the globe. In this study we analyze data from multi-model climate projections to analyze the occurrence of spatially compounding events and area affected in future climates under scenarios at +1.5 ∘C, +2.0 ∘C, +3.0 ∘C and higher levels of global warming using Earth System Model simulations from the 6th Phase of the Coupled Model Intercomparison Project. Since spatially compounding extreme events can strongly amplify societal impacts as economic supply chains are increasingly interdependent, we want to highlight that the world’s breadbasket regions are projected to be particularly affected by an increase in spatially co-occurring hot, wet or dry extreme events, posing risks to the global food security. We show that the spatial extent of top-producing agricultural regions being potentially threatened by climate extremes will increase drastically if global mean temperatures shift from +1.5 ∘C to +2.0 ∘C. Further we identify a large increase in the projected global land area concurrently affected by hot, wet or dry extremes with increased global warming posing risk to other industries and sectors in addition to the agricultural sector.


Introduction
Recently, adverse weather and climate extreme events have occurred near-simultaneously at many locations around the globe.For instance, in the 2018 summer season, 37.5% of the northern hemispheric highly-populated or agricultural area was concurrently affected by hot days (Vogel et al 2019).The simultaneous occurrence of climate extremes at different locations but affecting similar sectors (e.g.breadbaskets) in different regions are named as spatially compounding or concurrent extremes (Vogel et al 2019, Zscheischler et al 2020, Seneviratne et al 2021).Spatially compounding extremes are projected to become more frequent with increasing global warming, in particular above +2.0• C of global warming (Sarhadi et al 2018, Vogel et al 2019, Seneviratne et al 2021, Singh et al 2021, Rogers et al 2022).However, the investigation of spatially compounding events is underrepresented in current research.Consequently, the IPCC AR6 report noted that 'very few studies investigate which types of concurrent extreme events could occur under increasing global warming' (Seneviratne et al 2021).
The simultaneous occurrence of climate extremes in different agricultural regions may impose a great risk to the increasingly interdependent global food supply chain, threatening global food security (Zhou et al 2023).Current levels of food insecurity are often exacerbated by short-term food shortages and price spikes caused by weather extremes partly linked to climate change (Kerr et al 2022).The recently published Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report (AR6) concludes with high confidence that 'climate change has reduced food security [...] due to warming, changing precipitation patterns, reduction and loss of cryospheric elements, and greater frequency and intensity of climatic extremes, thereby hindering efforts to meet Sustainable Development Goals' (IPCC 2023).Direct negative impacts of global climate change on food production will increase with further warming (Kerr et al 2022), partly due to the fact that multiple breadbasket regions might be affected by climate extremes near-simultaneously in time (Fraser et al 2013, Levermann 2013, Puma et al 2015, Anderson et al 2019, Haqiqi et al 2021, Seneviratne et al 2021, Kornhuber et al 2023).
Literature on climate change impacts on agriculture (e.g.Gaupp et al 2019, Kornhuber et al 2020, 2023, Haqiqi et al 2021) mainly focuses on yields.Insights about the amount of cropping area being affected by climate extremes remain elusive, although this quantity is an essential factor in determining global food production (Iizumi and Ramankutty 2015).In this study, we shift the focus from crop yield to the future suitability of global cropland area for food production.To complement the existing literature, we investigate how the land area is affected by spatially concurrent climate extremes in a changing climate.In particular, we focus on spatially compounding events since a recent study (Vogel et al 2019) has stressed the relevance of spatially compounding climate extremes for global food security.Vogel et al (2019) focused on the future spatial extent of agricultural land area being concurrently affected by hot days.The investigation of other spatially extreme event types remains elusive.Therefore, we calculate the spatial extent of agricultural area being concurrently affected by hot, wet or dry extreme events in a changing climate.We first focus on 'breadbasket regions' representing the world's highest-producing crop regions.We limit our analysis to primary stable crops constituting large parts of the human diet: wheat, maize, rice and soybean.In a second step, we expand the spatial focus beyond breadbaskets and examine how the global land area is affected by spatially compounding climate extremes in a changing climate.

Breadbaskets
To study future changes in the spatial extent of agricultural area being concurrently affected by hot, wet or dry extreme events, we employ a pragmatic and robust definition of breadbasket regions for wheat, maize, rice and soybeans, since they constitute large parts of the human diet.We restrict our analysis to countries with at least a 1% share of global production during 1981-2010 and then focus on the gridcells that account for 90% of a country's national output (figure 1).
Yearly country-level data on crop production (tons from 1981 to 2010) from the Food and Agriculture Organization of the United Nations (FAO 2023) is used to determine the top producing countries.The identified countries comprise at least 81% of global production (see supplementary tables 2-5).
For each of the selected countries, the spatial extent of a country-level breadbasket is then based on the highest crop producing area sourced from Monfreda et al (2008), representing the average crop production between 1979 and 2003.The identified breadbasket regions are thus handled as time invariant, assuming no shift, expansion or reduction in cropping extent over time.We first interpolate the global crop production dataset to a 2.5 • × 2.5 • grid using second-order conservative remapping to facilitate comparison with ESM data.In a second step, the highest producing grid cells of each individual crop producing country contributing to 90% of a country's national production were selected as a contributor to the national breadbasket.

Global land area
To investigate the exposure of other sectors than agricultural production to future changes in hot, wet or dry extreme events, we investigate the global land area being affected by these extreme event types.We focus on the global land area excluding Antarctica and Greenland.(Kuma et al 2023).We equally weigh the contributing models.For detailed information on the considered ESMs, see supplementary material S1.For all models and simulation runs daily maximum temperature, daily total precipitation and daily precipitation-evapotranspiration (P-E) are analyzed to estimate the concurrent extreme weather area.All CMIP6 data are prepared through a centralized preprocessing (Brunner et al 2020), including the interpolation of the model outputs to a common 2.5 • grid using second-order conservative remapping, the handling of inconsistencies in the dimensions and differences in the file structure and the exclusion of files with very unrealistic values.

Methods
The extreme event indicators applied in this study vary based on the spatial focus and the extreme event type.We first describe the spatial-dependent and event type-dependent seasonal definitions of the extreme event indicators.Subsequently, we describe the computation of concurrent hot spell, wet spell and dry spell area.In the last step we describe the translation of historical and future simulation runs to global warming levels (GWLs).

Growing season selection for the breadbasket's area affected by hot spells, wet spells or dry spells
For the crop-specific analysis based on its breadbasket's area affected by hot spells, wet spells or dry spells, the crop calendar provided by Sacks et al (2010) is used to extract the crop-specific main growing season per grid cell.The crop calendar represents the average planting and harvesting dates between the 1990s and early 2000s.The applied growing seasons are thus handled as time invariant, as we assume no shifts in growing seasons over time.Before extracting the crop-specific and grid cell-specific growing seasons, we interpolated the crop calendar dataset to a 2.5 • × 2.5 • grid using second-order conservative remapping to facilitate comparison with ESM data.

Season selection for the global land area affected by hot spells, wet spells or dry spells
Regarding the global analysis of hot spells and wet spells, the seasonal definition depends on the location of the grid cell.For grid cells within tropical latitudes (⩽23.5 • ), hot spells and dry spells are calculated based on all twelve months of each year, whereas in extratropical latitudes (>23.5 • ), hot spells and dry spells are identified from the period June-August in the Northern Hemisphere and from the period December-February in the Southern Hemisphere.Wet spells are calculated based on the whole year.

Computation of concurrent hot spell, wet spell and dry spell area
Hot spells, wet spells and dry spells are characterized using a percentile-based event definition.The percentile climatology as well as the occurrence of the event type is calculated for each grid cell of the simulated data.
Hot spells are defined based on a 14-day moving average of daily maximum temperature and its subsequently extracted seasonal maximum.The 95 th percentile climatology is calculated based on the reference period 1850-1900.Each extracted seasonal 14day temperature maximum that exceeds the 95 th percentile climatology is defined as a season in which a hot spell occurs.
Wet spells are defined based on a five-day moving precipitation sum and its subsequently extracted seasonal maximum.The 95 th percentile climatology is calculated based on the reference period 1850-1900.Each extracted seasonal five-day precipitation sum that exceeds the amount of precipitation in the 95 th percentile climatology is defined as a season in which a wet spell occurs.
Dry spells are defined based on the 31-day precipitation-evapotranspiration (P-E) moving sum and its subsequently extracted seasonal minimum.The 95 th percentile climatology is calculated based on the reference period 1850-1900.Each extracted seasonal 31-day P-E minimum that undershoots the percentile climatology is defined as a season in which a dry spell occurs.
In addition, we investigate the multivariate occurrence of the above-mentioned extreme events.Each season between the respective selected period based on the location of the grid cell that is affected by at least one of the above-mentioned extreme events is defined as a season in which at least one of the extreme events occurs.
Referring to the seasonal definition of extreme events described in sections 3.1.1and 3.1.2,the seasonal definition depends on the spatial extent and the extreme event type.E.g. for analyzing the maizebreadbasket's area affected by extreme events, both the reference period and occurrence of hot, wet, and dry spells are limited to the growing season of maize.
We introduce the percentage of land area that is affected by hot spells, wet spells or dry spells as a metric for characterizing spatially co-occurring extreme events.The relative concurrent hot spell, wet spell or dry spell area is obtained by dividing the areaweighted mean of affected grid cells by the cropspecific breadbasket's extension or by the global land area.

Calculation of Global Warming Levels
We translate the historical and future simulation runs to global warming levels (GWL).In order to allow comparability with current literature, we follow the approach introduced by Arias et al (2021).We compute 20-year moving averages of global mean temperature (T glob ) for all CMIP6 models between 1850 and 2100 and determine changes relative to the baseline period 1850-1900.We estimate T glob as a weighted mean between near-surface air temperature (tas in CMIP6) over land and (dynamic) sea ice regions and sea surface temperature (ts in CMIP6) over the oceans.
The GWLs of +1.0 • C, +1.5 • C, +2.0 • C, +3.0 • C and +4.0 • C are determined as 20-year periods such that T glob in the center year is closest to the respective warming level.The intermediate GWL of 1.5 • C is chosen due to its policy-relevance regarding the legally binding Paris Agreement to limit global temperature increase to 1.5 • C above pre-industrial levels (IPCC 2018).

Future projections of concurrent breadbasket's area affected by hot, wet or dry spells
All investigated crop-specific breadbasket regions show a steady increase in concurrent land area being affected by hot, wet or dry spells (figure 2).The following paragraph refers to the median exposure to hot and/or wet and/or dry spells calculated based on the multi-model median and the median exposure of the 20 years contributing to the investigated warming level (model-dependent).At least 20% of the land area of all studied crop-specific breadbaskets except for winter wheat is calculated to be affected by hot spells under +1.0 • C global warming and increases up to 35% and approximately 50% under higher warming of +1.5 • C and +2.0 • C, respectively.The major rice-producing region is most affected by hot spells up to a GWL of +2.0 • C, whereas differences in affected crop type-specific land area decrease under higher warming levels.Under +3.0 • C and +4.0 • C warming, about 70% and 90% of land area, respectively, is affected by hot spells, regardless of the investigated crop.The winter wheat-production is the least affected by climate extremes.This is mainly due to a large part of the growing season occurring during winter, whereas pronounced increases in temperature extremes are projected to occur especially during the summer months (Seneviratne et al 2021).Under +1.0 • C of global warming, around 15% of the winter wheat-producing region is affected by a hot spell, increasing up to 64% under +4.0 • C of global warming.
The projected increase in the global breadbaskets' land area being affected by wet spells or dry spells is less pronounced compared to hot spells.While the median land area affected by wet spells is below 10% up to a global warming of +1.5 • C, regardless of the investigated crop type, especially the maize-, riceand soy-producing regions show more pronounced increases in land area affected by wet spells under higher levels of global warming.The median projects an increase in land area affected by wet spells of up to 20% under +4.0 • C of global warming regarding the maize-, rice-and soy-producing regions.
The multi-model median area affected by dry spells is below 10% up to a global warming of +2.0 • C, regardless of the investigated crop type, and increases up to 12% under higher levels of global warming.In contrast to the breadbaskets' area affected by wet spells, the land area affected by dry spells does not differ so much depending on the crop type.
The land area affected by at least one of the studied extreme events is already about 40% under +1.0 • C of global warming, regardless of the studied crop type (except for winter wheat).The land area affected by either a hot, wet or dry spell increases up to 55% under +1.5 • C of global warming, reaching 85% to over 90% of land area affected under higher levels of warming of up to +3.0 • C or +4.0 • C.

Future projections of global concurrent land area affected by hot, wet or dry spells
Figure 3 provides spatial insights into the changes in hot, wet or dry spells.Displayed is the number of extreme events occurring within a 20-year period per grid cell.Hence, for each GWL, there can be at maximum 20 events.Independent of the investigated extreme event type and location, a steady increase in the temporal sequence of hot, wet or dry spells can be seen under continued global warming.Globally, four out of 20 years (multi-model median) experience a hot spell under the present +1.0 • C of global warming.This increases up to eight and twelve out of 20 The global land area affected by climate extremes in figure 3 follows the patterns observed in the breadbaskets' area affected (figure 2), even though the former analysis is not constrained to crop-specific growing seasons.E.g. the grid cells corresponding to the rice-breadbasket (mainly south-east Asia and north-eastern South America) are one of the highest affected grid cells by hot spells.The temporal sequence of hot spells increases up to a global median of 18 out 20 years being affected by a hot spell under +3.0 • C of global warming, whereas under the higher level of warming up to +4 • C every year will be affected.The rice-breadbasket was also identified as being the region which is the highest affected by hot spells.Similar patterns emerge regarding the high exposure to wet and dry spells of grid cells corresponding to the maize and soy breadbaskets.In addition, figure 3 shows, that some regions located outside the identified breadbaskets are projected to also be particularly affected by climate extremes.Under the current and continued global warming of +1.The concurrent global land area affected by hot, wet or dry spells (figure 4) has increased steadily since the mid of last century and is projected to further intensify under the highest emission scenario.The global land area affected by hot spells is projected to increase stronger than the global land area affected by wet or dry spells.While the average concurrent global land area affected by hot, wet or dry spells covered around 5% (multi-model median) during the early-industrial period (1850-1900), hot spells are projected to cover 96%-98% (multi-model

Technical choices and limitations
This study provides a global assessment of different crop-dependent breadbaskets as well as the global land area affected by current and future spatially concurrent hot, wet and dry spells, considering a wide range of ESMs to account for model uncertainty.The analysis is built on a comprehensive set of global climate model simulations that cover the earlyindustrial period and extend to the end of the 21st century.Although the considered ESMs are limited with respect to their spatial resolution, we note that higher resolution models are typically not available at a global scale or do not cover the required timeperiod.Moreover, statistically downscaled ESM data (e.g.Karger et al 2017, Lange 2019, Gebrechorkos et al 2023), are often limited in terms of the number of considered models and also do not provide evapotranspiration, which is essential for our analysis.The study is subject to the reliability of and confidence in ESM comparisons representing a realistic longterm warming range of models (Tokarska et al 2020, Yazdandoost et al 2021, Kornhuber et al 2023).Recent literature reports on an underestimation of the magnitude of temperature and precipitation anomalies in CMIP6 models especially during a strongly meandering jet stream (Kornhuber et al 2023).This underestimation may translate into a conservative estimate of crop production risk in this study (Kornhuber et al 2023).
The investigated extreme events are defined relative to percentiles, which depend on a climatological baseline period (Thomas et al 2023).In order to find a reasonable trade-off between low probability-high impact events posing potential future threats to global food production from a climatological perspective and a robust statistical estimation of these rare events given the 50-year baseline period (e.g.Gessner et al 2021), we define extreme events as those that occur in the highest or lowest 5% of the selected baseline period.Further, the frequency and extent of events is subject to their defined duration.To test the dependency of our results to the duration definition, we conducted a sensitivity analysis (supplementary material figure 1), which shows only minor sensitivities to the event duration, not changing the overall conclusion.Recent studies focused on crop-and regionspecific (absolute) thresholds for extreme events during critical phenological phases (e.g.Schauberger et al 2017, Schmitt et al 2022).However, applying absolute thresholds to globally distributed agricultural land is not suitable since what is regarded as an extreme weather incident in one region of the world is not automatically considered a threat elsewhere (Schmitt et al 2022).Current and future crop production is not only modulated by climate change but by multiple factors such as agronomic technology innovations, agricultural adaption capacity to climate change, farmer decision-making, changes in cropping intensity or a shift in cropping areas or growing seasons due to gradual climate changes (Iizumi and Ramankutty 2015).However, these factors are not considered in our study as we focus on shifts in frequency and extent of concurrent climate extreme events over agricultural areas.

Implications and outlook
We demonstrated in this study that besides the projected increase of climate extremes happening nearsimultaneously around the globe, the world's breadbasket regions will be particularly affected by concurrent climate extremes.The increasing frequency of climate extremes is and will be exacerbating the pressure on scarce cropland resources, compounding existing challenges from rising demand and environmental constraints.Cropland is already a scarce resource under current climate conditions.Due to soil erosion, chemical and physical degradation, desertification, and encroaching human settlements, the global arable land area declined by one-third during the last 40 years (Hubacek and Sun 2001, Weinzettel et al 2013, Cameron et al 2015, Lasanta et al 2017, Wu et al 2018).The climatological suitability of current global food production areas is projected to further decrease by 31% until the end of this century due to a change in mean climatic factors for agriculture (Kummu et al 2021).Beyond climate change-driven impacts on future agricultural production, this sector is facing additional future challenges due to both increasing demands and environmental constraints.By 2050, the world's population is expected to grow by almost two billion (United Nations 2022), significantly amplifying the need for agricultural products.Additionally, resource-intensive diets are projected to become more prevalent, further straining land and water resources (Tilman and Clark 2014, Ringler and Zhu 2015, Bodirsky et al 2020).The growing demand for food production is compounded by the emerging reliance of nations on the land sector to meet climate change mitigation goals by producing biofuels and reforestation, both competing for land with the cultivation of food production (Bonsch et al 2016, Humpenöder et al 2018, Fujimori et al 2019).So far rising demand has been satisfied by increasing yields through methods such as nitrogen fertilizers, mechanization, and irrigation, among others (Eickhout et al 2006, Erisman et al 2008, Wu et al 2014).However, part of this approach is proving unsustainable as agricultural production already crosses the limits for sustainable human activities on the planet in terms of nitrogen and water usage (Rockström et al 2009, Gerten et al 2020).Relying on these historical methods of increasing yield may no longer be viable if we aim to stay within the safe space of human activity on Earth.
Here it is important to note that food production is a function of both crop yield (harvested production per unit of harvested area), cropping intensity (number of crops grown within a year) and cropping area: Nevertheless, we recognize that most existing work (e.g.Gaupp et al 2019, Kornhuber et al 2020, 2023, Haqiqi et al 2021) on climate change effects on global food production focuses on yield.However, cropping area is an important overlooked component of global food production due to the following two reasons: (1) Future food production may not solely rely on higher yields, as past increases already surpass environmental limits.However, expanding rather than intensifying food production might not provide a sufficient remedy due to the projected increase in cropping area being concurrently affected by climate extremes, as shown in this study.(2) Only focusing on crop yields may offer an incomplete picture due to the potential for compensating or compounding yield impacts with changes in harvested area (Iizumi and Ramankutty 2015).Yield response mainly reflects crops' underlying biophysical response, but economic conditions and farmer decision-making greatly influence which other components of crop production might be affected by climate (Roberts et al 2006, Iizumi and Ramankutty 2015, Schmitt et al 2022).Thus, the impact of weather extremes which result in farmers' behavioral response to crop abandonment due to damaged crops or complete crop failure can be blurred (Iizumi and Ramankutty 2015, Cui 2020, Schmitt et al 2022).
Our results emphasize the importance of climate change impacts on food production beyond yield.In particular, we showed that the spatial extent of agricultural land potentially threatened by climate extremes will increase drastically if global mean temperatures shift from +1.5 • C to +2.0 • C, and will be further amplified with every tenth degree of warming.If we want to keep the global cropland in a safe climatic space, ambitious climate action needs to be taken in order to limit global warming and the subsequent impacts on the global food sector.This holds true not only for the global food sector.An increase in global land area being near-simultaneously affected by climate extremes is highlighted in this study, posing risks to a wide range of other sectors besides agricultural production (e.g.ecosystem services, health (services), transport, global supply chains etc), if we do not limit global warming.

Figure 2 .
Figure 2. Concurrent breadbaskets' area (crop-dependent) being affected by hot, wet or dry spells, as well as the breadbasket's area being concurrently affected by a hot and/or wet and/or dry spell under different warming levels.The breadbaskets' area is divided into the primary staple crops: winter wheat, spring wheat, maize, rice, soy according to figure 1.The bar extent refers to the distribution based on the model-and year-to-year variability.
0 • C and +1.5 • C, the Sub-Saharan Africa, Southern Africa and Northern South America are affected by the highest temporal sequence of hot spells.The increase in temporal sequence is not as pronounced for wet and dry spells.The spatial patterns, however, are comparable to those of hot spells.Whereas under the present warming of +1.0 • C, one out of 20 years experiences a wet spell or a dry spell (global median), this increases up to 2 and 3 out of 20 years under +2.0 • C and +4.0 • C of global warming, respectively.The temporal sequence of the multivariate indicator is general much higher than the ones of the univariate indicators.Under the present warming of +1.0 • C, seven out of 20 years experience either a hot, wet or dry spell.This increases up to 10/14/19/20 out of 20 years under +1.5 • C/+2.0 • C/+3.0 • C/+4.0 • C of global warming.

Figure 3 .
Figure 3. Temporal sequence of the occurrence of hot, wet or dry spells for different warming levels.The last column shows the occurrence of a hot and/or wet and/or dry spell.Displayed is the number of years affected out of 20 years.Hot, wet and dry spells are calculated on an annual scale as described in section 3.1.2.The black polygons mark the location of the breadbasket regions (location of all identified breadbasket regions according to figure 1).

Figure 4 .
Figure 4. Modeled global concurrent hot, wet or dry spell area as well as the land area being concurrently affected by a hot and/or wet and/or dry spell.Left column: temporal evolution from 1850 to 2100 based on Coupled Model Intercomparison Project phase 6 models.Right column: concurrent land area affected as a function of global mean temperature T glob increase.For the models, we used a high-emission scenario (SSP5-8.5).We show the multi-model median, interquartile range, and the 10th-90th percentile range.
(Monfreda et al 2008)'Neill et al 2016, 2017)udy are included within the article (and any supplementary files).All original CMIP6 data(Eyring et al 2016, O'Neill et al 2016, 2017)used in this study are publicly available https://esgf-node.ipsl.upmc.fr/search/cmip6-ipsl/.The models and ensemble members used within this study are listed in tableS1.Yearly country-level data on crop production (FAO 2023) can be retrieved from the Food and Agriculture Organization of the United Nations website www.fao.org/faostat/en/#data.The harvested area data based upon(Monfreda et al 2008)is publicly available www.earthstat.org/cropland-pasture-area-2000/.The crop calendar (Sacks et al 2010) is publicly available https://sage.nelson.wisc.edu/data-and-models/datasets/crop-calendar-dataset/.et al 2019 A multi-model assessment of food security implications of climate change mitigation Nat.Sustain. 2 386-96 Gaupp F, Hall J, Mitchell D and Dadson S 2019 Increasing risks of multiple breadbasket failure under 1.5 • C and 2 • C global warming Agric.Syst.175 34-45 Gebrechorkos S et al 2023 A high-resolution daily global dataset of statistically downscaled CMIP6 models for climate impact analyses Sci.Data 10 611 Gerten D et al 2020 Feeding ten billion people is possible within four terrestrial planetary boundaries Nat.Sustain. 3 200-8 Gessner C, Fischer E M, Beyerle U and Knutti R 2021 Very rare heat extremes: quantifying and understanding using ensemble re-initialization J. Clim.34 1-46 Haqiqi I, Grogan D S, Hertel T W and Schlenker W 2021 Quantifying the impacts of compound extremes on agriculture Hydrol.Earth Syst.Sci. 25 551-64 Hubacek K and Sun L 2001 A scenario analysis of china's land use and land cover change: incorporating biophysical information into input-output modeling Struct.Change Econ.Dyn. 12 367-97 Humpenöder F et al 2018 Large-scale bioenergy production: how to resolve sustainability trade-offs?Environ.Res.Lett.13 024011 Iizumi T and Ramankutty N 2015 How do weather and climate influence cropping area and intensity?Glob.Food Secur. 4 46-50 IPCC 2023 Current status and trends Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change ed Core Writing Team, H Lee and J Romero (IPCC) pp 35-115 IPCC 2018 Summary for policymakers Global Warming of 1.5 • C.An IPCC Special Report on the impacts of Global Warming of 1.5 • C Above Pre-Industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change, Sustainable Development and Efforts to Eradicate Poverty ed V Masson-Delmotte et al (Cambridge University Press) pp 3-24 Karger D N, Conrad O, Böhner J, Kawohl T, Kreft H, Soria-Auza R W, Zimmermann N E, Linder H P and Kessler M 2017 Climatologies at high resolution for the earth's land surface areas Sci.Data 4 170122 Kerr R B et al 2022 Food, fibre, and other ecosystem products Climate Change 2022: Impacts, Adaptation and Vulnerability.Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change ed H-O Pörtner et al (Cambridge University Press) pp 713-906 Kornhuber K, Coumou D, Vogel E, Lesk C, Donges J F, Lehmann J and Horton R M 2020 Amplified rossby waves enhance risk of concurrent heatwaves in major breadbasket regions Nat.Clim.Change 10 48-53 Kornhuber K, Lesk C, Schleussner C F, Jägermeyr J, Pfleiderer P and Horton R M 2023 Risks of synchronized low yields are underestimated in climate and crop model projections Nat.