Record-breaking heat extremes are becoming more likely due to anthropogenic climate change, with their probability depending on the regional warming rate. Anthropogenic aerosol forcing modulates these warming rates, and aerosols are declining globally. However, the influence of aerosols on the probability of record-breaking heat extremes remains unclear. Here, we assess how aerosol trends alter record-breaking heat probabilities by combining idealized simulations, CESM2 large ensemble and single-forcing large ensemble simulations, Regional Aerosol Model Intercomparison Project (RAMIP) future aerosol scenarios, and ERA5/MERRA-2 reanalysis data. Reanalysis data provide empirical evidence for a link between changes in aerosol concentrations and record-breaking heat extremes. There is a statistically significant correlation between changes in sulfate aerosol optical depth (SO4AOD) and record occurrence after accounting for latitudinal differences in record probability, with a 0.5 percentage points increase in annual record-breaking probability for every 10% decrease in SO4AOD. For context, the global land-mean annual record-breaking probability during 2010--2025 is 6.7%. Consistent with this empirical relationship, idealized simulations show that increased warming rates can lead to higher record-breaking probabilities, with a lagged response. In CESM2, present-day record-breaking probabilities are amplified relative to GHG-only across most of the world, e.g. by 67% in Central Europe where aerosol forcing declines, whereas they are damped by 47% in South Asia where aerosol forcing increases. RAMIP simulations show that aerosol reductions would increase record-breaking probabilities, especially in regions with high remaining aerosol concentrations. For example, reducing global anthropogenic aerosol emissions from SSP3-7.0 to SSP1-2.6 results in an annual record-breaking heat probability of 11.4% instead of 6.7% in parts of South Asia in the 2040s. These results show that regional aerosol trends can substantially modulate record-breaking heat extremes by altering regional warming rates. Aerosol reductions can therefore temporarily amplify record-breaking heat probabilities, although greenhouse gas forcing remains the dominant driver of long-term increases.

The pursuit of a climate-neutral future accelerates the expansion of renewable energy sources such as wind and solar power. A key challenge for these technologies is their sensitivity to weather variability. Periods of extremely low renewable energy production coinciding with high demand, so-called variable renewable energy shortages (VRES), are especially critical for grid stability. VRES are commonly studied with observational data, which only represent one possible realisation of internal climate variability and are therefore not sufficient to adequately study extreme VRES. Additionally, the effect of non-stationarity under climate change on VRES is only starting to be investigated and is not yet fully understood. In this study, we evaluate a simple energy model for Germany using ERA5 reanalysis data and apply it to 3840 years of climate data from different Earth System Models (ESMs) under current climate and in a climate with +2 K global warming. We analyse the model outputs with regard to seasonality, event magnitude, duration, composition, and large-scale weather regimes. We find that VRES events exhibit a pronounced seasonality, with both occurrence and magnitude peaking in mid-winter. We identify variability in wind speed over Germany as the main driver of these events, while fluctuations in incoming solar radiation contribute only marginally during the winter months. The most extreme physically plausible VRES derived from ESM data are up to 6.11 % (1-day events), 8.83 % (7-day events), and 10.44 % (14-day events) more severe than those observed in the same climate period. The energy production only changes marginally under warming conditions, while the demand is decreasing due to higher winter temperatures. This results in decreasing VRES event numbers while the magnitude of potential worst-case events remains extreme. In addition, a bottom-up weather pattern clustering analysis identifies different blocked circulation regimes as a dominant driver in the formation of VRES over Germany. Among the weather patterns leading to VRES events, two blocking patterns are particularly relevant for longer-lasting VRES events due to their particularly cold temperatures over Germany. These results provide energy system planners with robust worst-case benchmarks for VRES severity under current and future warming and contribute to a deeper understanding of the climatic drivers behind periods of renewable energy shortfall.

Due to climate change, heatwaves are becoming more frequent and intense, with western Europe experiencing the strongest trends in the Northern Hemisphere mid-latitudes. Part of the temperature trends are caused by circulation changes, which are not accurately captured in climate models. Here we deploy Deep Learning techniques to classify European heatwaves based on their atmospheric circulation and to study their associated changes over time. We use a Variational Autoencoder (VAE) to reduce the dimensionality of the heatwave samples, after which we cluster them on their extraced features. The VAE is trained on large ensemble climate model simulations and we show that the VAE generalizes well to observed heatwave circulations in ERA5 reanalysis, without the need for transfer learning. The circulation features relevant for heatwaves in ERA5 are consistent with the climate model heatwaves. Regression analysis reveals that the Atlantic Plume type of heatwaves are becoming more frequent over time, while the Atlantic High heatwaves are becoming less frequent. We introduce new and straightforward interpretability methods to study the latent space, including feature importance identification and changes over time. We investigate which circulation features are associated with the most important nodes in the latent space and how the latent space changes over time. For example, we find that the Atlantic Low heatwave shows a deepening of the low pressure system off the Atlantic coast over time. Each heatwave type is undergoing unique changes in their circulation, highlighting the necessity to study each heatwave type separately. Our method can furthermore be used to boost specific aspects of extreme events, and we illustrate how heatwave circulation could change in the future if the current trends persist, with in some cases an intensification of features.

Due to climate change, heatwaves are becoming more frequent and intense, with western Europe experiencing the strongest trends in the Northern Hemisphere mid-latitudes. Part of the temperature trends are caused by circulation changes. Here we deploy Deep Learning techniques to classify European heatwaves based on their atmospheric circulation and to study their associated changes over time. We train a Variational Autoencoder (VAE) to reduce the dimensionality of the heatwave samples, on large ensemble climate model simulations. We show that the VAE generalizes well to observed heatwave circulations in ERA5 reanalysis, without the need for transfer learning. We then use probabilistic clustering techniques to find physically distinct heatwave types in ERA5 reanalysis. The circulation features relevant for heatwaves in ERA5 are consistent with the climate model heatwaves. Regression analysis reveals that the Atlantic Plume type of heatwaves are becoming more frequent over time, while the Atlantic High heatwaves are becoming less frequent. Then, we introduce new and straightforward interpretability methods to study the latent space, including feature importance identification for the clustering model. We investigate which circulation features are associated with the most important nodes in the latent space and how the latent space changes over time. For example, we find that the Atlantic Low heatwave shows a deepening of the low pressure system off the Atlantic coast over time. Each heatwave type is undergoing unique changes in their circulation, highlighting the necessity to study each heatwave type separately. Our method can furthermore be used to boost specific aspects of extreme events, and we illustrate how heatwave circulation could change in the future if the current trends persist, with in some cases an intensification of features.

We investigate the potential of utilizing the precipitation emulator MESMER-M-TP as a tool to gain insights into precipitation patterns derived from Earth System Models (ESMs). MESMER-M-TP generates spatially explicit, monthly mean precipitation fields (2.5°x2.5° resolution) by employing spatially explicit, monthly mean temperatures as input. The approach involves modeling local precipitation as the response variable of a generalized linear model (GLM) with local modes of temperature variability as predictive variables.The emulator is trained on 24 different ESMs from the CMIP6 dataset based on a single ensemble member across Shared Socioeconomic Pathways (SSPs). This results in a set of 24 distinct parameter sets for each month and location. These parameters link precipitation to temperature via the GLM, providing a basis for quantitatively analyzing inter-model differences and parametric uncertainties. We focus on three key aspects: (1) Investigating parameter distributions to identify locations and months with poor inter-model agreement and understanding how individual predictors contribute to overall discrepancies. (2) Utilizing a clustering-based approach to group the 24 climate models based on their parameters, revealing consistency with genealogy and code streams of CMIP6 model development. (3) Exploring the sensitivity of the emulator to parameter choices.This explorative analysis offers valuable insights into the intricacies of precipitation modeling in ESMs by providing a quantitative understanding of inter-model variations and exploring sampling strategies that take inter-model variations into acocunt.

High-impact climate damages are often driven by compounding conditions, such as elevated heat stress arising from combined high humidity and temperatures. To explore future changes in compounding hazards under several climate scenarios, climate emulators can provide light-weight, data-driven complements to Earth System Models (ESMs). Yet, only a few existing emulators jointly emulate multiple climate variables. We introduce MERCURY (Multi-resolution EmulatoR for CompoUnd climate Risk analYsis), a spatio-temporal, multi-resolution emulator designed for compound climate risk analysis. MERCURY employs image-compression-based techniques for memory-efficient emulation and consists of two main modules. The regional module represents the monthly, regional response of a given variable to yearly Global Mean Temperature using a probabilistic additive model, resolving regional cross-correlations. The resulting regional values are then jointly disaggregated to grid-cell level values using a lifting-scheme operator, founded on principles of Discrete Wavelet Transforms. We demonstrate MERCURY on the humid-heat metric, wet bulb globe temperature (WBGT), as derived from temperature and relative humidity emulations. The emulated WBGT spatial correlations correspond well to those of ESMs and the 95 and 97.5 quantiles of WBGT distributions are well captured, with an average of 5 deviation. MERCURY's setup allows for region-specific emulations from which one can efficiently "zoom" into the grid-cell level across multiple variables by means of the reverse lifting-scheme operator. This circumvents the traditional problem of having to emulate complete, global-fields of climate data and resulting storage requirements.

Simple climate models (SCMs) are widely used to simulate global mean temperature (GMT) trajectories across a wide range of emission scenarios by combining simplified representations of the carbon cycle and other Earth system processes. These simulations depend on uncertain Earth system parameters, and ensembles of SCM simulations are created by exploring plausible parameter sets, resulting in scenario-specific distributions of GMT for all considered years.In this work, we introduce RIME-X (Rapid Impact Model Emulator Extended), a novel emulator approach that extends SCM outputs by translating GMT distributions into distributions of regionally aggregated climate or climate impact indicators. RIME-X uses historical and scenario simulations from climate and impact modeling intercomparison projects, such as CMIP and ISIMIP, to record relationships between global warming levels and indicators. By extracting distributions of indicators at specific global warming levels from those records and combining them with the GMT distributions from SCM ensembles, RIME-X produces scenario-dependent distributions of these indicators over time.This framework integrates multiple sources of uncertainty along the modeling chain, including model uncertainty (from diverse climate or impact model records), Earth system parameter uncertainty (from SCM ensembles), and internal variability, depending on the indicator’s temporal resolution.RIME-X is broadly applicable to any indicator whose distribution is predominantly influenced by the global warming level, offering a versatile and efficient tool for assessing climate impacts across a variety of scenarios. We demonstrate the capabilities of RIME-X by emulating a diverse set of regionally aggregated climate and climate impact variables available from ISIMIP3 and beyond for the NGFS (Network for Greening the Financial System) climate scenarios.

Escalating impacts of climate change underscore the risk of crossing thresholds of socio-ecological systems and adaptation limits. However, limitations in the provision of actionable climate information may hinder an adequate response. Here we suggest a reversal of the traditional impact chain methodology as an end-user focused approach to link local climate risks to emission pathways. We outline the socio-economic and value judgment dimensions that can inform identification of such local risk thresholds and adaptation limits. We apply this approach to heat-mortality risks for the city of Berlin. To limit the likely maximum increase in the occurrence of heat days with expected health impacts to less than 50% compared to today, the remaining global carbon budget in 2020 is 700 Gt CO2. We argue that linking local risk threshold exceedance directly to global emission benchmarks can aid the understanding of the benefits of stringent emission reductions for societies and local decision-makers.

Global warming levels are politically relevant targets, and therefore, in public discussion and in climate science, these global warming levels are often taken as a reference for climate states. While the focus on global warming levels is a useful simplification in many cases, it becomes misleading when looking at temperature overshoot (or stabilization) scenarios. In temperature overshoot scenarios, greenhouse gas concentrations are eventually reduced leading to a decrease in global mean temperatures. In such scenarios, lagged effects, feedback mechanisms, and tipping points can result in considerably different climate states after the overshoot as compared to before at the same global warming level. Here we assess to what extent changes in regional climate signals are reversed in the period after peak warming when global mean temperature decreases. We analyze a multi-model ensemble of CMIP6 simulations of two overshoot scenarios, SSP5-34-OS and SSP119. In many regions, climate signals are decoupled from global mean temperatures in the decades after peak warming, leading to differences in regional climate signals between before and after the overshoot at the same global warming level. More dedicated climate simulations of overshoot scenarios would be required to better evaluate how long the influence of the overshoot would affect regional climate signals and to better understand the mechanisms behind these changes. The presented overview of regional climate signals in overshoot scenarios until 2100 already suggests that considerable implications of temperature overshoots for climate impacts are to be expected and that these implications need to be considered for adaptation planning and policy making.

Heatwaves are becoming more frequent because of climate change, and this trend is exacerbated in cities due to the urban heat island effect. With more than half of the world’s population living in cities, it is essential to quantify the future evolution of heat stress and develop smart adaptation strategies to counter its impacts. This requires the capturing of fine-grained variations in heat-related hazards within the urban fabric. However, the coarse resolutions of Earth System Models makes it difficult to model urban areas explicitly. Moreover, high-resolution modelling of future climate conditions in cities is often conducted for select cities only, in very focused studies or by private companies, thus limiting the availability of its results in the public domain. Additionally, there is limited understanding of the potential of climate-smart urban development for reducing heat stress.In the H2020 PROVIDE project, we use the urban boundary layer climate model UrbClim to generate projections of urban heat stress at a 100-meter resolution, for about 20 indicators in 140 urban centres across the world. UrbClim consists of a land surface scheme with simplified urban physics coupled to a 3-D atmospheric boundary layer module, and can represent the effect of varying land cover conditions on local climate. We consider three emission scenarios: 1) compatible with the 1.5°C goal of the Paris Agreement, 2) representative of the trend from current policies, and 3) an intermediary scenario. The forcing data corresponding to these scenarios is generated by coupling the emulator for Global Mean Temperature FaIR, with the Earth System Model emulator with spatially explicit representation MESMER. This allows us to account for uncertainties in the forcing data arising from both the response of Global Mean Temperature (GMT) to emissions, and the response of large-scale climate conditions above each city included in the study to rising GMT.The resulting database is integrated into the PROVIDE climate risk dashboard, an open-access and user-friendly online tool that allows visualization of global-to-local future climate impacts depending on mitigation outcomes. The dashboard also contains a module that allows its users to first select a critical heat stress level of their choice, and then get information about the emission scenarios that would enable to avoid exceeding that level in their city of interest. This more impact-centered perspective on the UrbClim results provides information on future heat stress in a way that better reflect how climate impact information is accounted for in local adaptation processes.Furthermore, we explore the potential for urban greening plans co-developed by urban planners and city-level stakeholders to reduce heat stress by running UrbClim at very high resolution (down to 1 meter) for the cities of Lisbon (Portugal), Bodø (Norway), Islamabad (Pakistan), and Berlin (Germany). These new results will eventually also be made available in the PROVIDE climate risk dashboard. Together with the insights from the urban planners and stakeholders’ needs, they offer more practical and policy-relevant insights for adaptation practitioners at the municipal level on the potential for climate-smart urban development to reduce heat stress.

Anthropogenic climate change encompasses shifts in weather and climate patterns that result in more severe extreme weather events such as tropical storms and heat waves. Observations and climate model simulations show that compound heat waves are becoming more frequent and intense with increasing global mean temperatures. Nevertheless, appropriate local and actionable climate information is scarce and may hinder an adequate adaptation response. Here, we use a reversal of the traditional impact chain methodology to find emissions constraints that avoid severe heat waves in Islamabad, Pakistan. We use high-resolution urban climate simulations from UrbClim, global climate simulations from CMIP6, and climate simulations from the simple climate model FaIR to estimate local risk threshold exceedances for a large set of emission scenarios. By doing so, we can link specific levels of local climatic impact-drivers to global climate trajectories and assess emission constraints that would avoid severe heat events in Islamabad. Connecting local risk threshold exceedance to global emission benchmarks can clarify the benefits of reduced emissions for society and decision-makers. Furthermore, our modeling framework allows to investigate different combinations of heat thresholds with occurrence frequencies and can easily be used to answer specific questions from various stakeholders.

Emulators of Earth system models (ESMs) are statistical models that approximate selected outputs of ESMs. Owing to their runtime efficiency, emulators are especially useful when large amounts of data are required, for example, for in-depth exploration of the emission space, for investigating high-impact low-probability events, or for estimating uncertainties and variability. This paper introduces an emulation framework that allows us to emulate gridded monthly mean precipitation fields using gridded monthly mean temperature fields as forcing. The emulator is designed as an extension of the Modular Earth System Model Emulator (MESMER) framework, and its core relies on the concepts of generalised linear models (GLMs). Precipitation at each (land) grid point and for each month is approximated as a multiplicative model with two factors. The first factor entails the temperature-driven precipitation response and is assumed to follow a gamma distribution with a logarithmic link function. The second factor is the residual variability in the precipitation field, which is assumed to be independent of temperature but may still possess spatial precipitation correlations. Therefore, the monthly residual field is decomposed into independent principal components and subsequently approximated and sampled using a kernel density estimation with a Gaussian kernel. The emulation framework is tested and validated using 24 ESMs from the sixth phase of the Coupled Model Intercomparison Project (CMIP6). For each ESM, we train on a single-ensemble member across scenarios and evaluate the emulator performance using simulations with historical and Shared Socioeconomic Pathways (SSP5-8.5) forcing. We show that the framework captures grid-point-specific precipitation characteristics, such as variability, trend, and temporal auto-correlations. In addition, we find that emulated spatial (cross-variable) characteristics are consistent with those of ESMs. The framework is also able to capture compound hot-dry and cold-wet extremes, although it systematically underestimates their occurrence probabilities. The emulation of spatially explicit coherent monthly temperature and precipitation time series is a major step towards a computationally efficient representation of impact-relevant variables of the climate system.

Without stringent reductions in emission of greenhouse gases in the coming years, an exceedance of the 1.5C temperature limit would become increasingly likely. This has given rise to so-called temperature overshoot scenarios, in which the global mean surface air temperature increase above pre-industrial levels exceeds a certain limit, i.e. 1.5C, before bringing temperatures back below that level. Despite their prominence in the climate mitigation literature, the implications of an overshoot for local climate impacts is still understudied. Here we present a comprehensive analysis of implications of an overshoot for regional temperature and precipitation changes as well as climate extremes indices. Based on a multi-model comparison from the Coupled Model Intercomparison Project (CMIP6) we find that temperature changes are largely reversible in many regions, but also report significant land-ocean and latitudinal differences after an overshoot. For precipitation, the emerging picture is less clear. In many regions the drying or wetting trend is continued throughout the overshoot irrespective of a change in the global mean temperature trend with resulting consequences for extreme precipitation. Taken together, our results indicate that even under a reversal of global mean temperature increase, regional climate changes may only be partially reversed in the decades after peak warming. We thus provide further evidence that overshooting of a warming level implies considerable risks on the regional level.

Heatwaves are becoming more frequent because of climate change, and this trend is exacerbated in cities due to the urban heat island effect. With more than half of the world’s population living in cities, it is essential to quantify the future evolution of heat stress and develop smart adaptation strategies to counter its impacts. This requires the capturing of fine-grained variations in heat-related hazards within the urban fabric. However, the coarse resolutions of Earth System Models makes it difficult to model urban areas explicitly. Moreover, high-resolution modelling of future climate conditions in cities is often conducted for select cities only, in very focused studies or by private companies, thus limiting the availability of its results in the public domain. Additionally, there is limited understanding of the potential of climate-smart urban development for reducing heat stress. In the H2020 PROVIDE project, we use the urban boundary layer climate model UrbClim to generate projections of urban heat stress at a 100-meter resolution, for about 20 indicators in 140 urban centres across the world. UrbClim consists of a land surface scheme with simplified urban physics coupled to a 3-D atmospheric boundary layer module, and can represent the effect of varying land cover conditions on local climate. We consider three emission scenarios: 1) compatible with the 1.5°C goal of the Paris Agreement, 2) representative of the trend from current policies, and 3) an intermediary scenario. The forcing data corresponding to these scenarios is generated by coupling the emulator for Global Mean Temperature FaIR, with the Earth System Model emulator with spatially explicit representation MESMER. This allows us to account for uncertainties in the forcing data arising from both the response of Global Mean Temperature (GMT) to emissions, and the response of large-scale climate conditions above each city included in the study to rising GMT. The resulting database is integrated into the PROVIDE climate risk dashboard, an open-access and user-friendly online tool that allows visualization of global-to-local future climate impacts depending on mitigation outcomes. The dashboard also contains a module that allows its users to first select a critical heat stress level of their choice, and then get information about the emission scenarios that would enable to avoid exceeding that level in their city of interest. This more impact-centered perspective on the UrbClim results provides information on future heat stress in a way that better reflect how climate impact information is accounted for in local adaptation processes. Furthermore, we explore the potential for urban greening plans co-developed by urban planners and city-level stakeholders to reduce heat stress by running UrbClim at very high resolution (down to 1 meter) for the cities of Lisbon (Portugal), Bodø (Norway), Islamabad (Pakistan), and Berlin (Germany). These new results will eventually also be made available in the PROVIDE climate risk dashboard. Together with the insights from the urban planners and stakeholders’ needs, they offer more practical and policy-relevant insights for adaptation practitioners at the municipal level on the potential for climate-smart urban development to reduce heat stress.

Global emission reduction efforts continue to be insufficient to meet the temperature goal of the Paris Agreement1. This makes the systematic exploration of so-called overshoot pathways that temporarily exceed a targeted global warming limit before drawing temperatures back down to safer levels a priority for science and policy2-5. Here we show that global and regional climate change and associated risks after an overshoot are different from a world that avoids it. We find that achieving declining global temperatures can limit long-term climate risks compared with a mere stabilization of global warming, including for sea-level rise and cryosphere changes. However, the possibility that global warming could be reversed many decades into the future might be of limited relevance for adaptation planning today. Temperature reversal could be undercut by strong Earth-system feedbacks resulting in high near-term and continuous long-term warming6,7. To hedge and protect against high-risk outcomes, we identify the geophysical need for a preventive carbon dioxide removal capacity of several hundred gigatonnes. Yet, technical, economic and sustainability considerations may limit the realization of carbon dioxide removal deployment at such scales8,9. Therefore, we cannot be confident that temperature decline after overshoot is achievable within the timescales expected today. Only rapid near-term emission reductions are effective in reducing climate risks.

Changing climatic conditions threaten forest ecosystems. Drought, disease and infestation, are leading to forest die-offs which cause substantial economic and ecological losses. In central Europe, this is especially relevant for commercially important coniferous tree species. This study uses climate envelope exceedance (CEE) to approximate species risk under different future climate scenarios. To achieve this, we used current species presence-absence and historical climate data, coupled with future climate scenarios from various Earth System Models. Climate scenarios tended towards drier and warmer conditions, causing strong CEEs especially for spruce. However, we show that annual averages of temperature and precipitation obscure climate extremes. Including climate extremes reveals a broader increase in CEEs across all tree species. Our study shows that the consideration of climate extremes, which cannot be adequately reflected in annual averages, leads to a different assessment of the risk of forests and thus the options for adapting to climate change.

With ongoing greenhouse gas emissions it becomes increasingly unlikely that the global mean temperature (GMT) can be stabilized at 1.5°C without considerable negative emissions. As a result, most emission scenarios that would allow to reach 1.5°C GMT at the end of the century are overshoot scenarios: In these scenarios GMT warms until net-zero emissions are reached and slowly starts to cool afterwards. Here we want to have a closer look at the local climate responses after peak warming to get a first idea of potential consequences of overshoots.The analysis is mainly based on the overshoot scenarios SSP119 and the SSP534-over from the “Coupled Model Intercomparison Project (Phase 6)”. We identify regions in which precipitation or temperature has an asymmetric response to GMT changes around peak warming. In some regions, and especially for temperature related variables, the asymmetries could result from lagged responses in the climate system. However, there are also a number of dynamic mechanisms that could influence local climate signals after peak warming and there are only few regions where analyzed earth system models (ESM) agree on the sign of change.In many regions, the projected trends in precipitation or temperature after peak warming are in the range of trends that can be found in control runs without anthropogenic forcings. Here, single model initial-condition large ensembles (SMILEs) are necessary to estimate the forced response in overshoot scenarios. For a comprehensive understanding of the mechanisms explaining these non-linear responses to GMT changes around peak warming, more large ensemble simulations of idealized overshoot scenarios for different ESMs would be required.

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.

Climate impacts have been studied intensively and our understanding of changes in climate impacts due to anthropogenic activity is impressive (see IPCC AR6). There is, however, a gap between the physical understanding of changes in climate impacts and availability of information that could directly be used by adaptation planners. We argue that this gap is to a large extent a result of the usual modeling chain that is based on a handful of representative emission scenarios.Most climate change studies take a small, predefined set of emission scenarios (SSP2-45, SSP1-26, SSP5-85 etc.) and calculate the global and regional climate impacts resulting from these. Focusing on a limited set of emission scenarios allows us to compare results from different modeling groups and lets us run detailed climate models on each scenario. However, this modeling approach does not align with relevant research questions such as: “How much can be emitted to avoid a certain impact?” Or “what are the emission constraints to limit the probability of experiencing a certain event until 2050 to 10%?”The presented reversal of the impact chain would help to answer these questions. The idea is to start from a clearly defined impact and evolve uncertainties backwards into the emission space. Doing so, we take the perspective of practitioners who know very well what impacts are of relevance and would like to know how these impacts are related to greenhouse gas emissions.