Search

The cloud feedback may result in amplification or damping of Arctic warming. Two common techniques used to diagnose the top‐of‐the‐atmosphere cloud feedback are the Adjusted Cloud Radiative Effect (AdjCRE) method and the Cloud Radiative Kernel (CRK) method. We apply both to CMIP5 and CMIP6 model data, finding that the AdjCRE calculated Arctic shortwave cloud feedback is twice as correlated with sea ice loss in CMIP5, and four times in CMIP6, as the CRK method. We find that the CRK method produces Arctic all‐sky residual percentages exceeding 20% in 15 of 18 models. We use the CRK method to decompose the feedback in CMIP5 and CMIP6 finding that its median value changed from negative to positive driven by a less‐negative cloud optical depth feedback. Despite its lack of closure, we conclude that the CRK method is better suited for Arctic SW feedbacks as it is less impacted by surface albedo changes. Plain Language Summary: The cloud feedback is the process by which cloud property changes in a warming climate can either further enhance warming or damp it. The Arctic is warming faster than the rest of the globe, and one of the largest sources of uncertainty in its climate projections is the cloud feedback. There are two popular methods to calculate the cloud feedback: the Adjusted Cloud Radiative Effect technique, and the Cloud Radiative Kernel technique. In this paper we compare the two methods in a suite of climate models by considering the extent to which changes in Arctic sea ice impact the cloud feedbacks. From this analysis we conclude that the Cloud Radiative Kernel method is less affected by sea ice loss. We then apply the Cloud Radiative Kernel technique to data from the two most recent generations of global climate models to investigate how polar day Arctic cloud feedbacks have changed between these generations. We find that the median value of these Arctic feedbacks is slightly positive in the newest generation of models, a change from slightly negative in the previous generation that is largely fueled by a weakening of the feedback associated with changes in cloud optical depth. Key Points: The Cloud Radiative Kernel method is less sensitive to surface albedo changes than the Adjusted Cloud Radiative Effect techniqueThe Cloud Radiative Kernel method provides poor radiative closure in a suite of global climate modelsThe median shortwave Arctic cloud feedback in recent climate models is slightly positive due to a weakened cloud optical depth feedback [ABSTRACT FROM AUTHOR]

Representing mixing state of black carbon (BC) is challenging for global climate models (GCMs). The Community Atmosphere Model version 6 (CAM6) with the four‐mode version of the Modal Aerosol Module (MAM4) represents aerosols as fully internal mixtures with uniform composition within each aerosol mode, resulting in high degree of internal mixing of BC with non‐BC species and large mass ratio of coating to BC (RBC, the mass ratio of non‐BC species to BC in BC‐containing particles). To improve BC mixing state representation, we coupled a machine learning (ML) model of BC mixing state index trained on particle‐resolved simulations to the CAM6 with MAM4 (MAM4‐ML). In MAM4‐ML, we use RBC to partition accumulation mode particles into two new modes, BC‐free particles and BC‐containing particles. We adjust RBC to make the modeled BC mixing state index (χmode) match the one predicted by the ML model (χML). On a global average, the mass fraction of BC‐containing particles in accumulation mode decreases from 100% (MAM4‐default) to 48% (MAM4‐ML). The globally averaged χmode decreases from 78% (MAM4‐default) to 63% (MAM4‐ML, 19% reduction) and agrees well with χML (66%). The RBC decreases by 52% for accumulation mode and better agrees with observations. The hygroscopicity drops by 9% for BC‐containing particles in accumulation mode, leading to a 20% reduction in the BC activation fraction. The surface BC concentration increases most (6.9%) in the Arctic, and the BC burden increases by 4%, globally. Our study highlights the application of the ML model for improving key aerosol processes in GCMs. Plain Language Summary: Black carbon (BC) mixing state is an important property for assessing impacts of BC on human health and climate. It describes how BC is distributed among the particle population and within individual particles. Due to computational costs, global climate models (GCMs) consider aerosols as either external mixtures (e.g., bulk models) or fully internal mixtures with uniform composition within a certain size range (e.g., an aerosol bin in sectional models or an aerosol mode in modal models). To improve BC mixing state representation, we coupled a machine learning (ML) model learned from particle‐resolved simulations to the online simulations of the Community Atmosphere Model version 6 (CAM6) with the four‐mode version of the Modal Aerosol Module (MAM4‐ML). We used the mass ratio of coating to BC (RBC) to determine the mass fraction of BC‐free particles and BC‐containing particles, and we adjusted RBC to make the modeled BC mixing state match ML model. The MAM4‐ML model reduces the degree of internal mixture of BC substantially. In addition, MAM4‐ML RBC is reduced by 52%, leading to a 20% reduction in BC activation fraction and a 4% increase in BC burden. This study proposes a method for the application of the ML to improve GCMs. Key Points: Machine learning model trained on particle‐resolved simulations has been coupled to CAM6 MAM4 to improve BC mixing state representationThe BC mixing state index is reduced by 19%, and the mass ratio of coating to BC is reduced by 52%, in better agreement with observationsDue to the drop in hygroscopicity of BC‐containing aerosol (9%), BC activation fraction decreased by 20%, and BC burden increased by 4% [ABSTRACT FROM AUTHOR]

Despite multiple advances in the understanding of the water cycle intensification in a warmer climate, climate models still diverge in their hydrological projections. Here we constrain annual runoff projections over individual and aggregated Arctic river basins. For this purpose, we use two ensembles of global climate models and two statistical methods: a regression scheme assuming similar runoff sensitivities at interannual versus climate change timescales, and a Bayesian method where models are used to derive a posterior runoff response conditioned on historical observations. While both techniques are shown to narrow model uncertainties, more or less substantially depending on rivers, the Bayesian method is less sensitive to the choice of the model ensemble and is more skillful when tested with synthetic observations. It has also been applied over the whole Arctic watershed, showing so far a limited narrowing of the inter‐model spread, but its skill will further improve with increasing climate change. Plain Language Summary: Despite considerable progress in understanding the intensification of the water cycle in response to global warming, projections of changes in the various components of the continental water balance remain highly model‐dependent. This is particularly the case for the evolution of runoff, which remains very sensitive to the way atmospheric and continental processes are represented. Here we evaluate and compare two statistical methods to constrain climate projections of annual runoff at high latitudes via available or synthetic observations. Our Bayesian method, based on historical observations, is found to be more robust than a previous method based on observed interannual climate variability. It underlines the importance of having reliable flow reconstructions to constrain runoff projections and of using empirical observational constraints with caution, as they are likely to lead to over‐confident projections. Key Points: Runoff sensitivity to temperature depends on the considered timescaleObservational constraints based on interannual variability can thus lead to overconfident projectionsObservational constraints based on historical trends can be effective if such trends are driven by human emissions of greenhouse gasses [ABSTRACT FROM AUTHOR]

Employing a model democracy in which each model is equally weighted may lead to a poor estimation of the true uncertainty in climate projections from phase 6 of the Coupled Model Intercomparison Project (CMIP6). The improvement and increase in number of CMIP6 models compared with previous phases of CMIP indicate that both model skill and independence need to be considered to provide convincing projections. In this study, we use a weighting scheme, which weights both the skill and independence of multi‐model simulations, to efficiently constrain the large uncertainty in the projection of the timing of an ice‐free Arctic. The uncertainty‐constrained projections of CMIP6 show that the multi‐model spread of the projected first year of an ice‐free Arctic can be reduced by about 29 years under the Shared Socioeconomic Pathway (SSP)3–7.0 scenario (a regional rivalry scenario), indicating a faster tendency to an ice‐free Arctic summer than projections based solely on model democracy. A fossil‐fuel‐based development scenario (i.e., SSP5‐8.5) leads to an ice‐free Arctic before the 2070s (ranging from 2038 to 2071), while an ice‐free Arctic occurs slightly later (by ∼10 years) under the SSP2‐4.5 scenario (i.e., intermediate scenario) and the SSP3‐7.0 scenario, but is inevitable this century. The sustainable development scenario (i.e., SSP1‐2.6) is likely to prevent the occurrence of an ice‐free Arctic. Internal variability strongly affects the projection estimated by the equally weighted ensemble; however, it has a negligible impact on the results obtained by the weighting scheme, thereby indicating that the results of this study are robust and convincing. Plain Language Summary: Climate models are a key tool used by scientists to estimate climate changes in the future. However, the outputs of such models carry inherent uncertainties that can affect the confidence in projections. These uncertainties may not be appropriately estimated by simple and traditional methods that do not adequately consider the model skill or independence (i.e., similarity to other models). In this study, we found that the uncertainties in Arctic sea‐ice projections can be largely reduced if the outputs of climate models are weighted according to their performance and independence. By doing so, we can increase the level of confidence in estimating the year when the Arctic is likely to see its first ice‐free summer (sea‐ice extent below 1 × 106 km2). It is indicated that a sustainable development scenario may avoid an ice‐free Arctic. In contrast, a fossil‐fuel‐based development scenario will lead to an ice‐free Arctic before the 2070s (ranging from 2038 to 2071); and an ice‐free Arctic will occur slightly later (by ∼10 years) under an intermediate or regional rivalry scenario, but is still inevitable this century. Key Points: The performance and interdependence of climate models may lead to a poor estimation of the true uncertainty in climate projectionsA weighting scheme considering both model skill and independence can reduce the spread in the first year of an ice‐free Arctic by ∼29 yearsThe weighted spread indicates that the first year of an ice‐free Arctic is likely to occur during 2040–2072 under the SSP3‐7.0 scenario [ABSTRACT FROM AUTHOR]

While the annual mean Arctic Amplification (AA) index varied between two and three during the 1970–2000 period, it reached values exceeding four during the first two decades of the 21st century. The AA did not change in a continuous fashion but rather in two sharp increases around 1986 and 1999. During those steps the mean global surface air temperature trend remained almost constant, while the Arctic trend increased. Although the "best" CMIP6 models reproduce the increasing trend of the AA in 1980s they do not capture the sharply increasing trend of the AA after 1999 including its rapid step‐like increase. We propose that the first sharp AA increase around 1986 is due to external forcing, while the second step close to 1999 is due to internal climate variability, which models cannot reproduce in the observed time. Plain Language Summary: The rate of Arctic warming varied in time, while the rate of global warming was nearly constant over the period 1960–2020. This led to a variable Arctic Amplification (AA) defined as a ratio of the Arctic warming trend to mean global warming trend. In addition, the Arctic experienced sudden changes in the rate of warming with step‐like changes in annual mean Arctic Amplification around the years 1986 and 1999. Climate models do not capture the second observed sharp increase of the AA. We suggest that the first AA increase around 1986 is primarily due to external forcing (increasing concentration of carbon dioxide) while the second sharp increase around 1999 is dominated by internal climate variability which current models cannot capture accurately within the time scale of the physical phenomena. Key Points: Annual mean Arctic Amplification (AA) within the period 1970–2020 changed in steep steps around 1986 and 1999. It reached values over 4.0Even those CMIP6 models best at reproducing the AA did not reproduce the second sudden increase near 1999We conjecture that the first AA sharp increase is due to external forcing, while the second is due to internal climate variability [ABSTRACT FROM AUTHOR]

Climate models are crucial for assessing climate variability and change. A reliable model for future climate should reasonably simulate the historical climate. Here, we assess the performance of CMIP6 models in reproducing statistical properties of observed annual maxima of daily precipitation. We go beyond the commonly used methods and assess CMIP6 simulations on three scales by performing: (a) univariate comparison based on L‐moments and relative difference measures; (b) bivariate comparison using Kernel densities of mean and L‐variation, and of L‐skewness and L‐kurtosis, and (c) comparison of the entire distribution function using the Generalized Extreme Value (GEV) distribution coupled with a novel application of the Anderson‐Darling Goodness‐of‐fit test. The results reveal that the statistical shape properties (related to the frequency and magnitude of extremes) of CMIP6 simulations match well with the observational datasets. The simulated mean and variation differ among the models with 70% of simulations having a difference within ±10% from the observations. Biases are observed in the bivariate investigation of mean and variation. Several models perform well with the HadGEM3‐GC31‐MM model performing well in all three scales when compared to the ground‐based Global Precipitation Climatology Centre data. Finally, the study highlights biases of CMIP6 models in simulating extreme precipitation in the Arctic, Tropics, arid and semi‐arid regions. Plain Language Summary: Annual maxima of daily precipitation are widely used to design critical infrastructures such as dams and stormwater networks. Climate change is expected to increase the frequency and intensity of extreme events. Climate model projections offer a glimpse into the future and can help assess potential changes and impacts. It is reasonable to assume that climate models simulating accurately the historical climate may also simulate well the future. Here, we assessed the latest generation of climate models, that is the CMIP6 models, to reproduce the historical annual maximum daily precipitation. We compared simulations and observations of extreme precipitation using advanced summary statistics, novel probability similarity measures, and robust statistical tests. The results indicate that models, in general, reproduce well the behavior of annual maximum precipitation, but biases exist especially in the simultaneous behavior of mean and variance. Shortcomings of CMIP6 models are highlighted in the Arctic, Tropics, arid, and semi‐arid regions. Key Points: Assessment of CMIP6 biases in precipitation extremes using L‐moments, novel probability similarity measures and statistical testsCMIP6 models reproduce fairly well observed annual maximum precipitation but biases exist in the joint behavior of mean and variabilityShortcomings of CMIP6 models are highlighted in the Arctic, Tropics, arid, and semi‐arid regions [ABSTRACT FROM AUTHOR]

State‐of‐the‐art climate models simulate a large spread in the projected decline of Arctic sea‐ice area (SIA) over the 21st century. Here we diagnose causes of this intermodel spread using a simple model that approximates future SIA based on present SIA and the sensitivity of SIA to Arctic temperatures. This model accounts for 70%–95% of the intermodel variance, with the majority of the spread arising from present‐day biases. The remaining spread arises from intermodel differences in Arctic warming, with some contribution from differences in the local sea‐ice sensitivity. Using observations to constrain the projections moves the probability of an ice‐free Arctic forward by 10–35 years when compared to unconstrained projections. Under a high‐emissions scenario, an ice‐free Arctic will likely (>66% probability) occur between 2036 and 2056 in September and between 2050 and 2068 from July to October. Under a medium‐emissions scenario, the "likely" date occurs between 2040 and 2062 in September and much later in the 21st century from July to October. Plain Language Summary: Arctic sea ice coverage has declined substantially over the past few decades and is projected to continue to decline over the next century. These projections, however, are marred by large uncertainties which arise primarily due to differences between climate models. In this study, we use a simple model that emulates the future evolution of Arctic sea ice as simulated by climate models to explain where this uncertainty comes from. We show that biases in simulating present‐day Arctic sea ice contribute most of the uncertainty, with differences in the simulated amount of Arctic warming from climate models contributing much of the rest. We use observations to constrain our simple model and show that under a high emissions scenario it is likely the Arctic will be free of sea ice in September sometime between 2036 and 2056 and from July to October sometime between 2050 and 2068. We also show that the emissions pathway impacts the length of ice free summers in the Arctic, indicating a low‐emissions pathway will reduce the likelihood of seeing ice‐free Arctic summers. Key Points: A model relating future sea‐ice area (SIA) to present SIA and local sea‐ice sensitivity is used to explain the intermodel spread in Arctic SIA projectionsBiases in simulating present‐day SIA contribute most to the spread, with model differences in Arctic warming contributing the restUnder a high‐emissions scenario, the Arctic will likely be ice‐free in September between 2036 and 2056 and from July to October between 2050 and 2068 [ABSTRACT FROM AUTHOR]

Given uncertainty in the processes involved in polar amplification, elucidating the role of poleward heat and moisture transport is crucial. The Polar Amplification Model Intercomparison Project (PAMIP) permits robust separation of the effects of sea ice loss from sea surface warming under climate change. We utilize a moist isentropic circulation framework that accounts for moisture transport, condensation, and eddy transport, in order to analyze the circulation connecting the mid‐latitudes and the Arctic. In PAMIP's atmospheric general circulation model experiments, prescribed sea ice loss reduces poleward heat transport (PHT) by warming the returning moist isentropic circulation at high latitudes, while prescribed warming of the ocean surface increases PHT by strengthening the moist isentropic circulation. Inter‐model spread of PHT into the Arctic reflects the tug‐of‐war between sea‐ice and surface‐warming effects. Plain Language Summary: A major conundrum in current climate science is to understand what Arctic changes imply for the climate and environment in mid‐latitude regions. The Polar Amplification Model Intercomparison Project (PAMIP) designed a set of climate model experiments to specifically answer this question in a carefully designed, idealized framework. PAMIP's approach is to separate historic and projected climate change into parts associated with Arctic sea ice loss and ocean surface warming and investigate how these two contributions can influence the atmosphere. To isolate these effects, only models can be used, because, in reality, sea ice loss and ocean surface warming are strongly linked together. This letter focuses on what the PAMIP experiments imply for the transport and redistribution of heat and moisture in the atmosphere. In PAMIP, we learn that Arctic sea ice loss causes the atmosphere to reduce the transport of dry and cold air away from the Arctic while ocean warming causes more transport of moist warm air toward the Arctic. These two effects are in a tug‐of‐war, suggesting that climate change can cause a mix of competing impacts on the global energy transport from the tropics to the Arctic. Key Points: Sea ice loss exports static energy from the Arctic by warming the equatorward branch of the global mass circulationSea surface warming imports static energy into the Arctic by increasing the mid‐latitude mass transport, overwhelming the sea ice effectThere is more spread amongst models for the effect of sea‐surface warming than for sea ice loss [ABSTRACT FROM AUTHOR]