Reducing Uncertainties in Climate Projections with Emergent Constraints:Concepts,Examples and Prospects

Models disagree on a significant number of responses to climate change,such as climate feedback,regional changes,or the strength of equilibrium climate sensitivity.Emergent constraints aim to reduce these uncertainties by finding links between the inter-model spread in an observable predictor and climate projections.In this paper,the concepts underlying this framework are recalled with an emphasis on the statistical inference used for narrowing uncertainties,and a review of emergent constraints found in the last two decades.Potential links between highlighted predictors are explored,especially those targeting uncertainty reductions in climate sensitivity,cloud feedback,and changes of the hydrological cycle.Yet the disagreement across emergent constraints suggests that the spread in climate sensitivity can not be significantly narrowed.This calls for weighting the realism of emergent constraints by quantifying the level of physical understanding explaining the relationship.This would also permit more efficient model evaluation and better targeted model development.In the context of the upcoming CMIP6 model intercomparison a growing number of new predictors and uncertainty reductions is expected,which call for robust statistical inferences that allow cross-validation of more likely estimates.     

Climate model dependence and the replicate Earth paradigm

Multi-model ensembles are commonly used in climate prediction to create a set of independent estimates, and so better gauge the likelihood of particular outcomes and better quantify prediction uncertainty. Yet researchers share literature, datasets and model code—to what extent do different simulations constitute independent estimates? What is the relationship between model performance and independence? We show that error correlation provides a natural empirical basis for defining model dependence and derive a weighting strategy that accounts for dependence in experiments where the multi-model mean would otherwise be used. We introduce the “replicate Earth” ensemble interpretation framework, based on theoretically derived statistical relationships between ensembles of perfect models (replicate Earths) and observations. We transform an ensemble of (imperfect) climate projections into an ensemble whose mean and variance have the same statistical relationship to observations as an ensemble of replicate Earths. The approach can be used with multi-model ensembles that have varying numbers of simulations from different models, accounting for model dependence. We use HadCRUT3 data and the CMIP3 models to show that in out of sample tests, the transformed ensemble has an ensemble mean with significantly lower error and much flatter rank frequency histograms than the original ensemble.     

Projections of daily mean temperature variability in the future: cross-validation tests with ENSEMBLES regional climate simulations

Because of model biases, projections of future climate need to combine model simulations of recent and future climate with information on observed climate. Here, 10 methods for projecting the distribution of daily mean temperatures are compared, using six regional climate change simulations for Europe. Cross validation between the models is used to assess the potential performance of the methods in projecting future climate. Delta change and bias correction type methods show similar cross-validation performance, with methods based on the quantile mapping approach doing best in both groups due to their apparent ability to reduce the errors in the projected time mean temperature change. However, as no single method performs best under all circumstances, the optimal approach might be to use several well-behaving methods in parallel. When applying the various methods to real-world temperature projection for the late 21st century, the largest intermethod differences are found in the tails of the temperature distribution. Although the intermethod variation of the projections is generally smaller than their intermodel variation, it is not negligible. Therefore, it should be preferably included in uncertainty analysis of temperature projections, particularly in applications where the extremes of the distribution are important.     

IPCC第六次气候变化评估中的气候约束预估方法

得益于第五次评估报告（AR5）以来约束预估研究的迅速发展,观测约束成为政府间气候变化专门委员会（IPCC）第一工作组（WGI）第六次评估报告（AR6）提升对未来预估约束的证据链中的重要一环。IPCC第一工作组第六次评估报告首次利用包括根据历史模拟温度升高幅度得到的观测约束、多模式预估以及第六次评估报告中更新的气候敏感度在内的多条证据链来约束全球地表温度未来变化的预估,减小了多模式预估的不确定性。文中回顾并介绍了IPCC第一工作组第六次评估报告中涉及的几种主要观测约束方法（多模式加权方法、基于归因结论的约束方法（ASK方法）、萌现约束方法）及其应用情况。在IPCC第一工作组第六次评估报告以及很多针对不同区域不同变量的预估研究中,观测约束方法均显示出了订正模式误差、改善模式预估的潜力。相比而言,目前中国在观测约束预估方面的研究还不多,亟待加强观测约束方法研究以及在中国区域气候变化预估中的应用,为中国应对气候变化的政策制定和适应规划提供更丰富、不确定性更小的未来气候信息。      

Climate model errors, feedbacks and forcings: a comparison of perturbed physics and multi-model ensembles

Ensembles of climate model simulations are required for input into probabilistic assessments of the risk of future climate change in which uncertainties are quantified. Here we document and compare aspects of climate model ensembles from the multi-model archive and from perturbed physics ensembles generated using the third version of the Hadley Centre climate model (HadCM3). Model-error characteristics derived from time-averaged two-dimensional fields of observed climate variables indicate that the perturbed physics approach is capable of sampling a relatively wide range of different mean climate states, consistent with simple estimates of observational uncertainty and comparable to the range of mean states sampled by the multi-model ensemble. The perturbed physics approach is also capable of sampling a relatively wide range of climate forcings and climate feedbacks under enhanced levels of greenhouse gases, again comparable with the multi-model ensemble. By examining correlations between global time-averaged measures of model error and global measures of climate change feedback strengths, we conclude that there are no simple emergent relationships between climate model errors and the magnitude of future global temperature change. Algorithms for quantifying uncertainty require the use of complex multivariate metrics for constraining projections.     

On the added value of regional climate modeling in climate change assessment

Regional climate models represent a promising tool to assess the regional dimension of future climate change and are widely used in climate impact research. While the added value of regional climate models has been highlighted with respect to a better representation of land-surface interactions and atmospheric processes, it is still unclear whether radiative heating implies predictability down to the typical scale of a regional climate model. As a quantitative assessment, we apply an optimal statistical filter to compare the coherence between observed and simulated patterns of Mediterranean climate change from a global and a regional climate model. It is found that the regional climate model has indeed an added value in the detection of regional climate change, contrary to former assumptions. The optimal filter may also serve as a weighting factor in multi-model averaging.     

Characterizing climate change risks by linking robust decision frameworks and uncertain probabilistic projections

There is increasing concern that avoiding climate change impacts will require proactive adaptation, particularly for infrastructure systems with long lifespans. However, one challenge in adaptation is the uncertainty surrounding climate change projections generated by general circulation models (GCMs). This uncertainty has been addressed in different ways. For example, some researchers use ensembles of GCMs to generate probabilistic climate change projections, but these projections can be highly sensitive to assumptions about model independence and weighting schemes. Because of these issues, others argue that robustness-based approaches to climate adaptation are more appropriate, since they do not rely on a precise probabilistic representation of uncertainty. In this research, we present a new approach for characterizing climate change risks that leverages robust decision frameworks and probabilistic GCM ensembles. The scenario discovery process is used to search across a multi-dimensional space and identify climate scenarios most associated with system failure, and a Bayesian statistical model informed by GCM projections is then developed to estimate the probability of those scenarios. This provides an important advancement in that it can incorporate decision-relevant climate variables beyond mean temperature and precipitation and account for uncertainty in probabilistic estimates in a straightforward way. We also suggest several advancements building on prior approaches to Bayesian modeling of climate change projections to make them more broadly applicable. We demonstrate the methodology using proposed water resources infrastructure in Lake Tana, Ethiopia, where GCM disagreement on changes in future rainfall presents a major challenge for infrastructure planning.     

Evaluation of delta change and bias correction methods for future daily precipitation: intermodel cross-validation using ENSEMBLES simulations

Due to inherent limitations in climate models, their output is biased in relation to observed climate and as such does not provide reliable climate projections. In this study, nine methods used to account for biases in daily precipitation are tested. First, cross-validation tests were made using a set of ENSEMBLES regional model simulations to gain insights in the potential performance of the methods in the future climate. The results show that quantile mapping type methods, being able to modify the shape of the precipitation distribution, often outperform other types of methods. Yet, as the performance depends on time of the year, location and part of the distribution considered, it is not possible to distinguish one universally best performing method. In addition, the improvement relative to the projections that would have been obtained assuming unchanged climate is relatively modest, particularly in the early twentyfirst century conditions. Further tests with different method combinations show that the projections could be potentially improved by using several well performing methods in parallel. In the second part of the study, contributions of method and model differences to the overall variation of precipitation projections are assessed. It is shown that although intermodel differences play an important role, uncertainties related to intermethod differences are substantial, particularly in the tails of the distribution. This suggests that method uncertainty should be taken into account when constructing daily precipitation projections, possibly by using several methods in parallel.     

Climate change signal and uncertainty in projections of ocean wave heights

In this study, projections of seasonal means and extremes of ocean wave heights were made using projections of sea level pressure fields conducted with three global climate models for three forcing-scenarios. For each forcing-scenario, the three climate models’ projections were combined to estimate the multi-model mean projection of climate change. The relative importance of the variability in the projected wave heights that is due to the forcing prescribed in a forcing-scenario was assessed on the basis of ensemble simulations conducted with the Canadian coupled climate model CGCM2. The uncertainties in the projections of wave heights that are due to differences among the climate models and/or among the forcing-scenarios were characterized. The results show that the multi-model mean projection of climate change has patterns similar to those derived from using the CGCM2 projections alone, but the magnitudes of changes are generally smaller in the boreal oceans but larger in the region nearby the Antarctic coastal zone. The forcing-induced variance (as simulated by CGCM2) was identified to be of substantial magnitude in some areas in all seasons. The uncertainty due to differences among the forcing-scenarios is much smaller than that due to differences among the climate models, although it was identified to be statistically significant in most areas of the oceans (this indicates that different forcing conditions do make notable differences in the wave height climate change projection). The sum of the model and forcing-scenario uncertainties is smaller in the JFM and AMJ seasons than in other seasons, and it is generally small in the mid-high latitudes and large in the tropics. In particular, some areas in the northern oceans were projected to have large changes by all the three climate models.     

Does the model regional bias affect the projected regional climate change? An analysis of global model projections

An analysis is presented of the dependence of the regional temperature and precipitation change signal on systematic regional biases in global climate change projections. The CMIP3 multi-model ensemble is analyzed over 26 land regions and for the A1B greenhouse gas emission scenario. For temperature, the model regional bias has a negligible effect on the projected regional change. For precipitation, a significant correlation between change and bias is found in about 30% of the seasonal/regional cases analyzed, covering a wide range of different climate regimes. For these cases, a performance-based selection of models in producing climate change scenarios can affect the resulting change estimate, and it is noted that a minimum of four to five models is needed to obtain robust precipitation change estimates. In a number of cases, models with largely different precipitation biases can still produce changes of consistent sign. Overall, it is assessed that in the present generation of models the regional bias does not appear to be a dominant factor in determining the simulated regional change in the majority of cases.     

Shortcomings in climate model simulations of the ENSO-Atlantic hurricane teleconnection

A number of recent studies have used model projections to investigate how the North Atlantic environment in which tropical storms develop, as well as hurricane activity itself, might change in a warming world. However, accurate projection of the North Atlantic environment in the future requires, at a minimum, accurate representation of its mean state and variability in the current climate. Here we examine one metric of Atlantic basin tropical cyclone variability—its well-documented association with the El Ni?o-Southern Oscillation (ENSO)—in reanalyses and Intergovernmental Panel of Climate Change (IPCC) 4th Assessment Report (AR4) twentieth century and Atmospheric Model Intercomparison Project simulations. We find that no individual model provides consistently good representation of ENSO-related variability in the North Atlantic for variables relevant to hurricane activity (e.g. vertical wind shear, genesis potential). Model representation of the ENSO influence is biased due to both inaccurate representation of ENSO itself and inaccurate representation of the response to ENSO within the North Atlantic. Among variables examined, ENSO impacts on vertical wind shear and potential intensity were most poorly simulated. The multi-model ensemble mean representation of North Atlantic environmental response to ENSO is better matched with reanalysis than most individual AR4 models; however, this mean response still possesses some considerable bias. A few models do provide comparable or slightly better simulation of these ENSO-North Atlantic teleconnections than the multi-model ensemble average; however, for both the multi-model mean and the well performing models, good simulation of the ENSO-related variability of genesis potential within portions of the North Atlantic does not stem from accurate representation of the ENSO-related variability of the individual environmental variables that comprise genesis potential (e.g. vertical wind shear, potential intensity).     

Sensitivity of climate change projections to uncertainties in the estimates of observed changes in deep-ocean heat content

The MIT 2D climate model is used to make probabilistic projections for changes in global mean surface temperature and for thermosteric sea level rise under a variety of forcing scenarios. The uncertainties in climate sensitivity and rate of heat uptake by the deep ocean are quantified by using the probability distributions derived from observed twentieth century temperature changes. The impact on climate change projections of using the smallest and largest estimates of twentieth century deep ocean warming is explored. The impact is large in the case of global mean thermosteric sea level rise. In the MIT reference (“business as usual”) scenario the median rise by 2100 is 27 and 43 cm in the respective cases. The impact on increases in global mean surface air temperature is more modest, 4.9 and 3.9 C in the two respective cases, because of the correlation between climate sensitivity and ocean heat uptake required by twentieth century surface and upper air temperature changes. The results are also compared with the projections made by the IPCC AR4’s multi-model ensemble for several of the SRES scenarios. The multi-model projections are more consistent with the MIT projections based on the largest estimate of ocean warming. However, the range for the rate of heat uptake by the ocean suggested by the lowest estimate of ocean warming is more consistent with the range suggested by the twentieth century changes in surface and upper air temperatures, combined with the expert prior for climate sensitivity.     

A multi-model ensemble of downscaled spatial climate change scenarios for the Dommel catchment, Western Europe

Regional or local scale hydrological impact studies require high resolution climate change scenarios which should incorporate some assessment of uncertainties in future climate projections. This paper describes a method used to produce a multi-model ensemble of multivariate weather simulations including spatial–temporal rainfall scenarios and single-site temperature and potential evapotranspiration scenarios for hydrological impact assessment in the Dommel catchment (1,350 km²) in The Netherlands and Belgium. A multi-site stochastic rainfall model combined with a rainfall conditioned weather generator have been used for the first time with the change factor approach to downscale projections of change derived from eight Regional Climate Model (RCM) experiments for the SRES A2 emission scenario for the period 2071–2100. For winter, all downscaled scenarios show an increase in mean daily precipitation (catchment average change of +9% to +40%) and typically an increase in the proportion of wet days, while for summer a decrease in mean daily precipitation (−16% to −57%) and proportion of wet days is projected. The range of projected mean temperature is 7.7°C to 9.1°C for winter and 19.9°C to 23.3°C for summer, relative to means for the control period (1961–1990) of 3.8°C and 16.8°C, respectively. Mean annual potential evapotranspiration is projected to increase by between +17% and +36%. The magnitude and seasonal distribution of changes in the downscaled climate change projections are strongly influenced by the General Circulation Model (GCM) providing boundary conditions for the RCM experiments. Therefore, a multi-model ensemble of climate change scenarios based on different RCMs and GCMs provides more robust estimates of precipitation, temperature and evapotranspiration for hydrological impact assessments, at both regional and local scale.     

Gradient in the climate change signal of European discharge predicted by a multi-model ensemble

In order to perform hydrological studies on the PRUDENCE regional climate model (RCM) simulations, a special focus was put on the discharge from large river catchments located in northern and central Europe. The discharge was simulated with a simplified land surface (SL) scheme and the Hydrological Discharge (HD) model. The daily fields of precipitation, 2 m temperature and evapotranspiration from the RCM simulations were used as forcing. Therefore the total catchment water balances are constrained by the hydrological cycle of the different RCMs. The validation of the simulated hydrological cycle from the control simulations shows that the multi-model ensemble mean is closer to the observations than each of the models, especially if different catchments and hydrological variables are considered. Therefore, the multi-model ensemble mean can be used to largely reduce the uncertainty that is introduced by a single RCM. This also provides more confidence in the future projections for the multi-model ensemble means. The scenario simulations predict a gradient in the climate change signal over Northern and Central Europe. Common features are the overall warming and the general increase of evapotranspiration. But while in the northern parts the warming will enhance the hydrological cycle leading to an increased discharge, the large warming, especially in the summer, will slow down the hydrological cycle caused by a drying in the central parts of Europe which is accompanied by a reduction of discharge. The comparison of the changes predicted by the multi-model ensemble mean to the changes predicted by the driving GCM indicates that the RCMs can compensate problems that a driving GCM may have with local scale processes or parameterizations.     

The use of the land-sea warming contrast under climate change to improve impact metrics

A favoured method of assimilating information from state-of-the-art climate models into integrated assessment models of climate impacts is to use the transient climate response (TCR) of the climate models as an input, sometimes accompanied by a pattern matching approach to provide spatial information. More recent approaches to the problem use TCR with another independent piece of climate model output: the land-sea surface warming ratio (φ). In this paper we show why the use of φ in addition to TCR has such utility. Multiple linear regressions of surface temperature change onto TCR and φ in 22 climate models from the CMIP3 multi-model database show that the inclusion of φ explains a much greater fraction of the inter-model variance than using TCR alone. The improvement is particularly pronounced in North America and Eurasia in the boreal summer season, and in the Amazon all year round. The use of φ as the second metric is beneficial for three reasons: firstly it is uncorrelated with TCR in state-of-the-art climate models and can therefore be considered as an independent metric; secondly, because of its projected time-invariance, the magnitude of φ is better constrained than TCR in the immediate future; thirdly, the use of two variables is much simpler than approaches such as pattern scaling from climate models. Finally we show how using the latest estimates of φ from climate models with a mean value of 1.6—as opposed to previously reported values of 1.4—can significantly increase the mean time-integrated discounted damage projections in a state-of-the-art integrated assessment model by about 15 %. When compared to damages calculated without the inclusion of the land-sea warming ratio, this figure rises to 65 %, equivalent to almost 200 trillion dollars over 200 years.     

Evaluating the performance of CMIP3 and CMIP5 global climate models over the north-east Atlantic region

One of the main sources of uncertainty in estimating climate projections affected by global warming is the choice of the global climate model (GCM). The aim of this study is to evaluate the skill of GCMs from CMIP3 and CMIP5 databases in the north-east Atlantic Ocean region. It is well known that the seasonal and interannual variability of surface inland variables (e.g. precipitation and snow) and ocean variables (e.g. wave height and storm surge) are linked to the atmospheric circulation patterns. Thus, an automatic synoptic classification, based on weather types, has been used to assess whether GCMs are able to reproduce spatial patterns and climate variability. Three important factors have been analyzed: the skill of GCMs to reproduce the synoptic situations, the skill of GCMs to reproduce the historical inter-annual variability and the consistency of GCMs experiments during twenty-first century projections. The results of this analysis indicate that the most skilled GCMs in the study region are UKMO-HadGEM2, ECHAM5/MPI-OM and MIROC3.2(hires) for CMIP3 scenarios and ACCESS1.0, EC-EARTH, HadGEM2-CC, HadGEM2-ES and CMCC-CM for CMIP5 scenarios. These models are therefore recommended for the estimation of future regional multi-model projections of surface variables driven by the atmospheric circulation in the north-east Atlantic Ocean region.     

The role of boundary organizations in climate change adaptation from the perspective of municipal practitioners

In climate change impact research it is crucial to carefully select the meteorological input for impact models. We present a method for model selection that enables the user to shrink the ensemble to a few representative members, conserving the model spread and accounting for model similarity. This is done in three steps: First, using principal component analysis for a multitude of meteorological parameters, to find common patterns of climate change within the multi-model ensemble. Second, detecting model similarities with regard to these multivariate patterns using cluster analysis. And third, sampling models from each cluster, to generate a subset of representative simulations. We present an application based on the ENSEMBLES regional multi-model ensemble with the aim to provide input for a variety of climate impact studies. We find that the two most dominant patterns of climate change relate to temperature and humidity patterns. The ensemble can be reduced from 25 to 5 simulations while still maintaining its essential characteristics. Having such a representative subset of simulations reduces computational costs for climate impact modeling and enhances the quality of the ensemble at the same time, as it prevents double-counting of dependent simulations that would lead to biased statistics.     

Regional and seasonal intercomparison of CMIP3 and CMIP5 climate model ensembles for temperature and precipitation

Regional and seasonal temperature and precipitation over land are compared across two generations of global climate model ensembles, specifically, CMIP5 and CMIP3, through historical twentieth century skills and multi-model agreement, and twenty first century projections. A suite of diagnostic and performance metrics, ranging from spatial bias or model-consensus maps and aggregate time series plots, to measures of equivalence between probability density functions and Taylor diagrams, are used for the intercomparisons. Pairwise and multi-model ensemble comparisons were performed for 11 models, which were selected based on data availability and resolutions. Results suggest little change in the central tendency or variability or uncertainty of historical skills or consensus across the two generations of models. However, there are regions and seasons, at different levels of aggregation, where significant changes, performance improvements, and even degradation in skills, are suggested. The insights may provide directions for further improvements in next generations of climate models, and in the meantime, help inform adaptation and policy.     

Extended regional climate model projections for Europe until the mid-twentyfirst century: combining ENSEMBLES and CMIP3

This study aims at sharpening the existing knowledge of expected seasonal mean climate change and its uncertainty over Europe for the two key climate variables air temperature and precipitation amount until the mid-twentyfirst century. For this purpose, we assess and compensate the global climate model (GCM) sampling bias of the ENSEMBLES regional climate model (RCM) projections by combining them with the full set of the CMIP3 GCM ensemble. We first apply a cross-validation in order to assess the skill of different statistical data reconstruction methods in reproducing ensemble mean and standard deviation. We then select the most appropriate reconstruction method in order to fill the missing values of the ENSEMBLES simulation matrix and further extend the matrix by all available CMIP3 GCM simulations forced by the A1B emission scenario. Cross-validation identifies a randomized scaling approach as superior in reconstructing the ensemble spread. Errors in ensemble mean and standard deviation are mostly less than 0.1 K and 1.0 % for air temperature and precipitation amount, respectively. Reconstruction of the missing values reveals that expected seasonal mean climate change of the ENSEMBLES RCM projections is not significantly biased and that the associated uncertainty is not underestimated due to sampling of only a few driving GCMs. In contrast, the spread of the extended simulation matrix is partly significantly lower, sharpening our knowledge about future climate change over Europe by reducing uncertainty in some regions. Furthermore, this study gives substantial weight to recent climate change impact studies based on the ENSEMBLES projections, since it confirms the robustness of the climate forcing of these studies concerning GCM sampling.     

统计降尺度法对华北地区未来区域气温变化情景的预估

迄今为止,大部分海气耦合气候模式（AOGCM）的空间分辨率还较低,很难对区域尺度的气候变化情景做合理的预测。降尺度法已广泛用于弥补AOGCM在这方面的不足。作者采用统计降尺度方法对1月和7月华北地区49个气象观测站的未来月平均温度变化情景进行预估。采用的统计降尺度方法是主分量分析与逐步回归分析相结合的多元线性回归模型。首先,采用1961～2000年的 NCEP再分析资料和49个台站的观测资料建立月平均温度的统计降尺度模型,然后把建立的统计降尺度模型应用于HadCM3 SRES A2 和 B2 两种排放情景, 从而生成各个台站1950～2099年1月份和7月份温度变化情景。结果表明:在当前气候条件下,无论1月还是7月,统计降尺度方法模拟的温度与观测的温度有很好的一致性,而且在大多数台站,统计降尺度模拟气温与观测值相比略微偏低。对于未来气候情景的预估方面,无论1月还是7月,也无论是HadCM3 SRES A2 还是B2排放情景驱动统计模型,结果表明大多数的站点都存在温度的明显上升趋势,同时7月的上升趋势与1月相比偏低。