Towards constraining climate sensitivity by linear analysis of feedback patterns in thousands of perturbed-physics GCM simulations

A linear analysis is applied to a multi-thousand member “perturbed physics" GCM ensemble to identify the dominant physical processes responsible for variation in climate sensitivity across the ensemble. Model simulations are provided by the distributed computing project, climate prediction.net . A principal component analysis of model radiative response reveals two dominant independent feedback processes, each largely controlled by a single parameter change. The leading EOF was well correlated with the value of the entrainment coefficient—a parameter in the model’s atmospheric convection scheme. Reducing this parameter increases high vertical level moisture causing an enhanced clear sky greenhouse effect both in the control simulation and in the response to greenhouse gas forcing. This effect is compensated by an increase in reflected solar radiation from low level cloud upon warming. A set of ‘secondary’ cloud formation parameters partly modulate the degree of shortwave compensation from low cloud formation. The second EOF was correlated with the scaling of ice fall speed in clouds which affects the extent of cloud cover in the control simulation. The most prominent feature in the EOF was an increase in longwave cloud forcing. The two leading EOFs account for 70% of the ensemble variance in λ—the global feedback parameter. Linear predictors of feedback strength from model climatology are applied to observational datasets to estimate real world values of the overall climate feedback parameter. The predictors are found using correlations across the ensemble. Differences between predictions are largely due to the differences in observational estimates for top of atmosphere shortwave fluxes. Our validation does not rule out all the strong tropical convective feedbacks leading to a large climate sensitivity.     

Radiative forcing and response of a GCM to ice age boundary conditions: cloud feedback and climate sensitivity

 A general circulation model is used to examine the effects of reduced atmospheric CO₂, insolation changes and an updated reconstruction of the continental ice sheets at the Last Glacial Maximum (LGM). A set of experiments is performed to estimate the radiative forcing from each of the boundary conditions. These calculations are used to estimate a total radiative forcing for the climate of the LGM. The response of the general circulation model to the forcing from each of the changed boundary conditions is then investigated. About two-thirds of the simulated glacial cooling is due to the presence of the continental ice sheets. The effect of the cloud feedback is substantially modified where there are large changes to surface albedo. Finally, the climate sensitivity is estimated based on the global mean LGM radiative forcing and temperature response, and is compared to the climate sensitivity calculated from equilibrium experiments with atmospheric CO₂ doubled from present day concentration. The calculations here using the model and palaeodata support a climate sensitivity of about 1 Wm^-2 K^-1 which is within the conventional range. Received: 8 February 1997 / Accepted: 4 June 1997     

Relative contribution of feedback processes to Arctic amplification of temperature change in MIROC GCM

The finding that surface warming over the Arctic exceeds that over the rest of the world under global warming is a robust feature among general circulation models (GCMs). While various mechanisms have been proposed, quantifying their relative contributions is an important task in order to understand model behavior. Here we apply a recently proposed feedback analysis technique to an atmosphere–ocean GCM under two and four times CO₂ concentrations which approximately lead to seasonally and annually sea ice-free climates. The contribution of feedbacks to Arctic temperature change is investigated. The surface warming in the Arctic is contributed by albedo, water vapour and large-scale condensation feedbacks and reduced by the evaporative cooling feedback. The surface warming contrast between the Arctic and the global averages (AA) is maintained by albedo and evaporative cooling feedbacks. The latter contributes to AA predominantly by cooling the low latitudes more than the Arctic. Latent heat transport into the Arctic increases and hence evaporative cooling plus large-scale condensation feedback contributes positively to AA. On the other hand, dry-static energy transport into the Arctic decreases and hence dynamical heating feedback contributes negatively to AA. An important contribution is thus made via changes in hydrological cycle and not via the ‘dry’ heat transport process. A larger response near the surface than aloft in the Arctic is maintained by the albedo, water vapour, and dynamical heating feedbacks, in which the albedo and water vapour feedbacks contribute through warming the surface more than aloft, and the dynamical heating feedback contributes by cooling aloft more than the surface. In our experiments, ocean and sea ice dynamics play a secondary role. It is shown that a different level of CO₂ increase introduces a latitudinal and seasonal difference into the feedbacks.     

Climate sensitivity and cloud feedback processes imposed by two different external forcings in an aquaplanet GCM

The sensitivity of climate to an increase in sea surface temperature (SST) and CO₂, as well as cloud feedback processes, is analyzed through a series of aquaplanet experiments listed in the Coupled Model Intercomparison Project. Rainfall is strengthened in a +4K anomaly SST experiment due to the enhanced surface evaporation; while in a quadruple CO₂ experiment, precipitation and total cloud cover are appreciably weakened. In both the +4K and quadruple CO₂ (4xCO₂) experiments, the Hadley cell is impaired, with an increase in moderate subsidence and a decrease in the frequency of strong convective activity. Regarding cloud radiation forcing (CRF), the analysis technique of Bony et al. (Climate Dynamics, 22:71–86, 2004) is used to sort cloud variables by dynamic regimes using the 500-hPa vertical velocity in tropical areas (30°S–30°N). Results show that the tropically averaged CRF change is negative and is dominated mainly by the thermodynamic component. Within convective regimes, the behavior of longwave CRF is different in the +4K and 4xCO₂ experiments, with positive and negative changes, respectively. The globally averaged CRF also reveals a negative change in both aquaplanet and Earthlike experiments, implying that clouds may play a role in decelerating global warming. The calculated climate sensitivity demonstrates that our results are close to those obtained from other models, with 0.384 and 0.584?Km²?W^?1 for aquaplanet and Earthlike experiments, respectively.     

On the generation of climate model ensembles

Climate model ensembles are used to estimate uncertainty in future projections, typically by interpreting the ensemble distribution for a particular variable probabilistically. There are, however, different ways to produce climate model ensembles that yield different results, and therefore different probabilities for a future change in a variable. Perhaps equally importantly, there are different approaches to interpreting the ensemble distribution that lead to different conclusions. Here we use a reduced-resolution climate system model to compare three common ways to generate ensembles: initial conditions perturbation, physical parameter perturbation, and structural changes. Despite these three approaches conceptually representing very different categories of uncertainty within a modelling system, when comparing simulations to observations of surface air temperature they can be very difficult to separate. Using the twentieth century CMIP5 ensemble for comparison, we show that initial conditions ensembles, in theory representing internal variability, significantly underestimate observed variance. Structural ensembles, perhaps less surprisingly, exhibit over-dispersion in simulated variance. We argue that future climate model ensembles may need to include parameter or structural perturbation members in addition to perturbed initial conditions members to ensure that they sample uncertainty due to internal variability more completely. We note that where ensembles are over- or under-dispersive, such as for the CMIP5 ensemble, estimates of uncertainty need to be treated with care.     

The impact of new land surface physics on the GCM simulation of climate and climate sensitivity

 Recent improvements to the Hadley Centre climate model include the introduction of a new land surface scheme called “MOSES” (Met Office Surface Exchange Scheme). MOSES is built on the previous scheme, but incorporates in addition an interactive plant photosynthesis and conductance module, and a new soil thermodynamics scheme which simulates the freezing and melting of soil water, and takes account of the dependence of soil thermal characteristics on the frozen and unfrozen components. The impact of these new features is demonstrated by comparing 1×CO₂ and 2×CO₂ climate simulations carried out using the old (UKMO) and new (MOSES) land surface schemes. MOSES is found to improve the simulation of current climate. Soil water freezing tends to warm the high-latitude land in the northern Hemisphere during autumn and winter, whilst the increased soil water availability in MOSES alleviates a spurious summer drying in the mid-latitudes. The interactive canopy conductance responds directly to CO₂, supressing transpiration as the concentration increases and producing a significant enhancement of the warming due to the radiative effects of CO₂ alone. Received: 16 March 1998 / Accepted: 4 August 1998     

Origins of differences in climate sensitivity, forcing and feedback in climate models

We diagnose climate feedback parameters and CO₂ forcing including rapid adjustment in twelve atmosphere/mixed-layer-ocean (“slab”) climate models from the CMIP3/CFMIP-1 project (the AR4 ensemble) and fifteen parameter-perturbed versions of the HadSM3 slab model (the PPE). In both ensembles, differences in climate feedbacks can account for approximately twice as much of the range in climate sensitivity as differences in CO₂ forcing. In the AR4 ensemble, cloud effects can explain the full range of climate sensitivities, and cloud feedback components contribute four times as much as cloud components of CO₂ forcing to the range. Non-cloud feedbacks are required to fully account for the high sensitivities of some models however. The largest contribution to the high sensitivity of HadGEM1 is from a high latitude clear-sky shortwave feedback, and clear-sky longwave feedbacks contribute substantially to the highest sensitivity members of the PPE. Differences in low latitude ocean regions (30°N/S) contribute more to the range than those in mid-latitude oceans (30–55°N/S), low/mid latitude land (55°N/S) or high latitude ocean/land (55–90°N/S), but contributions from these other regions are required to account fully for the higher model sensitivities, for example from land areas in IPSL CM4. Net cloud feedback components over the low latitude oceans sorted into percentile ranges of lower tropospheric stability (LTS) show largest differences among models in stable regions, mainly due to their shortwave components, most of which are positive in spite of increasing LTS. Differences in the mid-stability range are smaller, but cover a larger area, contributing a comparable amount to the range in climate sensitivity. These are strongly anti-correlated with changes in subsidence. Cloud components of CO₂ forcing also show the largest differences in stable regions, and are strongly anticorrelated with changes in estimated inversion strength (EIS). This is qualitatively consistent with what would be expected from observed relationships between EIS and low-level cloud fraction. We identify a number of cases where individual models show unusually strong forcings and feedbacks compared to other members of the ensemble. We encourage modelling groups to investigate unusual model behaviours further with sensitivity experiments. Most of the models fail to correctly reproduce the observed relationships between stability and cloud radiative effect in the subtropics, indicating that there remains considerable room for model improvements in the future.     

CO2 climate sensitivity and snow-sea-ice albedo parameterization in an atmospheric GCM coupled to a mixed-layer ocean model

The snow-sea-ice albedo parameterization in an atmospheric general circulation model (GCM), coupled to a simple mixed-layer ocean and run with an annual cycle of solar forcing, is altered from a version of the same model described by Washington and Meehl (1984). The model with the revised formulation is run to equilibrium for 1 × CO₂ and 2 × CO₂ experiments. The 1 ×CO₂ (control) simulation produces a global mean climate about 1° warmer than the original version, and sea-ice extent is reduced. The model with the altered parameterization displays heightened sensitivity in the global means, but the geographical patterns of climate change due to increased carbon dioxide (CO₂) are qualitatively similar. The magnitude of the climate change is affected, not only in areas directly influenced by snow and ice changes but also in other regions of the globe, including the tropics where sea-surface temperature, evaporation, and precipitation over the oceans are greater. With the less-sensitive formulation, the global mean surface air temperature increase is 3.5 °C, and the increase of global mean precipitation is 7.12%. The revised formulation produces a globally averaged surface air temperature increase of 4.04 °C and a precipitation increase of 7.25%, as well as greater warming of the upper tropical troposphere. Sensitivity of surface hydrology is qualitatively similar between the two cases with the larger-magnitude changes in the revised snow and ice-albedo scheme experiment. Variability of surface air temperature in the model is comparable to observations in most areas except at high latitudes during winter. In those regions, temporal variation of the sea-ice margin and fluctuations of snow cover dependent on the snow-ice-albedo formulation contribute to larger-than-observed temperature variability. This study highlights an uncertainty associated with results from current climate GCMs that use highly parameterized snow-sea-ice albedo schemes with simple mixed-layer ocean models.The National Center for Atmospheric Research is sponsored by the National Science Foundation.     

Empirical Estimates of Global Climate Sensitivity： An Assessment of Strategies Using a Coupled GCM

A control integration with the normal solar constant and one with it increased by 2.5% in the National Center for Atmospheric Research （NCAR） coupled atmosphere-ocean Climate System Model were conducted to see how well the actual realized global warming could be predicted just by analysis of the control results. This is a test, within a model context, of proposals that have been advanced to use knowledge of the present day climate to make ＂empirical＂ estimates of global climate sensitivity. The scaling of the top-of-the-atmosphere infrared flux and the planetary albedo as functions of surface temperature was inferred by examining four different temporal and geographical variations of the control simulations. Each of these inferences greatly overestimates the climate sensitivity of the model, largely because of the behavior of the cloud albedo. In each inference the control results suggest that cloudiness and albedo decrease with increasing surface temperature. However, the experiment with the increased solar constant actually has higher albedo and more cloudiness at most latitudes. The increased albedo is a strong negative feedback, and this helps account for the rather weak sensitivity of the climate in the NCAR model. To the extent that these model results apply to the real world, they suggest empirical evaluation of the scaling of global-mean radiative properties with surface temperature in the present day climate provides little useful guidance for estimates of the actual climate sensitivity to global changes.     

IPCC AR6对地球气候系统中反馈机制的新认识

气候反馈反映了气候系统内部对外界干扰的适应过程,在很大程度上影响对未来气候变化的预估。本文对政府间气候变化专门委员会（IPCC）第六次评估报告（AR6）中有关气候反馈的内容进行了梳理。相比第五次评估报告（AR5）,AR6对云反馈的认识有了较大提高,尤其是副热带海洋上空低云的反馈。AR6认为在高信度上云反馈参数为正值,即对气候变化起到一种放大效应。不过,云反馈的不确定范围在所有反馈机制中依然是最大的。除了普朗克反馈外,其他反馈机制（包括水汽、温度直减率、地表反照率、云、生物地球物理和非CO₂生物地球化学反馈）均在正值区间或零附近,总体上对气候变化起到放大效应。AR6对总的气候反馈的估计值为-1.16 W·m^-2·℃^-1,5%～95%的置信区间为[-1.81,-0.51] W·m^-2·℃^-1。随着气候平均态的增暖,气候反馈参数很可能会更靠近正值。     

A GCM study of future climate response to aerosol pollution reductions

We use the global atmospheric GCM aerosol model ECHAM5-HAM to asses possible impacts of future air pollution mitigation strategies on climate. Air quality control strategies focus on the reduction of aerosol emissions. Here we investigate the extreme case of a maximum feasible end-of-pipe abatement of aerosols in the near term future (2030) in combination with increasing greenhouse gas (GHG) concentrations. The temperature response of increasing GHG concentrations and reduced aerosol emissions leads to a global annual mean equilibrium temperature response of 2.18 K. When aerosols are maximally abated only in the Industry and Powerplant sector, while other sectors stay with currently enforced regulations, the temperature response is 1.89 K. A maximum feasible abatement applied in the Domestic and Transport sector, while other sectors remain with the current legislation, leads to a temperature response of 1.39 K. Increasing GHG concentrations alone lead to a temperature response of 1.20 K. We also simulate 2–5% increases in global mean precipitation among all scenarios considered, and the hydrological sensitivity is found to be significantly higher for aerosols than for GHGs. Our study, thus highlights the huge potential impact of future air pollution mitigation strategies on climate and supports the need for urgent GHG emission reductions. GHG and aerosol forcings are not independent as both affect and are influenced by changes in the hydrological cycle. However, within the given range of changes in aerosol emissions and GHG concentrations considered in this study, the climate response towards increasing GHG concentrations and decreasing aerosols emissions is additive.     

Using second-order approximation to incorporate GCM uncertainty in climate change impact assessments

This study presents a method to incorporate uncertainty of climate variables in climate change impact assessments, where the uncertainty being considered refers to the divergence of general circulation model (GCM) projections. The framework assesses how much bias occurs when the uncertainties of climate variables are ignored. The proposed method is based on the second-order expansion of Taylor series, called second-order approximation (SOA). SOA addresses the bias which occurs by assuming the expected value of a function is equal to the function of the expected value of the predictors. This assumption is not valid for nonlinear systems, such as in the case of the relationship of climate variables to streamflow. To investigate the value of SOA in the climate change context, statistical downscaling models for monthly streamflow were set up for six hydrologic reference stations in Australia which cover contrasting hydro-climate regions. It is shown that in all locations SOA makes the largest difference for low flows and changes the overall mean flow by 1–3%. Another advantage of the SOA approach is that the individual contribution of each climate variable to the total difference can be estimated. It is found that geopotential height and specific humidity cause more bias than wind speeds in the downscaling models considered here.     

Coupling between subtropical cloud feedback and the local hydrological cycle in a climate model

In HadGEM2-A, AMIP experiments forced with observed sea surface temperatures respond to uniform and patterned +4 K SST perturbations with strong positive cloud feedbacks in the subtropical stratocumulus/trade cumulus transition regions. Over the subtropical Northeast Pacific at 137°W/26°N, the boundary layer cloud fraction reduces considerably in the AMIP +4 K patterned SST experiment. The near-surface wind speed and the air-sea temperature difference reduces, while the near-surface relative humidity increases. These changes limit the local increase in surface evaporation to just 3 W/m² or 0.6 %/K. Previous studies have suggested that increases in surface evaporation may be required to maintain maritime boundary layer cloud in a warmer climate. This suggests that the supply of water vapour from surface evaporation may not be increasing enough to maintain the low level cloud fraction in the warmer climate in HadGEM2-A. Sensitivity tests which force the surface evaporation to increase substantially in the +4 K patterned SST experiment result in smaller changes in boundary layer cloud and a weaker cloud feedback in HadGEM2-A, supporting this idea. Although global mean surface evaporation in climate models increases robustly with global temperature (and the resulting increase in atmospheric radiative cooling), local values may increase much less, having a significant impact on cloud feedback. These results suggest a coupling between cloud feedback and the hydrological cycle via changes in the patterns of surface evaporation. A better understanding of both the factors controlling local changes in surface evaporation and the sensitivity of clouds to such changes may be required to understand the reasons for inter-model differences in subtropical cloud feedback.     

On the dependence of climate sensitivity on convective parametrization

Two sensitivity experiments, in which CO₂ is doubled and sea-surface temperatures are enhanced, were carried out using a general circulation model to determine the influence of the convective parametrization on simulated climate change. In the first experiment, a non-penetrative layer-swapping convection scheme is used; in the second, a penetrative scheme is used. It is found that the penetrative scheme gives the greater upper tropospheric warming (over 4.5 K compared to 4 K) and the greater reduction in upper tropospheric cloud, consistent with recent CO₂ sensitivity studies. However, there is a 0.7 Wm^–2 greater increase in net downward radiation at the top of the atmosphere in the experiment with the non-penetrative scheme, implying a larger tropical warming which is inconsistent with recent CO₂ studies. Other possible explanations for discrepancies between recent studies of the equilibrium climate response to increasing CO₂ are considered and discussed. The changes in the atmospheric fluxes of heat and moisture from the tropical continents in the model with the penetrative scheme differ from those found using the non-penetrative scheme, and those in an equilibrium experiment using the penetrative scheme. Thus, changes in circulation may explain the apparent discrepancy in the current experiments, but prescribed sea-surface temperature experiments may not provide a reliable indication of a model's equilibrium climate sensitivity. Offprint requests to: JFB Mitchell     

The albedo of temperate and boreal forest and the Northern Hemisphere climate: a sensitivity experiment using the LMD GCM

A deforestation experiment is performed using the Laboratoire de Meteorologie Dynamique Atmospheric General Circulation Model (LMD GCM) to determine the climatic role of the largest vegetation formation in the Northern Hemisphere, localized mostly north of latitude 45°N, which is called the temperate and boreal forest. For this purpose, an iterative albedo scheme based on vegetation type, snow age, snowfall rate and area of snow cover, is developed for snow-covered surfaces. The results show a cooling of Northern Hemisphere soil and an increase in the snow cover when the forest is removed, as found by previous similar experiments.In our study this cooling is related to different causes, depending on the season. It is linked to modifications in the soil radiative properties, like surface albedo, due to the disappearance of forest, and consequently, to a greater exposure of the snow-covered soil underneath. It is also related to alterations in the hydrological cycle, observed mainly in summer and autumn at middle latitudes. The model shows a strong sensitivity to the coupled surface albedo — soil temperature — fractional snow cover response in the spring. A later and longer snowmelt season is also detected.This study adds to our understanding of climatic variation on longer time scales, since it is widely accepted that the formation and disappearance of different vegetation formations is closely related to climatic evolution patterns, in particular on the time scale of the glacial oscillations.     

Incorporating circulation statistics in bias correction of GCM ensembles: hydrological application for the Rhine basin

An adapted statistical bias correction method is introduced to incorporate circulation-dependence of the model precipitation bias, and its influence on estimated discharges for the Rhine basin is analyzed for a historical period. The bias correction method is tailored to time scales relevant to flooding events in the basin. Large-scale circulation patterns (CPs) are obtained through Maximum Covariance Analysis using reanalysis sea level pressure and high-resolution precipitation observations. A bias correction using these CPs is applied to winter and summer separately, acknowledging the seasonal variability of the circulation regimes in North Europe and their correlation with regional precipitation rates over the Rhine basin. Two different climate model ensemble outputs are explored: ESSENCE and CMIP5. The results of the CP-method are then compared to observations and uncorrected model outputs. Results from a simple bias correction based on a delta factor (NoCP-method) are also used for comparison. For both summer and winter, the CP-method offers a statistically significant improvement of precipitation statistics for subsets of data dominated by particular circulation regimes, demonstrating the circulation-dependence of the precipitation bias. Uncorrected, CP and NoCP corrected model outputs were used as forcing to a hydrological model to simulate river discharges. The CP-method leads to a larger improvement in simulated discharge in the Alpine area in winter than in summer due to a stronger dependence of Rhine precipitation on atmospheric circulation in winter. However, the NoCP-method, in comparison to the CP-method, improves the discharge estimations over the entire Rhine basin.     

Carbon cycle–climate feedback sensitivity to parameter changes of a zero-dimensional terrestrial carbon cycle scheme in a climate model of intermediate complexity

Summary A series of sensitivity runs have been performed with a coupled climate–carbon cycle model. The climatic component consists of the climate model of intermediate complexity IAP RAS CM. The carbon cycle component is formulated as a simple zero-dimensional model. Its terrestrial part includes gross photosynthesis, and plant and soil respirations, depending on temperature via Q ₁₀-relationships (Lenton, 2000). Oceanic uptake of anthropogenic carbon is formulated is a bi-linear function of tendencies of atmospheric concentration of CO₂ and globally averaged annual mean sea surface temperature. The model is forced by the historical industrial and land use emissions of carbon dioxide for the second half of the 19th and the whole of the 20th centuries, and by the emission scenario SRES A2 for the 21st century. For the standard set of the governing parameters, the model realistically captures the main features of the Earth’s observed carbon cycle. A large number of simulations have been performed, perturbing the governing parameters of the terrestrial carbon cycle model. In addition, the climate part is perturbed, either by zeroing or artificially increasing the climate model sensitivity to the doubling of the atmospheric CO₂ concentration. Performing the above mentioned perturbations, it is possible to mimic most of the range found in the C4MIP simulations. In this way, a wide range of the climate–carbon cycle feedback strengths is obtained, differing even in the sign of the feedback. If the performed simulations are subjected to the constraints of a maximum allowed deviation of the simulated atmospheric CO₂ concentration (pCO_2(a)) from the observed values and correspondence between simulated and observed terrestrial uptakes, it is possible to narrow the corresponding uncertainty range. Among these constraints, considering pCO_2(a) and uptakes are both important. However, the terrestrial uptakes constrain the simulations more effectively than the oceanic ones. These constraints, while useful, are still unable to rule out both extremely strong positive and modest negative climate–carbon cycle feedback.     

On the observational determination of climate sensitivity and its implications

We estimate climate sensitivity from observations, using the deseasonalized fluctuations in sea surface temperatures (SSTs) and the concurrent fluctuations in the top-of-atmosphere (TOA) outgoing radiation from the ERBE (1985–1999) and CERES (2000–2008) satellite instruments. Distinct periods of warming and cooling in the SSTs were used to evaluate feedbacks. An earlier study (Lindzen and Choi, 2009) was subject to significant criticisms. The present paper is an expansion of the earlier paper where the various criticisms are taken into account. The present analysis accounts for the 72 day precession period for the ERBE satellite in a more appropriate manner than in the earlier paper. We develop a method to distinguish noise in the outgoing radiation as well as radiation changes that are forcing SST changes from those radiation changes that constitute feedbacks to changes in SST. We demonstrate that our new method does moderately well in distinguishing positive from negative feedbacks and in quantifying negative feedbacks. In contrast, we show that simple regression methods used by several existing papers generally exaggerate positive feedbacks and even show positive feedbacks when actual feedbacks are negative. We argue that feedbacks are largely concentrated in the tropics, and the tropical feedbacks can be adjusted to account for their impact on the globe as a whole. Indeed, we show that including all CERES data (not just from the tropics) leads to results similar to what are obtained for the tropics alone — though with more noise. We again find that the outgoing radiation resulting from SST fluctuations exceeds the zerofeedback response thus implying negative feedback. In contrast to this, the calculated TOA outgoing radiation fluxes from 11 atmospheric models forced by the observed SST are less than the zerofeedback response, consistent with the positive feedbacks that characterize these models. The results imply that the models are exaggerating climate sensitivity.     

On the estimation of systematic error in regression-based predictions of climate sensitivity

An extension of a regression-based methodology for constraining climate forecasts using a multi-thousand member ensemble of perturbed climate models is presented, using the multi-model CMIP-3 ensemble to estimate the systematic model uncertainty in the prediction, with the caveat that systematic biases common to all models are not accounted for. It is shown that previous methodologies for estimating the systematic uncertainty in predictions of climate sensitivity are dependent on arbitrary choices relating to ensemble sampling strategy. Using a constrained regression approach, a multivariate predictor may be derived based upon the mean climatic state of each ensemble member, but components of this predictor are excluded if they cannot be validated within the CMIP-3 ensemble. It is found that the application of the CMIP-3 constraint serves to decrease the upper bound of likelihood for climate sensitivity when compared with previous studies, with 10th and 90th percentiles of probability at 1.5 K and 4.3 K respectively.