Search

The year 2014 broke the record for the warmest yearly average temperature in Europe. Attributing how much this was due to anthropogenic climate change and how much it was due to natural variability is a challenging question but one that is important to address. In this study, we compare four event attribution methods. We look at the risk ratio (RR) associated with anthropogenic climate change for this event, over the whole European region, as well as its spatial distribution. Each method shows a very strong anthropogenic influence on the event over Europe. However, the magnitude of the RR strongly depends on the definition of the event and the method used. Across Europe, attribution over larger regions tended to give greater RR values. This highlights a major source of sensitivity in attribution statements and the need to define the event to analyze on a case-by-case basis.

[1] This paper aims to test the robustness of the detection and attribution of anthropogenic climate change using four different empirical models that were previously developed to explain the observed global mean temperature changes over the last few decades. These studies postulated that the main drivers of these changes included not only the usual natural forcings, such as solar and volcanic, and anthropogenic forcings, such as greenhouse gases and sulfates, but also other known Earth system oscillations such as El NiQ no Southern Oscillation (ENSO) or the Atlantic Multidecadal Oscillation (AMO). In this paper, we consider these signals, or forced responses, and test whether or not the anthropogenic signal can be robustly detected under different assumptions for the internal variability of the climate system. We assume that the internal variability of the global mean surface temperature can be described by simple stochastic models that explore a wide range of plausible temporal autocorrelations, ranging from short memory processes exemplified by an AR(1) model to long memory processes, represented by a fractional differenced model. In all instances, we conclude that human-induced changes to atmospheric gas composition is affecting global mean surface temperature changes.

A number of studies have set out to obtain a range of atmosphere and ocean model behavior by perturbing parameters in a single climate model (perturbed physics ensemble: PPE). Early studies used shallow layer slab ocean or flux-adjusted coupled ocean-atmosphere models to obtain a broad range of behavior as characterized by climate sensitivity. A recent study reports a relatively narrow range of sensitivities (2.2–3.2°C) in a PPE of 35 coupled models without flux adjustment, raising the question whether previous broad ranges were an artifact of the use of models that were not in top-of-atmosphere (TOA) energy balance. Moreover, no PPE experiment has reported a large spread of behavior of the ocean compared to that exhibited in a multi-model ensemble (MME) such as Coupled Model Intercomparison Project phase 3 (CMIP3). In this work, we randomly perturb model parameters of a coupled ocean-atmosphere general circulation model using a space-filling design containing 10,000 combinations. The ensemble is run over the distributed computing platform of climateprediction.net under fixed pre-industrial forcing without flux adjustment. We resample a second, 20,000-member, ensemble with perturbations conditioned on the TOA fluxes from the first ensemble to not drift significantly from a realistic base state while targeting a range of behavior. Models within the targeted ensemble show realistic regional control climates when compared to the CMIP3 ensemble, although there is a bias in global mean surface temperature. The range of predicted equilibrium climate sensitivities of the targeted ensemble is substantially smaller than that obtained with flux adjustment, but larger than the range in the CMIP3 ensemble or in the 35-model un-flux-adjusted PPE in a recent study mentioned above. The Atlantic meridional overturning circulation in the targeted ensemble exhibits a spread in strength as wide as that found in the CMIP3 ensemble. We conclude that flux adjustment is not a pre-requisite for obtaining a broad spread of behavior in a perturbed physics ensemble.

[1] We perform a comparison of temperature changes in the free atmosphere over the period 1961–2010 using a high-top (lid at 84 km) and low-top (lid at 40 km) version of the HadGEM2 atmosphere-ocean global climate model. Model simulations of historical climate change that include key anthropogenic and natural external forcings are compared with three different radiosonde data sets. We also apply a regression-based “optimal fingerprinting” method to determine whether model-predicted temperature-change signals in response to external forcing are identifiable in observations. This method employs simulations that isolate the signals associated with different sets of external forcings (well-mixed GHGs and natural external factors). In both high- and low-top models, we obtain positive detection of the signals arising from anthropogenic influences. We find statistically significant differences between the latitude-height temperature signals simulated by the high- and low-top models, particularly in the tropical lower stratosphere. The scaling factors associated with each of the separate forcings are more tightly constrained in the high-top over the low-top model, but otherwise are similar in nature. Finally, we show that in detecting the GHG fingerprint, it is the entire vertical temperature profile that is important, not simply the lower-most tropospheric levels.

[1] To the extent that deficiencies in GCM simulations of precipitation are due to persistent errors of location and timing, correcting the spatial and seasonal distribution of features would provide a physically based improvement in inter-model agreement on future changes. We use a tool for the analysis of medical images to warp the precipitation climatologies of 14 General Circulation Models (GCMs) closer to a reanalysis of observations, rather than adjusting intensities locally as in conventional bias correction techniques. These warps are then applied to the same GCMs' simulated changes in mean climate under a CO2 quadrupling experiment. We find that the warping process not only makes GCMs' historical climatologies more closely resemble reanalysis but also reduces the disagreement between the models' response to this external forcing. Developing a tool that is tailored for the specific requirements of climate fields may provide further improvement, particularly in combination with local bias correction techniques.

Augmenting previous papers about the exceptional 2011-15 California drought, we offer new perspectives on the ‘snow drought’ that extended into Oregon in 2014 and Washington in 2015. Over 80% of measurement sites west of 115°W experienced record low snowpack in 2015, and we estimate a return period of 400-1000 years for California’s snowpack under the questionable assumption of stationarity. Hydrologic modeling supports the conclusion that 2015 was the most severe on record by a wide margin. Using a crowd-sourced superensemble of regional climate model simulations, we show that both human influence and sea surface temperature anomalies contributed strongly to the risk of snow drought in Oregon and Washington: the contribution of SST anomalies was about twice that of human influence. By contrast, SSTs and humans appear to have played a smaller role in creating California’s snow drought. In all three states, the anthropogenic effect on temperature exacerbated the snow drought.

Understanding how precipitation varies as the climate changes is essential to determining the true impact of global warming. This is a difficult task not only due to the large internal variability observed in precipitation but also because of a limited historical record and large biases in simulations of precipitation by general circulation models (GCMs). Here we make use of a technique that spatially and seasonally transforms GCM fields to reduce location biases and investigate the potential of this bias correction to study historical changes. We use two versions of this bias correction - one that conserves intensities and another that conserves integrated precipitation over transformed areas. Focussing on multimodel ensemble means, we find that both versions reduce RMS error in the historical trend by approximately 11% relative to the Global Precipitation Climatology Project (GPCP) data set. By regressing GCMs' historical simulations of precipitation onto radiative forcings, we decompose these simulations into anthropogenic and natural time series. We then perform a simple detection and attribution study to investigate the impact of reducing location biases on detectability. A multiple ordinary least squares regression of GPCP onto the anthropogenic and natural time series, with the assumptions made, finds anthropogenic detectability only when spatial corrections are applied. The result is the same regardless of which form of conservation is used and without reducing the dimensionality of the fields beyond taking zonal means. While "detectability" is dependent both on the exact methodology and the confidence required, this nevertheless demonstrates the potential benefits of correcting location biases in GCMs when studying historical precipitation, especially in cases where a signal was previously undetectable.

Tropical Instability Waves are a prominent feature in analyses of tropical Pacific sea surface temperature and sea level height variability, and may play a significant role in the heat, salt and momentum budgets of the equatorial oceans. Here we show that they may be strongly influenced by forcing from the atmosphere. In experiments with an ocean GCM we find that a phase shift in the subseasonal wind forcing gives rise to a corresponding phase shift in the TIWs, indicating a systematic influence of the high frequency wind forcing.

Previous workers have used the radiosonde record of atmospheric vertical temperature structure to detect anthropogenic influence on climate [e.g. Santer et al., 1996a]. However, none of these studies explicitly considered the influence of natural forcing factors. Whilst decadal changes of radiative solar activity are small, they could be amplified by various mechanisms (e.g. Haigh, 1996), potentially accounting for a significant fraction of the observed signal. Gillett et al. [2000] show that the diagnostics used to detect anthropogenic influence on atmospheric temperature structure are strongly influenced by solar forcing. Here we repeat the analysis of Allen and Tett, [1999] (AT99) including a model-simulated solar signal, to assess whether the observed tropospheric warming might be explained by solar forcing and the observed stratospheric cooling by the effects of ozone depletion. It transpires that previously reported results are robust to the inclusion of solar forcing since the greenhouse gas response is detectable even in the presence of solar forcing and stratospheric ozone depletion, allowing for an arbitrary amplitude of both solar and ozone signals.

In the last decade, satellite altimetric measurements of sea surface height (SSH) and infrared radiometric measurements of sea surface temperature (SST) have provided a wealth of information about ocean circulation and atmosphere-ocean interactions. SSH is a depth-integrated quantity dependent upon the temperature and salinity structure of the water column and on the depth independent barotropic contribution. SST from infrared radiometers is a surface parameter representing the temperature of the top few microns of the ocean surface. Hence any relationship between SST and SSH provides dynamical information about the coupling between the ocean surface and subsurface. It also offers a promise of new techniques such as interpolating SSH data using SST and of improved calculations of eddy kinetic energy. We use SST data from the along-track scanning radiometer on ERS-1 and SSH data from the TOPEX/POSEIDON instrument to examine the relationship between SST and SSH anomalies within the South Atlantic region for 1993 and 1994. We find that positive (=0.2-0.6) spatial cross correlations between SST and SSH anomalies at zero lag are present throughout the region at large scales (wavelengths >1000 km). Small-scale correlations, however, are high (=0.7) only in areas associated with fronts and mesoscale variability. These small-scale correlations are seasonal, being strongest in winter and weakest in summer. We discuss the application of these correlations to various techniques requiring the synergistic use of SSH and SST data.

Westward-propagating waves with periods of 20–30 days and wavelengths of ∼ 1,100km are a prominent feature of sea-surface temperatures (SSTs) in the equatorial Pacific and Atlantic Oceans. They have been attributed to instabilities due to current shear. We compare SST observations from the spaceborne Along Track Scanning Radiometer (ATSR) and TOGA-TAO moored buoys with SSTs from a model of the tropical Pacific forced with observed daily windstress data. The phases of the strongest “Tropical Instability Waves” (TIWs) in the model are in closer correspondence with those observed than we would expect if these waves simply developed from infinitesimal disturbances (in which case their phases would be arbitrary). If we filter out the intraseasonal component of the windstress, all phase-correspondence is lost. We conclude that the phases of these waves are not arbitrary, but partially determined by the intraseasonal winds. The subsurface evolution of the model suggests a possible control mechanism is through interaction with remotely-forced subsurface Kelvin and Rossby waves. This is supported by an experiment which shows how zonal wind bursts in the west Pacific can modify the TIW field, but other mechanisms, such as local feedbacks, are also possible.

[1] The Antarctic polar night jet has intensified during spring in recent decades due to stratospheric ozone depletion and rising greenhouse gas (GHG) concentrations and this has had substantial effects on the region's climate. GHG concentrations will rise over the 21st century whereas stratospheric ozone is expected to recover and there is uncertainty in future southern hemisphere (SH) circulation trends. We examine sensitivity to the physics parameterisation of the 21st century SH circulation projection of a coupled atmosphere-ocean General Circulation Model and the sensitivity of the contribution from stratospheric ozone recovery. Different model parameterizations give a greater range of future trends in the position of the tropospheric jet than has been found in previous multi-model comparisons. Ozone recovery causes a weakening and northward shift of the DJF tropospheric jet. Varying the physics parameterization affects the zonal wind response to ozone recovery of the SON stratosphere by ∼10% and that of the DJF troposphere by ∼25%. The projected future SAM index changes with and without ozone recovery and the SAM index response to ozone recovery alone are found to be strongly positively correlated with projected 21st century global warming.

[1] In the summer 2010 Western Russia was hit by an extraordinary heat wave, with the region experiencing by far the warmest July since records began. Whether and to what extent this event is attributable to anthropogenic climate change is controversial. Dole et al. (2011) report the 2010 Russian heat wave was “mainly natural in origin” whereas Rahmstorf and Coumou (2011) write that with a probability of 80% “the 2010 July heat record would not have occurred” without the large-scale climate warming since 1980, most of which has been attributed to the anthropogenic increase in greenhouse gas concentrations. The latter explicitly state that their results “contradict those of Dole et al. (2011).” Here we use the results from a large ensemble simulation experiment with an atmospheric general circulation model to show that there is no substantive contradiction between these two papers, in that the same event can be both mostly internally-generated in terms of magnitude and mostly externally-driven in terms of occurrence-probability. The difference in conclusion between these two papers illustrates the importance of specifying precisely what question is being asked in addressing the issue of attribution of individual weather events to external drivers of climate.

The first multi thousand member "perturbed physics" ensemble simulation of present and future climate, completed by the distributed computing project climateprediction.net, is used to search for constraints on the response to increasing greenhouse gas levels among present day observable climate variables. The search is conducted with a systematic statistical methodology to identify correlations between observables and the quantities we wish to predict, namely the climate sensitivity and the climate feedback parameter. A sensitivity analysis is conducted to ensure that results are minimally dependent on the parameters of the methodology. Our best estimate of climate sensitivity is 3.3 K. When an internally consistent representation of the origins of model-data discrepancy is used to calculate the probability density function of climate sensitivity, the 5th and 95th percentiles are 2.2 K and 6.8 K respectively. These results are sensitive, particularly the upper bound, to the representation of the origins of model-data discrepancy. Copyright 2005 by the American Geophysical Union.

[1] Observed globally-averaged land precipitation changes over the 20th century are compared with simulations of the HadCM3 climate model using an “optimal fingerprinting” method. We find that observed changes in precipitation are too large to be consistent with model-generated internal variability and are consistent with (attributable to) the combination of natural and anthropogenic forcings applied to the model. By comparing precipitation observations to shortwave and longwave forcing timeseries, we find that most of the forced variation in precipitation appears to be driven by natural shortwave forcings. We are unable to detect a response to anthropogenic longwave forcings in isolation. Finally, we seek to explain these results in terms of perturbations to the energy budget of the troposphere.

[2] The IPCC Third Assessment Report [Mitchell et al., 2001] describes the application of ‘‘optimal fingerprinting’’ techniques [Hasselmann, 1997] to the detection of a combined greenhouse gas plus sulphate aerosol (GS) response over the past 50 years. The response patterns simulated by seven climate models were found to be detectable in observations of surface temperature; the amplitudes of these patterns in the observations were found to be inconsistent with zero. However, there were considerable differences in the magnitude of the response between the models, with some simulating a response consistent with that observed, while others predicted a response significantly larger than that observed. When these techniques were used to estimate the amplitudes of the greenhouse gas (G) and sulphate aerosol (S) response patterns separately, these inter-model differences became larger, with simultaneous detection of G and S possible with some models, but not with others. These results from multiple models are synthesized only qualitatively by Mitchell et al. [2001]. Here we suggest a method for doing so more quantitatively. [3] Lambert and Boer [2001] compared coupled model climatologies of surface air temperature with observations using the CMIP ensemble. They found that overall the mean climatology of the models matched the observations better than that of any individual model. Similarly in seasonal forecasting, Krishnamurti et al. [1999] and Kharin and Zwiers [2002] argued that a weighted sum of multiple model predictions performs better than predictions using an individual model. These results might be explained if each model has independent errors, each giving different characteristic biases in model output. Thus, much as we can reduce the effects of initial condition uncertainty by averaging over an ensemble of integrations with perturbed initial conditions, so we might also account for model uncertainty by averaging over multiple models. Such an argument is also likely to apply to the anthropogenically-forced responses of these models. Thus here we describe how a mean of the response patterns of five climate models (HadCM2, HadCM3, ECHAM3, CGCM1 and CGCM2) may be used to detect greenhouse gas and sulphate aerosol influences in surface air temperature. [4] In one standard approach to the detection of anthropogenic influence on surface temperature [Allen et al., 2002], signal-to-noise optimised observations are regressed against a modelled response pattern, using a total least squares fit. Climate model output enters the analysis at three points. First, output from integrations of the model with prescribed time-varying forcings is used to derive the signal patterns of climate response. Second, output from a control integration is used to estimate the autocovariance matrix representing internal variability. This covariance matrix is used to derive the EOF basis used for truncation of the signal patterns and signal-to-noise optimisation. Third, an independent section of control data is used to estimate the uncertainty in the derived regression coefficients. In this study, output from multiple models is used in all three stages.

We assess the robustness of previous optimal detection and attribution studies considering zonal-mean temperatures. Principal results, which have consistently pointed towards a demonstrable anthropogenic influence on recently observed upper air temperatures, are confirmed. Importantly our detection results are not critically dependent on the inclusion of stratospheric as well as tropospheric temperatures. We find that detection is dependent on input field pre-processing choices, and on the choice of detection algorithm. There are a number of cases where either no signals are detected, or results fail a consistency test.

[1] The Arctic Oscillation index has increased significantly over the past forty years, and such an increase has been simulated in response to greenhouse gas increases in several climate models. However, it has been suggested that an atmospheric model with an upper boundary in the upper stratosphere or mesosphere is required to simulate a realistic response, and that predictions made with standard climate models are hence unreliable. Here we show that a climate model with a 30-km upper boundary shows no increase in its surface Arctic Oscillation response to doubled carbon dioxide when its upper boundary is raised to 80 km. Neither model version shows a significant Arctic Oscillation response to stratospheric ozone depletion.