GC43A-0933
Projected Evolution of Atlantic Tropical Cyclone Activity in the 21st Century
Vecchi and Soden (2007) have shown that the vertical wind shear projected by multi-model ensembles for the 21st century increases in the tropical North Atlantic, favoring a reduced tropical cyclone activity, contrary to some expectations based on the increased maximum potential intensity (MPI) of hurricanes. By considering the results of several recent studies, together with the projections of the Intergovernmental Panel on Climate Change (IPCC), there emerges a consistent physical rationale for expecting that, on the time scale of the 21st century, Atlantic tropical cyclone activity could gradually decrease overall. Recent research based on observations and model studies confirms that when the tropical North Atlantic is cool (warm) and the Atlantic warm pool is small (large), both the vertical wind shear and the moist static stability over the main development region (MDR) for hurricanes are increased while tropical cyclone activity decreases, accordingly. Whether or not this relationship holds as tropical Atlantic sea surface temperatures (SST) trend upwards, should depend on the relative rate of SST increase in the Atlantic as compared with the Pacific and Indian Oceans. If, as suggested by the IPCC multi- model data set, the North Atlantic warms more slowly than other oceans as the Atlantic meridional overturning circulation (AMOC) decreases, then a physically consistent scenario suggests that both the vertical wind shear and the moist static stability over the MDR should gradually increase, thus causing tropical cyclone activity to decrease. This should occur on the centenary time scale for which the AMOC is expected to decrease, but could be preceded by shorter term changes in activity in either direction due to multidecadal climate fluctuations. It is also possible that the expected decrease could be at least partially offset by an increase in the MPI of major hurricanes due to the absolute increase in local SST.
GC43A-0934
Quantifying the change in extreme seasonal precipitation events under global warming using a grand ensemble experiment
Estimates of future precipitation extremes are subject to much uncertainty. Uncertainties in emission rates, climate model structures, parameterisations and initial conditions add to the uncertainties in the prediction of extremes simply due to rarity of such events. Here, we produce probabilistic predictions of seasonal changes in extreme precipitation for regions across the globe using results from the climateprediction.net experiment. In this experiment, a coupled atmosphere-ocean global climate model was run in ‘grand ensemble' mode for a transient integration from 1920 to 2080, varying parameter values, initial conditions and forcing scenarios. Here, we examine changes separately for each forcing scenario but examine the uncertainties introduced by different parameterisations and initial conditions. We use extreme value analysis to define extremes of precipitation. We fit the Generalized Extreme Value distribution to annual maxima using L-moments, to estimate extremes with return values of between 5 and 50 years for moving 30-yr time slices within the transient integration. We then apply the principle of equal weighting of the results from different models in the production of probability distributions of change for different regions of the globe. This allows us to establish which regions are most sensitive to the impacts of global warming on precipitation extremes, and the likely rates of change.
GC43A-0935
Solar-cycle response in global climate models assessed by IPCC AR4
Do IPCC models underestimate the earth's climate response to solar variation? The adequacy of the IPCC models in this regard can be checked by examining their response to the 11-year solar cycle, whose forcing has been measured accurately by satellites since 1978. In response to this cyclic forcing previous generation of GCMs produced a very weak warming signal at the Earth's surface, of the order of 0.06 C from solar min to solar max, leading to cautionary comments even in AR4 that current models may underestimate solar response, in both long-term and 11-year variations. Haig [1996] pointed out that the previous attempts at GCM simulation of the 11-year solar cycle used fixed sea-surface temperature, which may have inhibited the various climate feedback processes. The GCMs assessed by IPCC AR4 incorporates coupled ocean and atmosphere dynamics, but due to a lack of study at this point it is not clear if the solar cycle response is correctly reproduced in these models. Using multi-model archived outputs in the repository, we compare model temperature responses in models with and without solar cycle forcing. These are additionally compared with the recent observational result of Camp and Tung [2007], who found a globally averaged warming of almost 0.2 C from solar min to solar max.
GC43A-0936
Utilizing the Koeppen climate classification to assess the future climate change
It is suggested that global warming due to anthropogenic greenhouse gasses will cause a large change in the mean temperature and precipitation patterns of the future. One way to quantify the impact of this change is to use the climate classification method. Classifying the climate into regions with distinct properties instead of using only physical properties such as temperature and precipitation helps to give an objective view of how climate change affects the environment such as the land-surface types and vegetation. TheKoeppen climate classification has a long history of application and modification and is known to give a robust classification of the mean climate that closely follows the distribution of vegetation types. In this study, we apply theKoeppen climate classification on the result of 19 Atmosphere-Ocean GCM results provided by the PCMDI for the upcoming IPCC - AR4. By applying this method to the long-term future projection of climate models, instability of a particular climate region and its expected change in the longer timescales are quantified. The classification is performed on the 20th century simulation (20C3M) and the SRES-A1B / A2 scenario based on the long-term monthly climatology. The overall changes in classifications as well as inter-model distribution is calculated for all each model and the skill weighted ensemble mean. Results show that due to warmer climate and increase in moisture, large area of western Russian region and north America experience a shift from aDf (snow / fully moist) climate to Cf (Warm temperate / fully moist) classification which is in good agreement with the stronger NAO/AO phase in the north Atlantic. On the other hand, coastal Greenland region changes from a Ef (Polar frost) classification to Ef (Polar tundra) classification, which is in good agreement with the SST and sea-ice distribution. In contrast, northern China undergoes a change from Cf classification to Cw (Warm temperate / winter dry) classification which marks a drying of this region. Weakening of the Aleutian low and a strongerENSO signal among models may have contributed to this result. In the presentation, major changed in classification and its physical background is highlighted.
GC43A-0937
Towards an Objective Characterization of Climate Model Performance
This study is based on previous work where we measured the performance of models in terms of their ability to simulate the observed climate mean state for a wide range of quantities. We have shown that the mean of a multi-model ensemble consistently outperforms any individual simulation. Now an important question is how to construct an optimally weighted multi-model mean which maximizes the strengths while minimizing the weaknesses discovered in the model conglomerate. Consequently, we explore ways to reduce our rather comprehensive choice of climate quantities into a much smaller subset. Our goal is to derive an unbiased description of model performance retaining a significant proportion of information while neglecting a considerable amount of data redundancy. Statistical methods as diverse as cluster analysis and principal component analysis are shown to be successful in producing a minimal collection of climate quantities which are distinctly useful in model evaluation. To the first order, this subset consists of two variables: one primarily representing model physics, while the second mainly represents model dynamics. We apply these results to the IPCC-AR4 ensemble and demonstrate how it can be used to construct an optimally weighted average of many models.
GC43A-0938
An Archive of Downscaled WCRP CMIP3 Climate Projections for Planning Applications in the Contiguous United States
Incorporating climate change information into long-term evaluations of water and energy resources requires analysts to have access to climate projection data that have been spatially downscaled to "basin-relevant" resolution. This is necessary in order to develop system-specific hydrology and demand scenarios consistent with projected climate scenarios. Analysts currently have access to "climate model" resolution data (e.g., at LLNL PCMDI), but not spatially downscaled translations of these datasets. Motivated by a common interest in supporting regional and local assessments, the U.S. Bureau of Reclamation and LLNL (through support from the DOE National Energy Technology Laboratory) have teamed to develop an archive of downscaled climate projections (temperature and precipitation) with geographic coverage consistent with the North American Land Data Assimilation System domain, encompassing the contiguous United States. A web-based information service, hosted at LLNL Green Data Oasis, has been developed to provide Reclamation, LLNL, and other interested analysts free access to archive content. A contemporary statistical method was used to bias-correct and spatially disaggregate projection datasets, and was applied to 112 projections included in the WCRP CMIP3 multi-model dataset hosted by LLNL PCMDI (i.e. 16 GCMs and their multiple simulations of SRES A2, A1b, and B1 emissions pathways).
GC43A-0939
Seasonal Forecast Datasets: a Resource for Calibrating Regional Climate Change Projections
Probabilistic projections of regional climate change are currently being used for long-term planning. These projections are of value provided the associated probabilities are trustworthy. However, by the nonlinear nature of climate, finite computational models of climate are inherently deficient in their ability to simulate regional climatic variability with enough accuracy. Therefore, in the light of such generic deficiencies, the hypothesis of whether regional climate-change projections are untrustworthy should be tested. A calibration method is proposed whose basis lies in the notion of seamless prediction: if essentially the same ensemble forecasting system can be validated probabilistically on timescales where validation data exist, ie on daily, seasonal and decadal timescales, then climate-change probabilities obtained with the same systems could be objectively modified or calibrated using probabilistic forecasts on shorter timescales. Specifically, calibrated probabilities of regional climate change are derived from analyses of the statistical reliability of seasonal probabilistic predictions obtained from multi-model ensembles. The method is demonstrated by calibrating probabilistic projections from the multi-model ensemble of the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC) using reliability analyses from the seasonal-forecast DEMETER multi-model dataset. The focus is on climate-change projections of regional precipitation, though the methodology is more general. The examples provide some justification for the development of seamless prediction systems across weather and climate timescales.
GC43A-0940
Paleo-Constraints In Ensemble Climate Modelling
The aim of this work is to place a paleo-constraint on climate sensitivity in the climateprediction.net ensemble. Sensitivity is here defined as the equilibrium temperature response to doubling of pre-industrial carbon dioxide concentrations. Some of the climateprediction.net models have shown a large sensitivity and by applying these models to past climates, we are testing if these models and their corresponding sensitivity are realistic. The general circulation model results are compared to paleo-observations. The period studied here is the mid-Holocene, i.e. ~6000 years before present (6kyBP). Why use the mid- Holocene climate to benchmark our models? The current climate is not in equilibrium, it is changing. The previous period with a stable climate was the mid-Holocene, which is reasonably well known through paleo- observations. Four proof-of-concept runs are presented, which show that mid-Holocene simulations do distinguish between different models. A large ensemble of paleo-climate models will be distributed on climateprediction.net exploring paleo-boundary conditions including ice-sheets, vegetation and ocean heat flux convergence values. http://climateprediction.net
GC43A-0941
On a Possible Bias of Climate Change Predictions Based on the Multi-model Ensemble Approach.
The response of the climate system to prescribed changes in the concentrations of greenhouse gases and aerosols depends on the characteristics of the system such as the climate sensitivity, rate of heat uptake by the ocean and strength of aerosol forcing. All these characteristics are highly uncertain, leading to the uncertainty in future climate. Different approaches have been used for producing probability distributions for the changes in surface air temperature (SAT) and other climate variables, including the multi-model ensemble approach. Distributions obtained using the latter approach may be biased for a number of reasons. The ranges of possible SAT changes given in IPCC AR4 were adjusted to account for the fact that values of climate sensitivity and rate of oceanic heat uptake for AOGCMs used in AR4 simulations do not cover the full uncertainty ranges. These SAT ranges are, however, based on the multi-model ensemble means. The rates of the oceanic heat uptake for most of AR4 AOGCMs lie in the upper part of the range suggested by observations. As a result the means of the multi-model ensemble are likely to be biased toward low warming, especially in the simulations with large increase in radiative forcing (SRES A2). Here we evaluate the uncertainty in climate response to prescribed changes in greenhouse gas concentrations using the MIT Integrated Global System Model (IGSM2.2).We carried out three 250 member ensembles for SRES scenarios B1, A1B and A2. Probability distributions for the climate sensitivity, strength of aerosol forcing and the rate of ocean heat uptake were obtained by comparing 20th century climate as simulated by the IGSM with available observations. Our simulations suggest that by the end of the 21st century (2090-2099) surface air temperature is likely to increase above the present level (1980-1999) by 1.6C to 2.3C for B1, 2.5C to 3.6C for A1B and 3.4C to 4.6C for A2. Corresponding ranges for a sea level rise due to thermal expansion are: 9 cm to 18 cm for B1, 13 cm to 25 cm for A1B and 16 cm to 29 cm for A2. The mean values of surface warming obtained in our simulations for these three scenarios, 2.1C, 3.2C and 4.0C respectively, are noticeably higher than the AR4 multi-model ensemble means, 1.8C, 2.4C and 3.4C. The upper bounds of the possible SAT increases obtained in simulations with the MIT IGSM2.2 significantly exceed the warming simulated by AR4 AOGCMs. Thus for the high emissions scenario (A2) the results of all AOGCMs lie in the low half of the range (below the mean) suggested by the MIT IGSM and below the 67% percentile for the other two scenarios. The situation is the opposite for the sea level rise due to thermal expansion. The results presented illustrate the importance of properly sampling the full range of uncertainties in the input parameters when predicting future climate change by means of numerical simulations.
GC43A-0942
Skill-Change Relations and Inter-Model Scatter in CMIP3 Surface Field Projections
CMIP3 projections of future climate change agree fairly well on global changes of mean surface temperature (T) and precipitation (P), but there are still considerable differences on regional scales. An interesting question is whether these uncertainties can be somehow constrained. Although the global multi-model mean skill for T and P is higher than the skill of the best model we show that there are indeed some robust relations between simple global skill measures of the present-day representation of T and the projected changes in T especially in the tropics and near the poles. Local skill-change relations on the other hand reveal mainly model biases (e.g. caused by excessive snow pack in spring). While these are useful for identifying model deficiencies and stimulating model developments, they are not necessarily useful for constraining uncertainties in future changes. The amplitude and inter-model scatter (IMS) of projected late 21st century changes in T and P extremes is analyzed using the recently introduced Climate Change Index (CCI). The IMS expressed as the inter-model standard deviation varies between 0.37 for the western subtropical Pacific and 4.69 in the northern Atlantic deep water formation region (a factor of 12.8). It is generally smallest in the subtropics where the CCI shows average values. Large IMS is found in the ENSO region, Amazon basin and the polar regions. In the former two regions, the scatter is mainly related to differences in the changes of P extremes. For the polar regions, the large scatter is caused by different changes in temperature and precipitation extremes. This seems mainly related to model- dependent differences in sea ice, snow and possibly cloud changes.
GC43A-0943
Assessment of the use of Current Climate Patterns to Evaluate Regional Enhanced Greenhouse Response Patterns of Climate Models
Output of multiple global climate models is often considered in the generation of regional climate change projections. The reliability of the regional responses of models to enhanced greenhouse forcing is often assessed by comparing their current climate simulation against observations. The rationale for this assessment is that a model should be able to reproduce key aspects of the present climate if it is to be used to provide guidance for future changes in climate. However, the best way to assess the current climate of a model is unresolved. One can assess regional average or grid point model biases for the variable and season for which projections are to be prepared. However, a model that performs well for a target variable, season and location, may perform poorly for another variable, season or location, in which case model processes would be suspect. Other approaches consider spatial patterns and multiple variables, but this then raises the issue of how large an area the patterns should cover and what set of variables should be considered. We demonstrate how such a pattern-based approach can be evaluated by investigating the relationship between inter-model similarity in patterns of current regional climate and inter-model similarity in patterns of regional enhanced greenhouse response using data from the CMIP3 multi-model database. By making the assumption that the real world behaves like a typical climate model, we can use these results to assess whether the testing of the current climate patterns of the models against observations can be used to discriminate amongst enhanced greenhouse results of the models. Correlations of moderate magnitude are common in our results, indicating the value of testing the current regional climate simulation of models. Notably, relationships vary significantly regionally (e.g. are weakest in the tropics), cross variables (e.g. current climate mean sea level pressure is related to temperature and precipitation change in some extra-tropical regions) and can be insensitive to whether the current climate is assessed in the region concerned or globally. The presentation will also consider how the approach may be applied to performance-weighting of models in the generation of regional climate projections and in the assessment of the independence of output from different models. http://www.agu.org/journals/gl/gl0714/2007GL030025/2007GL030025.pdf
GC43A-0944
The Hadley circulation changes of the Atmosphere Ocean GCM under global warming
The mean meridional tropical circulation of the Atmospheric Ocean coupled General Circulation Model (AOGCM) used in the Fourth IPCC assement (AR4) report is diagnosed, focusing on the CMIP3 simulation. In the conditions of a doubling in the carbon dioxide concentration, the AOGCMs show a gentle but significant weakening of the Hadley circulation for the winter cell in both hemispheres, accompanied by a poleward extension of the Hadley circulation area. The particular AOGCM IPSL-CM4 show similar changes for the austral winter but different for boreal winter. The northern hemisphere Hadley cell of the IPSL-CM4 AOGCM is stronger and show little changes in its extension, under global warming. The conditions explaining the modification of the Hadley circulation are analyzed using detailed outputs from the IPSL-CM4 AOGCM. At the first order, the changes of the Hadley circulation can be seen as a balance between the modification of the dry static stability, the latent and the radiative heating, and the eddy polaward transport, as a response of the SST changes. The changes of the southern hemisphere cell of the AOGCM IPSL-CM4 are due to the stronger increase of the dry static stability than the radiative cooling changes. For the northern hemisphere changes, an increase in the zonal component of the SST explain stronger convection and latent heating, which overcome the effects of the dry static increase.
GC43A-0945
Use of CMIP3 simulations to estimate the changes of temperature indicators over France and Europe during the 21st century
Projections of changes in temperature are essential to assess the impact of climate change on the energy supply sector as heating and cooling, energy demand highly depends on temperature. A selection of temperature indicators and their changes are examined for several simulations using SRES Emission Scenario A2 from the CMIP3 archive. We compare the present day simulated indicators to those in European Center for Medium-Range Weather Forecasts (ECMWF) ERA40 reanalysis The results are analysed for six areas over Europe and two time periods during the 21st century. We focus our study on changes in number and duration of hot and cold events and on changes in heating degree-days and cooling degree-days, which are commonly used to estimate the weather-related variations in energy consumption. Results are presented for the different models with some comparisons to the regional model simulations from the European PRUDENCE project to evaluate uncertainties.
GC43A-0946
Assessing Climate Change Risks Using a Multi-Model Approach
We quantify the risks of climate-induced changes in key ecosystem processes during the 21st century by forcing a dynamic global vegetation model with multiple scenarios from the IPCC AR4 data archive using 16 climate models and mapping the proportions of model runs showing exceedance of natural variability in wildfire frequency and freshwater supply or shifts in vegetation cover. Our analysis does not assign probabilities to scenarios. Instead, we consider the distribution of outcomes within three sets of model runs grouped according to the amount of global warming they simulate: < 2 degree C (including committed climate change simulations), 2-3 degree C, and >3 degree C. Here, we are contrasting two different methods for calculating the risks: first we use an equal weighting approach giving every model within one of the three sets the same weight, and second, we weight the models according to their ability to model ENSO. The differences are underpinning the need for the development of more robust performance metrics for global climate models.
GC43A-0947
Assessing the low-frequency variability in ensemble simulations of western North American hydroclimate for 20th and 21st century
Recent availability of multimodel coupled global climate ensemble simulations of the 20th and 21st century climate affords a detailed diagnosis, detection and attribution of regional hydroclimatic change. We examine the interannual-to decadal and longer-term variations in the western North American region for the 20th and 21st century. Two aspects are specifically examined: a. Changing low frequency character of hydroclimatic variations and implications for water supply reliability and frequency of hydrologic extremes. b. the nature of ENSO teleconnections for western North America in the coupled climate model simulations, and the projected changes for the 21st century. Within this context, the utility of multimodel ensembles for water resources management and planning is discussed.
GC43A-0948
First Results of the Climateprediction.net BBC Climate Change Experiment
Climateprediction.net is a simulation project harnessing the power of idle PCs to forecast the climate of the 21st century. In collaboration with the BBC, we asked volunteer members of the public to download a climate model from the project website and to run it locally on their PC. Each model forms a single member of a massive, perturbed-physics ensemble in the world's largest climate forecasting experiment. As well as discussing the design of the experiment, we present some of our first results.
GC43A-0949
A kinematic representation of spatiotemporal scalar fields to support mining dynamics in data output from general circulation models.
Effective information analytics is required to decipher massive data output from general circulation models (GCMs) and other spatially explicit models of environmental dynamics. Common approaches with discrete space or time constructs overlook the fundamental characteristics of continuity in dynamics. We propose a representation that centers on the concept of kinetics to effectively capture dynamics by the direction and amount of change in space and time, i.e. velocity. In scalar fields, such as temperature, output from GCMs, we first define isolines to represent spatial variations of the fields. We then determine the velocity of isoline movement across time. Similarly, features of hot or cold spots can be identified from a scalar field of temperature. Velocity determined by the direction and amount of boundary change can capture deformation and movement of these features. The kinematic representation enables the comparison of multiple GCM's output at an increased level of abstraction through isoline and feature identification, and furthermore it enables the analogous comparison of climate change suggested by data output from multiple general circulation models. A comparison of the Center National Weather Research global coupled system and the National Center for Atmospheric Research Community Climate System Model output for IPCC scenario A2 was made. Our results indicate that climate change patterns from the two models are well correlated except for a high latitude band spanning Greenland, the northern Atlantic region and northern Eurasia. The kinematic representation presented by this paper enabled the spatiotemporal analysis of massive data sets, highlighting smaller spatial regions where further research may be productive in understanding the differences between different GCM models.
GC43A-0950
Characterizing AOGCM Ability to Simulate Northern Hemisphere Teleconnection Patterns
An important aspect of coupled atmosphere-ocean general circulation models (AOGCM) is their ability to simulate variability in regional and global atmospheric dynamics. This is particularly true for recurring teleconnection patterns known to be correlated with surface climate anomalies. Here, we evaluate the ability of all IPCC AR4 historical 20C3M AOGCM simulations for which the required output fields are available to simulate five present- day patterns of large-scale atmospheric internal variability in the Northern Hemisphere: the Arctic Oscillation (AO), the North Atlantic Oscillation (NAO), the Pacific/North American Oscillation (PNA), the Pacific Decadal Oscillation (PDO), and the El Niño - Southern Oscillation (ENSO). We evaluate these patterns in two ways: first, in terms of their characteristic temporal variability, and second, in terms of their magnitude and spatial locations. We find that historical total forcing simulations from nearly all of the AOGCMs are able to produce seasonal spatial patterns that clearly resemble the teleconnection patterns that result when identical calculation methods are applied to ECMWF ERA-40 and NCEP / NCAR reanalysis fields. However, many AOGCMs tend to either over- or underestimate the strength of the patterns, and also tend to rotate the AO, NAO, PNA, and PDO about the polar region. Furthermore, based on spectral analysis of the time series of each index, AOGCMs also vary in their ability to simulate the temporal variability of the teleconnection patterns, with some models oscillating too fast and others too slow relative to observed. We conclude, therefore, that although historical simulations from all the AOGCMs examined here were able to produce patterns that resemble those seen in reanalysis fields, significant biases still remain in model representation of these patterns, including biases in the strength and/or location of the patterns as well as in their temporal variability.
GC43A-0951 [WITHDRAWN]
Changes in the Subduction of Southern Ocean Water Masses in ten IPCC Models
The Southern Ocean's Subantarctic Mode and Antarctic Intermediate water masses dominate the oceanic uptake of anthropogenic CO2 and store heat, freshwater and dissolved gases. We analyse the ocean's response to changes in surface forcing using two different methods and across ten models from the IPCC Fourth Assessment Report. We compare the 1950's 20th century mean climate and the 2090s A2 scenario, where CO2 concentration reaches 860 ppm by the year 2100. Subduction rates are analysed on density surfaces (which evolve with time) and on surfaces that are fixed in time, defined by the position of density surfaces in the 1950s. We diagnose the subduction and entrainment rates of mode and intermediate water masses on both surfaces. The model means show a decrease in the subduction rates resulting from warming and freshening at the ocean's surface in the South-east Indian and Pacific Oceans. A corresponding cooling and freshening is found on density surfaces within the ocean interior in these two water masses (upper 1500 m) and compare well with observations. There is a gain in the net surface buoyancy flux at mode water densities due to the increased northward Ekman transport. The surface buoyancy flux decrease at intermediate water densities is dominated by warmer surface waters and to a lesser extent by the surface heat fluxes and through increases in the vertical stratification below the mixed layer. Our multi-model analysis shows a decrease in the renewal rates of Southern Ocean water masses and hence affecting the uptake of anthropogenic CO2.
GC43A-0952
Improving Probabilistic Rainfall Projections for Australia
One of the aims of developing new climate projections for Australia is to better address the requirements of stakeholders - particularly those who require less uncertainty and/or probabilistic information to work with. Recent projections differ from those issued previously, but there are two risks that need addressing. If the uncertainty is only narrowed slightly, then this may not satisfy some stakeholder requirements. Furthermore, if the differences between old and new projections are not large, then this could be construed as evidence that the problem has been solved (as much as it can be), and that further work in this area is not justified. Here we argue that it is possible to derive more reliable and less uncertain information about the future which may avoid these risks. One of the important steps in developing projections is deciding how to weight various model results. This is a controversial topic and there are a range of opinions, both here and overseas, on how make sense of a large number of results, some of which are often contradictory. Some argue for no weighting – on the grounds that a large number of results represent a real sample of possible outcomes and should not be tampered with. On the other hand, there are compelling reasons to downgrade, if not entirely dismiss, some model results based on their failure to satisfy some obviously necessary criteria. Current techniques already dismiss some model results and weight others, so the problem becomes one of deciding how to do this in a more consistent fashion, and investigating whether this makes a difference. We show how a critical approach to model performance can result in severe weighting of models and that this leads to significantly different projections for the Murray Darling Basin region. These are characterised by a shift in the probabilities towards much drier conditions.
GC43A-0953
Climate Clues From the Caspian Sea: Gas Hydrates and Seafloor Deformation
The presence of buried gas hydrates in the South Caspian Sea has been interpreted using 2-D seismic reflection data. Evidence for buried gas hydrates consists of a shallow (300-500 m below seafloor) comparatively high velocity zone approximately parallel to the seafloor bounded by a positive-polarity reflector at the top and a high- amplitude negative-polarity reflector at the base. The position of gas hydrates fall within the stability field predicted for the South Caspian Basin. New industry quality 3-D seismic data are being used to determine the relationship between the presence of gas hydrates and an approximately 2,500 km2 late-Pleistocene zone of seafloor deformation and submarine slumping in the South Caspian Sea, named the Absheron Allochthon. Well-logs from the South Caspian Sea offshore Azerbaijan are being correlated with the 3-D seismic data to determine the age and origin of the Absheron Allochthon. The history of sea level changes in the Caspian Sea in the past ~700 ka emphasizes a major ~100 m drop in sea level during the latest deglaciation (late Pleistocene). Preliminary results show that the Absheron Allochthon may have formed through catastrophic failure of the western South Caspian continental slope as a result of dissociation of underlying buried gas hydrates. If this proves to be true, then repeated, remarkably rapid global warming events in the Caspian Sea during the late Quaternary were likely due to rapid marine gas hydrate (clathrate) dissociation rather than exhalation from wetlands.