HABILIDADES LINGÜÍSTICAS Y APRENDIZAJES MATEMÁTICOS:
DESARROLLO ATÍPICO
2.5. SÍNTESIS DEL CAPÍTULO SEGUNDO
Evidence that humankind is affecting the climate system is now overwhelming (IPCC, 2007). However, projections of the magnitude of possible future changes are limited by an incomplete knowledge of the skill of models for making these projections. Models are tested on their ability to reconstruct the observational period (Braconnot et al., 2012), however the magnitude of climate change in this period is relatively small (~0.75°C - IPCC, 2007). Palaeoclimate offers a solution to this with time periods that display a range of changes to global temperatures that can be reconstructed from palaeo-data enabling the dynamics of warmer world climates to be investigated (Haywood et al., 2009a; Braconnot et al., 2012). Although no perfect analogue from the geological record exists for the likely projections of 21st century climate change (Haywood et al., 2011b), the Pliocene, specifically the mid-Pliocene Warm Period (mPWP) is the most parsimonious epoch to study in this regard. It has a similar to modern continental configuration, elevated atmospheric CO2 levels and warmer global mean temperatures compared to modern. The mPWP also has the largest global palaeo-environmental reconstruction for a warmer than modern climate enabling models to be more thoroughly tested than any other time in the geological record. As a result, the mPWP is an important geological period to study with models and palaeo- data to understand the climate system of a warmer than modern world.
The George Box axiom, “All models are wrong, but some models are useful” (Box & Draper, 1987) is ably demonstrated by mPWP modelling studies. The majority of mPWP climate model based studies have utilised a single climate model (Section 1.3.2) and have compared the results from the simulations to the PRISM palaeo-data. Data- model comparisons from these single models have highlighted strengths and weaknesses of individual climate models and enabled the identification of regions where a climate model is unable to reproduce the data (Section 1.3.4), but are unable to explain why. However, no account of the inherent uncertainty in climate model simulations has been accounted for in this previous work. A range of climate projection
studies investigating structural and parameter uncertainty have displayed the scale of the uncertainty on results produced by a climate models (Section 1.4.2 & 1.4.3).
Two initiatives have been undertaken to tackle this weakness in mPWP climate modelling. The first, is the Pliocene Model Intercomparison Project (PlioMIP – Haywood et al., 2010; 2011a; 2013a), which has produced a multi-model ensemble tackling the structural uncertainty in Pliocene climate modelling bringing together modelling groups from across the globe. The second is the subject of this thesis, the perturbed physics ensembles, designed to investigate parameter uncertainty in mPWP climate simulations. Although unconnected, the two are complementary, focussed on increasing our understanding in the mismatches between model and data reconstructions of the mPWP. The main component of the data-model mismatch for the mPWP focusses around the inability of fully coupled atmosphere-ocean general circulation models (AOGCMs) to simulate the high latitude warming of the mPWP climate, especially through the North Atlantic (Section 1.3.4).
Existing PPEs have focussed on future climate change projections (i.e. Collins et al., 2006; 2011), with work underway to investigate PPEs at the Last Glacial Maximum (Gregoire et al., 2011) and the Eocene (Sagoo et al., 2013). The Pliocene PPE presented here represents the first investigation of parameter and boundary condition uncertainty in a warmer than modern climate with similar continental configuration. It is the first palaeo-PPE to be tested against SST, SAT and vegetation biome data, using these three datasets to produce a combined ranking of the ensemble members.
1.5.1. Aims and Objectives
The thesis will investigate the contribution of parameter uncertainty and boundary condition uncertainty on the simulation of mPWP climate. Parameter and boundary condition uncertainty represent two of the three areas of model uncertainty relevant to palaeoclimate (see Section 1.4.1) forming vertices of the “PMIP Triangle” (Haywood et al., 2013a). The ensembles will be assessed in comparison with palaeoclimate data. The focus of these data-model comparisons will be to determine how the ensemble members vary in comparisons to individual palaeo-datasets and also across the combined range of palaeo-datasets available for use. The variation in ensemble member performance will enable the assessment of the ensemble and the impact of the two forms of uncertainty upon the simulation of the mPWP climate within HadCM3. The aim of the thesis is to identify the ensemble members which reduce the existing
range of palaeo-data utilised in this thesis. By utilising both sets of boundary conditions used in previous mPWP simulations, our PPE can also identify which boundary conditions produce the best data-model comparisons across the different palaeo-datasets used and in the combined rankings. The PRISM3D boundary conditions represent an improved understanding of the mPWP, but it is important to test whether they produce a stronger representation of the mPWP climate than PRISM2 boundary conditions. Similarly, the potential range of values of mPWP atmospheric CO2 will be included within the assessment of ensemble member performance, allowing for uncertainty in this boundary condition.
The goals of the investigation are:
1. To investigate the mPWP global and regional climate responses to the parameter perturbations, including the effect on these responses from the potential range for mPWP atmospheric CO2 concentrations.
2. To investigate the effect of changing the physical boundary conditions within the model, from PRISM2 to PRISM3D and the interaction between the boundary condition changes and the parameter perturbations
3. To investigate the effect parameter perturbations and boundary condition changes have on data-model comparisons to vegetation derived surface air temperature, sea surface temperature and vegetation biome data.
The perturbed physics ensemble presented here represents a thorough investigation of the uncertainty in mPWP climate simulations with AOGCMs due to the representation of sub-grid scale parameterisations and the physical boundary conditions. It is unlikely that any single ensemble member will provide a perfect reconstruction of the mPWP climate, however the range of performance across the ensembles and DMCs will provide useful information in where mPWP modelling can be improved. All models are wrong, but some members of our PPE are useful for reducing data-model mismatches.