The Earth’s climate system includes atmosphere, hydrosphere, cryosphere, land surface, biosphere and the interactions between these components. The climate system is also subject to external influences such as incoming solar radiation (Baede et al, 2001). Climate models have been developed to mathematically represent physical processes and interactions in the climate system. Models may incorporate either a single component of the climate system or couple several components (Baede et al, 2001). The equations, which model these complex physical processes are solved using numerical methods, written in computer code, and run on powerful computers (Baede et al, 2001).
(For more on CATE’s computer facilities, click here)
In recent years, climate models have been able to incorporate physical processes in greater complexity, such as coupling of climate system components and enhanced model resolution. This is made possible with the massive improvements in computer processing power (McGuffie & Henderson-Sellers, 2005), (Maslin, 2005). Further development of supercomputers should allow further increases in spatial, vertical and temporal resolutions. Scientific observations are used to validate and refine model outputs (Cess et al, 1989).
One method of validating climate models is to run these models over the past millennium and determine whether they can simulate the changes in past climates obtained from proxy data. This method was used to test the performance of the climate models used in the IPCC Fourth Assessment Report (AR4) (Randall et al, 2007). Inter-model agreement on climate change and the skill of models in representing present day climate can also be used to validate climate models (Räisänen, 2007). The example shown below shows the ability of 14 models used to simulate temperature change in the last millennium (Randall et al, 2007).
Figure 1. Global mean near-surface temperatures over the 20th century from observations (black) and as obtained from 58 simulations produced by 14 different climate models driven by both natural and human-caused factors that influence climate (yellow). The mean of all these runs is also shown (thick red line). Temperature anomalies are shown relative to the 1901 to 1950 mean. Vertical grey lines indicate the timing of major volcanic eruptions (Randall et al, 2007)
Despite advances in scientific knowledge and computing, climate models often cannot offer sufficient resolution to model sub-grid scale processes, which occur below the temporal or spatial resolution of a model (Räisänen, 2007). Parameterisation of sub-grid scale processes remains an important method in climate models (Harvey, 2000), (McGuffie & Henderson-Sellers, 2005).
Climate models are subject to uncertainty from a variety of sources. The behaviour of the climate system over time is sensitive to the initial conditions in the model. This is due to the non-linear nature of the equations representing dynamic and physical processes in the atmosphere (Leutbecher & Palmer, 2007). Longer scale climate simulations, made over periods of decades, are less sensitive to variations in initial conditions than short time scale predictions (Tebaldi & Knutti, 2007). The finite nature of computer processing power also results in truncation errors in computing solutions to the physics and dynamics equations used in models (Leutbecher & Palmer, 2007), (Collins, 2007).
An estimate of model uncertainty can be derived from ensemble runs in climate models (Stainforth et al, 2005). One method is the use of perturbed physics ensembles (PPE). Each member of an ensemble of single model simulations is run over the same time period. Individual ensemble members have different values for model inputs for which the input values are uncertain e.g. parameterised physical processes or uncertain feedback processes. There is evidence that multi-model ensembles (MME) increase the skill, reliability and consistency of model forecasts (Trebaldi & Knutti, 2007). Model intercomparison projects such as the Coupled Model Inter-comparison Project (CMIP) are carried out to compare model performance using MMEs (Covey et al, 2003).These projects collate, compare and evaluate climate model results (Randall et al, 2007).
A range of computer models of various levels of complexity are used in climate research projects (McGuffie and Henderson-Sellars, 2005). The most sophisticated models are the General Circulation Models (GCMs). GCMs are the main models used to study potential anthropogenic influences on current and future climate (Solomon et al, 2007). Simple climate models (SCMs) are less complex and therefore, require less programming effort and are not computationally expensive in terms of hardware and runtime (Kump et al, 1999). The IPCC AR4 report (Solomon et al, 2007) utilises data from a mix of models: SCMs, Earth Models of Intermediate Complexity (EMICs) and Atmospheric-Ocean Coupled General Circulation Models (AOGCMs) (Solomon et al, 2007).
SCMs have been used as decision support tools by policy makers, to evaluate climate impacts from a wide range of sources; from total anthropogenic emissions down to sectoral emissions such as aviation. Two widely used SCMs; MAGICC for all anthropogenic sources and LinClim for the aviation sector [Link to both models] are available at CATE.
Model for the Assessment of Greenhouse-gas induced Climate Change (MAGICC), which is co-developed by Dr Sarah Raper, has been used since the IPCC Second Assessment Report (SAR) onwards (Kattenberg et al, 1996). The simple linear climate response model, LinClim (Lim et al., 2007), was developed and is currently maintained by CATE. LinClim has been used in aviation assessments such as the EU SSA ATTICA (Lee et al, 2010) and Lee et al., 2009.
MAGICC is a simple climate model to examine time-dependent effects. The model used emissions scenarios to calculate time-dependent concentrations, radiative forcing, temperature response and sea-level rise from different perturbations. MAGICC can be used in the formulation of climate metrics for total anthropogenic sources (Lee, 2010).
LinClim has a simple aviation emissions model that uses global fuel from aircraft emissions model such as FAST and published/calculated emissions indices to calculate historical, present day and future emissions trends for CO2, NOx, H2O and particles (black carbon and sulphate). Unlike emissions species that is related to aviation fuel burnt, distance travelled is used as proxy to calculate climate impacts from contrails and contrails-cirrus. The built-in carbon-cycle model in LinClim is used to calculate the CO2 concentrations due to aircraft emissions. The current version of LinClim is tuned to the carbon-cycle results published by Maier-Reimer and Hasselmann (1987). LinClim calculates the aviation impacts of non-CO2 species and contrails/contrails-cirrus by scaling the emissions or distance travelled to an externally calculated or published base year radiative forcing. Temperature response can be calculated by LinClim using radiative forcing and climate model parameters which are tuned to GCM experiments. At present, LinClim has a range of parameters for transient, pulse, 2xCO2 and 4xCO2 experiments from various GCMs, including those that participated in the CMIP3 exercise. The following figures show example applications of LinClim, from fuel to CO2 temperature response.
Historical fuel and example of two fuel scenarios
CO2 emissions from aviation fuel
CO2 concentrations from aviation emissions
Background CO2 concentrations for the RCP scenarios
Aviation RF for the CO2 concentrations depicted above for aviation emissions, run against two RCP scenarios
Aviation temperature response for the CO2 RFs depicted above using ECHAM4 and ECHAM5 tuning parameters
The RF and temperature results from LinClim, such as those illustrated in the figures above can be used to calculate Global Warming Potential (GWP) and Global Temperature Potential (GTP). In addition to emissions, these metrics form the basis of comparing:
The following interactive presentation illustrates the history of climate modeling