Climate Model Assessment
We will start by highlighting that the IPCC appears to base many of its statements and conclusions on climate models, such that we might also assume that in their assessment climate models already do a good job in predicting the future of climate change and support the conclusion that climate change can be broadly attributed to anthropogenic emissions of greenhouse gases, e.g. CO2. For they claim that:
“The development of climate models has resulted in more realism in the representation of many quantities and aspects of the climate system and it is extremely likely that human activities have caused more than half of the observed increase in global average surface temperature since the 1950s”
As a general assessment, it might be stated that climate models are indeed an important and improving tool, which can help climate scientists, and others, understand the complexity of climate change in terms of collating both qualitative and quantitative data. However, as previously argued, climate ‘models’ are by definition a simplification of the many potential factors influencing climate change, such that the accuracy of model predictions still has to be questioned as outlined in the following bullets.
- Inherent limitation in the mathematical modelling of climate dynamics
makes accurate prediction difficult and maybe near impossible as the
timescales are stretched further into the future.
- It is possible that future enhancement to climate models will not
only have to consider the implication of even more climate change mechanisms,
as outlined, but also expand the details concerning chemical and biological
processes that influence climate over longer time periods.
- Climate models are invariably limited by processing capacity, which
then limits the resolution or granularity of each climate cell within
the model, both in terms of size and time. This appears to be especially
true when trying to accurately model all the implications associated
- Of course, any inaccuracy in the model at the micro-level of the
cell might then be extrapolated as projections are extended into the
future. However, by the very nature of it being a future prediction,
there is often little comparative data by which to assess its accuracy.
- It might be argued that the variance in the output of different
models to a doubling of atmospheric CO2 might also suggest the scope
of potential errors, such that a degree of caution, rather than certainty,
must be highlighted.
- There this outline will highlight that there is still a large variance
in climate model outputs, which might be attributed to the limitation
in the simulations of pressure, wind, clouds, temperature, precipitation,
ocean currents, sea ice, permafrost, etc.
- As previously highlighted, while some climate models have included some low-impact assessment of various solar effects, it is unclear that they have accounted for any secondary amplification processes, e.g. sunspots triggering solar flares affecting the solar wind leading to changes in cosmic ray effects on cloud formation.
Today, most models simulate atmospheric flow using mathematical equations based on mass distribution as a function of wind velocity. These equations then have to be mapped onto a spherical grid of cells that only supports a limited modelling of the total atmospheric depth and layer composition. Equally, the flow equations have to be subject to modification on a scale below that of the cell grid, such that it might compensate for turbulence, latent heat, cloud formation and dynamic heating linked to solar and infrared radiation, which interact with atmospheric gases, aerosols, and clouds. In addition, accurate simulation of ocean currents may be more important than the atmosphere when it comes to the transport of global energy.
Note: Many of the most sophisticated climate models includes what is called the Ocean General Circulation Model . This sub-model simulates the circulation of the oceans and a dynamic thermal reservoir through which energy is exchanged with the atmosphere. This interaction is generally assumed to dominate the evolution of the climate system as it is believed to regulate heat, moisture, and momentum exchanges between the ocean and atmosphere.
Land surface also regulates heat based on soil moisture and vegetation type, although possibly not as important as the mechanisms taking place in the atmosphere and oceans, clearly any inaccuracy might then also affect the accuracy of any climate prediction being extended to the end of the 21st century. Of course, it is known that ice sheets also have an important role in the evolution of the climate system as their size may affect the total energy within the climate system as a whole due to reflected energy and the latent heat lost as water changes from a liquid to solid. However, it appears to be generally accepted that the climate system has to be dominated by solar processes, both direct and indirect, which regulate the input and output of energy. This includes not only the most obvious effects of Earth’s orbital eccentricity, axial tilt and precession, but other possibly more indirect mechanisms, e.g. solar winds, cosmic ray and maybe even magnetic pole reversal.
What else might be important to this general assessment?
As a macro-level description, Earth’s climate transports the energy received at the tropics to higher latitudes. However, it is again highlighted that some basic level of global thermal equilibrium requires inbound solar radiation in the visible spectrum to be lost to the infrared spectrum, which is then radiated back into space. Of course, in the context of climate change in the long history of Earth, the equilibrium of input and output energy has never been complete stable. However, we might again begin to appreciate the full scope of complexity that we are expecting these climate models to be able to predict. If we accept that climate models are inherently limited in their ability to represent this complexity, we might come to a more realistic assessment of their accuracy and the degree to which climate policy should depend on their output.
How might we then assess the importance of today’s climate model?
In general, models have to simplify the complexity of the real-world for the reasons outlined. However, we might realise that for any climate model to have value it has to attempt to represent this complexity, otherwise its results will be too limited. While models can be statistical or dynamic, climate models are generally dynamic in their computation in order to simulate the heat flow and turbulence in the atmosphere and oceans. As outlined, most climate models are based on a limited 3D cell grid system, where the initial state is established for some point in time and then allowed to evolve in time based on the equations that are constrained by certain boundary conditions. For example, atmosphere-ocean circulation models might be comprised of 7 basic mathematical equations with 7 basic variables that help simulate the state of the atmosphere over time. This model is assumed to simulate atmospheric motions and processes that conform to the laws of physics, i.e.
- laws of thermodynamics
- conservation of mass
- conservation of energy
- conservation of momentum
- kinetic theory of gases.
The equations involved have to be sequentially solved for each cell across the entire 3D grid space for each time step, such that a change in one cell can be mapped onto its neighbour cells according to the laws of physics and the direction of time. However, as alluded to, there are processes, which do not have precise formulation, i.e. they are too chaotic and may only be represented by some statistical approximation. Equally, some mechanisms are known to fall outside the resolution of the model, either in terms of the size of the cell or its temporal granularity. In either case, the climate model may have to resort to what is called ‘ parameterization ’ that might also be called an educated guess. Of course, any computation that goes ‘off-grid’ might be seen as a potential source of error, which has the potential to be extrapolated into ever-larger discrepancies as the simulation is extended in time.
How is the initial state of the climate model established?
While we might assume that this only requires a reasonable assessment of the current state-of-play of the Earth’s climate, even this process might be subject to complexity and debate. First, it is debated as to whether there is enough reliable climate data to establish the initial conditions from which the climate model might proceed. For while some geographies in the developed world may have sufficient climate data, this does not necessarily extend to all geographies in the developing world or account for the fact that 70% of planet Earth is cover by oceans of variable depth and salinity. While this problem is being addressed by including satellite information that does have global coverage, this data often appears to deviate from earlier ground-based measurements.
Note: There is now some controversy over the attempts to reconcile the differences between satellite data and historic climate data, especially when there are accusations that this process is deliberately manipulating the results to best support the case for global warming. In part, this accusation might be seen as the basis for the two views of climate change previously outlined, but repeated above.
Of course, it might also be highlighted that any uncertainty in the initial state of the model will only be amplified as the computations flow though the cell structure in both space and time, such that it can lead to unacceptable inaccuracies in as little as 4-days, let alone 50-years or more. Again, such statements are not made to belittle or undermine the research and development going into climate models, only highlighting that caution should prevail over certainty of any predictions. Unfortunately, there often appears to be little discussion of these issues within the confines of the 95% consensus and even less by the IPCC concerning the limitations in all its climate models.
So what about IPCC predictions that have suggested a global temperature rise between 1-6oC by 2100?
Such predictions are usually based on a statistical average of multiple runs of the models with slightly different initial conditions. Of course, as we might expect, the more times the model is run and over ever-longer timescales, the spread of the results invariably gets wider. However, the IPCC’s 2007 position appears to be based on the following stated position:
“Climate models are based on well-established physical principles and have been demonstrated to reproduce observed features of recent climate, and past climate changes. There is considerable confidence that Atmosphere-Ocean General Circulation Models provide credible quantitative estimates of future climate change, particularly at continental and larger scales”
However, while this assessment carries no weight of authority, the IPCC’s confidence appears incongruous to all the issues outlined through this discussion. For this reason, Appendix-A provides linked references to the wider debate surrounding climate change.