To better understand how accurately the models predict future climate might we can look at how the same models show what the climate was like in the past. These simulations can be compared with known past climate records, and if there is a good match we know that there is a good chance the same modesl can predict future climate.


The aim of Experiment 1 in this particular study was to place a paleo-constraint on climate sensitivity. Sensitivity was 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 paleo-forcings to the models, we are testing if the models and their corresponding sensitivity are realistic. The general circulation model results are compared to paleo-observations. All the models are run with a range of initial conditions, parameters and forcings. By providing a framework for the evaluation of the climate models, we are testing if they are able to simulate climates that were different from today. This will improve our confidence in the models projections for future climates.

The period of focus 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 in time with a relatively stable climate was the mid-Holocene, when the climate was stable for a period of about 2000 years and the forcing on the climate is well known. The 6kyBP climate is reasonably well known through paleo-observations.

The model used is HadSM3, a state-of-the-art climate model from the Hadley Centre for Climate Prediction and Research. An Atmospheric General Circulation Model is coupled to a slab ocean. Mid-Holocene boundary conditions are applied to the model, i.e. the orbital configuration is altered to represent that of 6000 years ago. This redistributes the latitudinal and seasonal distribution of incoming solar radiation at the top of the atmosphere in the model. The methane concentrations are also lowered. Experiments 1 and 2 have somewhat different boundary conditions in the mid-Holocene phase, as described below. The model participants downloaded consisted of 4 different phases, each of 15 model years and with a unique set of initial conditions.

The four phases were:

  • Calibration step (phase 1)

Phase 1 is the calibration phase of the experiment. In this phase, the temperature of the surface of the ocean is artificially held constant. The movement, or flux, of heat, in or out of the ocean that is required to keep the ocean at a constant temperature is calculated. This is an easy solution to having a very simple ocean in the model, which cannot actually store heat in the way that a real, deep, complex ocean can. The dates given to this phase are 1810-1825.

    • Pre-industrial CO2 step (phase 2)

This is the control phase. This involves running the model for 15 years with the levels of CO2 in the model atmosphere kept constant at pre-industrial levels, 282ppm. Unlike phase 1, here the temperature of the ocean surface is allowed to vary, according to how much energy the ocean receives and emits. However, it is safe to assume that the amount of heat flowing into the oceans is the same as in phase 1, so the heat fluxes calculated in phase 1 are applied. Unless the atmosphere starts doing something very different, and the energy balance at the top of the atmosphere is changed, the temperature of the whole atmosphere should therefore stay the same. If this is the case, the globally averaged surface temperature should also be approximately constant and not change substantially from year to year or drift off to a very different temperature, and we say that the model is stable. The dates given to this phase are 1825-1840.

    • Mid-Holocene step (phase 3)

In this phase the model is forced with pre-historic conditions. The orbital configuration is altered for that of 6000 years before present. I.e. the tilt of the Earth is increased by 0.5 degree and the precession of the equinoxes (the position of the Earth in its orbit around the Sun at the equinoxes) and eccentricity (the shape of the Earth’s orbit around the Sun) of the Earth are changed in accordance with the so-called Milankovitch cycles. Additionally, the methane concentrations are lowered by about 15% compared to the pre-industrial. The model is again run for 15 years and is given the same model dates as in the control phase of 1825-1840.

  • Double CO2 step (phase 4)

In this phase the levels of greenhouse gases are doubled and the model is run for a further 15 years. In a good model, the atmosphere should adjust to this change in forcing and eventually settle in a new stable, equilibrium state (which may be the same, warmer or cooler). The dates given to this phase are 2050-2065.

By comparing the single and doubled CO2 steps, we can calculate the climate sensitivity of the models – this is the difference between the globally averaged surface temperature in the model with pre-industrial CO2 and in that with doubled CO2. This is a useful indicator of how a climate model behaves, although it is slightly artificial, as of course carbon dioxide values in the atmosphere do not remain constant for 15 years, but change continuously.

The difference in the model results between the mid-Holocene and control models are compared to the paleo-observations available. A range of paleo-observations is used to test the models; pollen data, lake level data from lake sediment cores, macro fossil data and temperature data deduced from foraminifera, molluscs, diatoms, driftwood, whale fossils and whale bones. The observations have shown that the global monsoon systems were stronger during the mid-Holocene. The most notable feature of the mid-Holocene climate is the increase in the moisture budget across northern Africa. There were extensive wetlands and lakes across the current hyper-arid Sahara. In addition to looking for these spatial features in the model results, the resulting change in seasonality from the altered orbital configuration can be seen in the global mean timeseries of temperature and precipitation rate.

There are two parts to this particular paleo-experiment; firstly the models will be distributed with the four phases as described above. In the second part of the experient these models will be distributed again, only there is a change to the mid-Holocene phase. Phase 3 has altered orbital configuration and reduced methane concentrations as before, now with the additional change of the inclusion of an ice sheet in Eastern North America and the Hudson Bay is expanded. These local changes are hopefully going to improve the simulated climate in Eastern North America, as climate models to date have had difficulties capturing the locally cooler climate of the mid-Holocene. With this second experiment the motivation is how well can we simulate past climates and the models are benchmarked against a set of robust features of the mid-Holocene climate as seen in the geological evidence.

Model Used


Lead Scientist

Helene Muri