There is some scepticism about the performance of the general circulatrion models of the climate and their predictions of future climate changes. This page has a description of the performance of the Japanese model and illustrates some of the reasons for scepticism.

One Model Performance

A cause for concern about the output of GCMs is the jagged nature of the graphs of future temperature changes against time. Normally the output of a computer programme is smoother than any actual measurements because the measurements are subject to various uncertainties and errors that are not present in the programme. It does seem that there is a built-in randomness. One consequence is that successive runs of the same climate model can yield very different values for warming trends. These trends may vary by an order of magnitude or more, and even their sign may vary. Modellers, therefore, carry out several runs and then publish the ‘model ensemble mean,’ the arithmetic average of the individual trend values generated by the several runs. For example, results from the Japanese MRI model for five runs [Santer et al., Int. J. Climatol. 2008, doi:1002/joc.1756] for the tropical region between 20°S and 20°N are shown in Figure 12.10. The individual trends range from 0.042 to 0.371 K/decade with an ensemble mean trend of 0.28 K/decade.

 

 

Five runs of the Japanese MRI model and the ensemble mean

The actual temperature measurements over the same period are known as are the atmospheric concentrations of CO2. The temperature anomaly is expected to be connected with the logarithm of the CO2 concentration (Beer-Lambert Law). Plotted against each other, the temperature anomaly and the natural logarithm of the CO2 concentration give a good straight line with the equation:

ΔT(CO2) = 1.38 ln (CO2/ppmv) – 8.13

Used to produce the temperature anomaly at each point of the mean CO2 concentration for the years concerned, this equation produces a plot of temperature anomaly against time with the equation:

ΔT(CO2) = 0.0059 × year – 11.83

The best linear equation linking the temperature anomaly with time is:

ΔT = 0.0059 × year – 11.84

The excellent agreement between the two equations indicates that the temperature anomaly is highly correlated with the CO2 concentration. So, at 0.059°C per decade, this value for the trend is only 21% of that given by the ensemble mean of the computer runs.