Carbon budgets - Environmental Audit Committee Contents

Supplementary memorandum submitted by the Met Office

  During the oral evidence session on 23 June 2009, the Committee asked the Met Office to provide a supplementary diagram to demonstrate the Committee on Climate Change's (CCC) cautionary approach to climate sensitivity distribution. Also included in this note is clarification on the issue of the inclusion of coupled/non-coupled feedbacks in the models used by the CCC, and a short summary of what we have done and what we intend to do with respect to further use of a complex 3-dimensional earth system model and the benefits of further work.

  Following the oral evidence session, Mr Aubrey Meyer wrote to Dr Jason Lowe at the Met Office requesting clarification on some of the evidence presented to the Committee. Mr Meyer copied his questions to the members of the EAC and we have, for completeness, included a summary of our response at Annex A.


  The models used by the Committee on Climate Change did include the feedback of climate change on the carbon cycle—it was a coupled model as defined in section 10.4.1 of the IPCC AR4 study.

  The CCC chose a particular climate sensitivity estimate when deciding on emissions targets. The climate sensitivity uncertainty estimate chosen was:

    (a) derived from a combination of the most sophisticated model available combined with a wide range of observations;

    (b) a precautionary choice in that it provides a lower probability of staying below 2C than other alternatives climate sensitivity uncertainty estimates.

  Although the CCC climate change scenarios were produced using a simple earth system model its performance was validated against a more complex model.

  Since these scenarios have been produced the simple model has been further evaluated against a more complex model and proven to skilful.


  Estimates of risk of exceeding a given temperature (eg 2°C), by a certain year, for a given emissions pathway, are a consequence of uncertainty in the climate projections.

  The uncertainty in the simple earth system model used to generate the global climate projections for the CCC was expressed in turns of three key parameters. At present, we do not know the precise value of these parameters, but we do have information on their ranges:

  Climate sensitivity: a measure of how much the average global temperature will eventually rise if atmospheric CO2 concentrations were to double.

  Ocean diffusivity: a measure of how effectively heat is mixed between the upper ocean and the deep ocean. This has a significant impact in the rate of surface warming.

  Carbon cycle-climate feedback: a measure of how much climate change can alter the natural flows of carbon between the atmosphere, the land and the ocean.

  For a given stabilisation concentration of greenhouse gases in the atmosphere, the climate sensitivity determines the stabilisation temperature—although this temperature may take several decades or longer to be reached. Climate sensitivity and ocean diffusivity together determine how long it takes the temperature to reach stabilisation once the concentration of greenhouse gases and aerosols has stablised. For a given pathway of future emissions, all three parameters determine the evolution of the greenhouse gas concentration over time.

  The modeling approach used by CCC can be described as "coupled" in that it includes the feedback of climate change onto the carbon cycle.

  This is the definition of "coupled" as used in section 10.4.1 of the IPCC AR4 WG1 and the C4MIP study. .


  Figure 1 below presents the probability of the equilibrium warming exceeding 2°C (y-axis) for a range of stabilisation greenhouse gas concentrations expressed as equivalent carbon dioxide concentrations (x-axis). Each line on the plot is for a different estimate of the uncertainty in climate sensitivity. Eleven different estimates of climate sensitivity are shown, giving eleven different estimates of the risk of exceeding 2°C. Using any of the lines on this figure it is possible to quote the stabilization equivalent CO2 concentration that gives a 50/50 chance of exceeding 2°C. As each climate sensitivity uncertainty distribution leads to a different result, care must be taken when interpreting the CCC results in terms of probability.

Figure 1. The probability of eventually exceeding 2°C for a range of different climate sensitivity estimates. This figure is adapted from Fig 28.5 Meinshausen et al. 2006.[8] For clarity red squares have been superimposed on the Murphy et al[9] result.

  The 50/50 result in the CCC simulations is obtained using the Murphy et al. climate sensitivity distribution. The Murphy et al. climate sensitivity uncertainty distribution was chosen for two reasons. First, it combines our most sophisticated type of model (complex 3-dimensional models) with a wide range of observations. Second, for stabilisation around 450ppm, it provides a lower probability of staying below 2°C than alternative estimates of climate sensitivity uncertainty, ie it is a precautionary choice. Above around 430ppm, it is clear that the Murphy et al. distribution gives the highest chance of exceeding the temperature target. Although this discussion is based on stabilisation temperature, a similar argument can be developed for the temperatures in the CCC scenarios at 2100.


What was done at the time of producing the CCC simulations?

  A simple climate model (with coupled climate-carbon cycle feedback) was set up and demonstrated to reproduce the response of more complex earth system models for scenarios of increasing greenhouse gas concentration. Having demonstrated model skill, a range of simulations was made using the simple climate model. Several alternative sets of emissions scenarios were used and the results also contain information on uncertainty/risk.

What has been done since?

  As part of the Met Office integrated climate programme (ICP), an idealised simulation with a complex three dimensional earth system model was carried out. This was idealised in the sense that emissions reductions were very fast and only carbon dioxide was treated. The simple climate model used in the CCC simulations was then compared against this new complex model simulation. The good agreement showed that the simple model has skill for scenarios in which emissions are reduced significantly (as well as that already demonstrated for cases when concentrations of greenhouse gases continue to rise rapidly). This increased our confidence in the suitability of the simple modeling approach.

What should still be done?

  We recommend that a small number of simulations be set up of the CCC emission scenarios using the complex three-dimensional earth system model. This has several purposes. First, it provides a further check of the simple model for the precise multi-gas scenarios used in the CCC work. Second, it provides extra information on climate variability and any sudden surprises, for instance, changes in the ocean circulation. Third, it provides regional information, so that it is possible to understand which regions are warming most rapidly and to examine the local projected changes in carbon cycle feedback. As a by-product, this approach may produce information useful to the adaptation sub-committee of the CCC.

  At present, we do not recommend repeating the entire simple model experiment set with complex three-dimensional models to estimate the risk. Such a project would be comparable in scale to the recent UKCP09 analysis and is unlikely to provide significantly better global risk estimates. However, this position should be reviewed as climate science and/or model understanding develops, and if the requirement for global adaptation information increases.

Annex A


  Mr Meyer's questions centered on climate model results shown in the IPCC AR4 WG1 report. The specific questions and our responses are reproduced below and we have included some background information, also supplied to Mr Meyer, to facilitate a broader contextual understanding.

  Question One: "As I pointed out in the written evidence from GCI that you said that you looked at, my reading of the figure from IPCC AR4 Chapter 10 is that with `coupling' introduced, the image in fact shows the extent of the need to reduce the full-term emissions contraction-event associated with a given reference curve for concentrations. Can you confirm that that is your understanding please?".

  Met Office Response: The graph taken from fig 10.21 of the IPCC AR4 WG1 report shows the results of three models. The Hadley Centre curve shows a simple model set up to replicate the more complex Hadley Centre model used in C4MIP. The simple model was then used to study the emissions that lead to a stabilisation level for CO2 of 450ppm for a single pathway. For this particular pathway, and only considering CO2, the curve does show when coupling of climate to the carbon cycle is included, as it was by the CCC, emission levels would have to reduce further to achieve a given stabilisation level of CO2 concentrations. However, given that all the models in C4MIP and fig 10.21 are considered credible we believe the appropriate scientific approach is to include information from the full range of available models not just the results of a single (worst case) model. To that extent the Hadley SM curve on the graph is not, by itself, a good indication of the need to reduce emissions targets further than was indicated in the CCC simulations.

  Question Two: "In the example graphic taken from the IPCC AR4 in what is tagged as the C4 MIP 'Hadley SM' model with runs for 450 ppmv it shows very clearly that what in the IPCC image is called:

    "uncoupled" for 450 ppm requires a 50% cut in carbon emissions globally by 2050 and "coupled" for 450 ppmv requires an 80% cut in carbon emissions globally by 2050.

  Can you confirm that that is your understanding of this image please?"

  Met Office Response: As explained above, fig 10.21 does not show results from C4MIP, rather it shows outputs from three simpler climate models which also include interactions between the carbon cycle and climate. Furthermore, using the results of a single mode for a pathway of a particular shape and only considering CO2 to make general conclusions about global emissions reduction targets for a single year, 2050, is not appropriate. It is also important when discussing percentage emission reductions by 2050 to state the year to which they are relative. The CCC expressed their recommendations for UK emissions relative to 1990.

  Question Three: "You went on to say, "The precise values we use to work out the magnitude of the coupling comes from elsewhere in IPCC and from a study referred to as a C4MIP study, which to date is the most comprehensive analysis of that particular type of feedback onto the carbon cycle". The runs in question and highlighted in the attached graphic from the IPCC AR4 bear the tag "Hadley SM". Can you as a member of the UKMO Hadley Centre please explain to me what `elsewhere in the IPCC' refers to?"

  Met Office Response: Chapter 7 of the AR4 WG1 report summarises the results of the C4MIP project while table 7.4 presents the range of coupling factors for all 11 of the models used. C4MIP is mentioned extensively in section 10.4.1 of the IPCC AR4 WG1 report (the section from which you have taken fig 10.21).


  The C4MIP project, summarised in Chapter 7 of the IPCC AR4 WG1 report, set out to understand the importance of coupling the carbon cycle to climate change and its impact on the evolution of atmospheric concentrations of CO2. Eleven models that explicitly represented the interaction between climate and the carbon cycle were used in the project.

  Each model was driven by a single emissions scenario—SRES A2—and was run twice, once with climate coupled to the carbon cycle and once without. Each model simulation produced an evolving estimate of the total atmospheric concentration of CO2. By comparing the coupled to the uncoupled simulations, it was possible to gain an indication of the importance of feedback between climate and the carbon cycle.

  All of the models run in C4MIP demonstrated that coupling the climate to the carbon cycle is important and that by 2100 climate change leads to the biosphere being less able to absorb CO2. A key result from the study was the significant variation across the models in the size of this effect, demonstrating significant uncertainty in representing the climate-carbon cycle feedback. Although C4MIP found the Hadley Centre model showed the strongest feedback effect, the other ten models are also credible and their results cannot therefore be ruled out.

  This parity between the eleven models meant it was important, in the work carried out for the Committee on Climate Change, that the results from all C4MIP models were used to select the strength of interaction between the climate and carbon cycle. Several different future emissions scenarios were then run through a climate model (which has a treatment of the carbon cycle), in each case estimating uncertainty in temperature and greenhouse gas concentrations. The full uncertainty range was due in part to the range of climate-carbon cycle feedbacks in C4MIP. For each emissions scenario an output from the simulations was a probability distribution showing how likely different amounts of 21st century warming will be. The Committee on Climate Change then selected the emissions scenario that showed a 50% chance of limiting warming to approximately 2C above pre-industrial levels at 2100, as well as reducing the risk of a 4C rise to very low levels.

  Before the simulations for the Committee on Climate Change, the Hadley Centre and two other modeling centres had already carried out studies specifically to evaluate the impact of climate change on carbon cycle feedbacks, and therefore the emissions required to reach atmospheric stabilisation at a number of concentration levels. These are shown in fig 10.21 in the IPCC AR4 WG1 report. Emissions pathways were based on CO2 only, unlike the more realistic Committee on Climate Change simulations that included aerosols and other Kyoto gases. Also relevant is that the Hadley simple model simulations in fig 10.21 were constrained so that atmospheric CO2 followed a particular pathway to 450ppm. In the Committee on Climate Change simulations, the atmospheric concentrations were not constrained in the same way. Instead, the emphasis was placed on the pathway of global temperature rise. It is important to recognise the limitation of the experiments reported in fig 10.21—which were largely to gain an understanding of the nature of the coupling between climate and carbon cycle rather than to provide definitive guidance on emissions reduction targets.

  The models used by the Committee on Climate Change did include a coupling between climate and the carbon cycle and took full account of the `coupled' model research presented in the AR4 WG1 report, the C4MIP study and related research.

July 2009

8   Meinshausen, M et al. Multi-gas emissions pathways to meet climate target in Avoiding Dangerous Climate Chang, ch 28, edited by Schellnhuber, Cramer, Nakicenovic, Wigley & Yohe. Back

9   Murphy, J M et al. Nature 430 768¸772 (2004). Back

previous page contents

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2010
Prepared 14 January 2010