Testing ice sheet models and modelled estimates of Earth's climate sensitivity using Miocene palaeoclimate data

Climate models use computational techniques to mimic real physical and chemical processes in the climate system in order to predict future climate change. Such models have been used to quantify how sensitive Earth's climate is to atmospheric carbon dioxide levels. Knowing this 'climate sensitivity' is essential for politicians to set goals for future CO2 emissions that will keep Earth's climate within 'safe' limits. Until recently, the estimates for this climate sensitivity have been based on models that look only at the short-term (e.g., years-decades) effects of increasing CO2 and rising temperatures. They do not include other longer-term effects, such as melting ice sheets or changing global vegetation cover. An example of such a feedback is the melting of the Greenland ice-sheet, which in addition to causing rising sea levels will also cause further regional warming. The problem with ignoring such components of the climate system is that there are large uncertainties regarding the timescales on which they operate. Recently there have therefore been suggestions from the academic community that estimates of Earth's sensitivity to atmospheric CO2 levels should include all feedbacks within the climate system: both those that operate fast and those that operate more slowly. This more comprehensive view of the relationship between Earth's climate and pCO2 is termed 'Earth System Sensitivity'. The best way to estimate Earth System Sensitivity is to use intervals in the geological past when we know that CO2 and temperature were different to today. However, thus far this approach has led to very different estimates, largely due to uncertainties in the levels of atmospheric pCO2 reconstructed for these intervals. In the first part of the proposed work we will generate new pCO2 records using a relatively new method of estimating pCO2 (using the ratio of boron isotopes within marine planktonic microfossils) that has recently been refined. We will also use several methods for reconstructing temperatures from the same interval in the past, so that we can calculate Earth System Sensitivity. The second part of the proposed work is to use data to test a computational ice sheet model. Ice sheets are dynamically complex, and sophisticated models are required in order to predict their response to changing climate and therefore their effect on global sea level. The models need to be tested if we are to have confidence in their predictions. The long timescales involved with ice sheet dynamics means that we cannot test ice sheet models with real-time observational data. The best way to test ice sheet models is to use them to predict ice sheet changes for a period in the geological past where we have good records of Earth's temperature gradients, and compare the model results with well-constrained records of ice sheet growth for the same interval. In the proposed work we will use the Middle Miocene Climate Transition to test an ice sheet model. We know that CO2 decreased, climate cooled, and the Antarctic ice sheet expanded at this time (~14 million years ago). In this work we will obtain new, accurate records of pCO2 and temperature to drive our models. The models will then predict ice sheet changes, which we can compare to an existing record of ice sheet growth across the climate transition. If the model and the data are in good agreement then our confidence in the ice sheet model will be increased. If the model and the data are not in good agreement, then this work could lead to the identification of certain parameters within the model that may need to be adjusted. This may then lead to improved future predictions of ice sheet, and hence sea level change.