AGU 2010 Abstract
Plio-QUMP: Quantifying Uncertainty in Model Predictions for the Pliocene James Pope1*, Matthew Collins2,3, Alan Haywood1, Harry Dowsett4, Daniel Hill5 & Daniel Lunt6 1. School of Earth & Environment, University of Leeds, Leeds, UK 2.
Met Office, Hadley Centre, Exeter, UK
3. School of Mathematics, University of Exeter, Exeter, UK 4. United States Geological Survey, Reston, Virginia, US 5. British Geological Survey, Keyworth, Nottingham, UK 6. BRIDGE, School of Geographical Sciences, University of Bristol, Bristol, UK *Contact:
[email protected] The mid-Piacenzian Warm Period is an interval of Earth history when global temperatures were sustained at 2 to 3°C above pre-industrial values, caused at least in part by higher levels of CO2 in the atmosphere (~380 to 425 ppmv). With a palaeo-geography almost identical to today, this interval provides an opportunity to examine the potential long-term effects of global warming through the use of modelling studies and data comparison. However, the uncertainty in model predictions of mid-Piacenzian climate has not been explored. The aim of Plio-QUMP (Quantifying Uncertainty in Model Predictions for the Pliocene) is to produce uncertainty estimates for model predictions in two stages. The first stage will create a series of ensembles based on Perturbed Physics in the UK Met Office Climate Model. The second stage of the project will explore the uncertainty in geological boundary conditions (e.g. initial trace gas concentrations, sea-surface temperatures, vegetation cover, orography, run-off, soils) some of which have been provided by the USGS Pliocene Research Interpretations and Synoptic Mapping (PRISM) Group. Here we show results from an initial perturbed physics ensemble using the Hadley Centre Coupled Climate Model Version 3 (HadCM3). The ensemble was created using a single climate model and perturbing the parameterisations of sub-grid scale processes (elements of the climate system that happen on a scale smaller than the resolution of the climate model). Three simulations where run, a Pliocene Control and a High Sensitivity and Low Sensitivity simulation. The high sensitivity simulation perturbs the settings of 33 parameters in the model so that in combination they produce the highest possible equilibrium climate sensitivity, with the opposite in place for the low sensitivity simulation. Anomaly plots for the sensitivity runs minus the control simulation shown here indicate that the High Sensitivity run when compared to geological proxy data improves the model against the data in
higher latitudes and in tropical oceans. However problems are encountered across some continental areas such as North America. The simulations will be further tested using vegetation and ice sheet models, to enable further comparison between the models and the data using as many available data sources as possible. Although the results from the high sensitivity simulation are more consistent with the direction the model needs to develop to address specific data-model discrepancies present, the low sensitivity simulation shows useful trends in continental areas. This indicates that a mixture of parameters as yet untested will probably prove to be a more effective version of the model compared to these extreme end members tested so far. This will be the development and future work of the project.