Session: Glacial Climates (LGM, Last deglaciation, Ice sheet uncertainties, Glacial-interglacial cycles)
Author: Paul Valdes / firstname.lastname@example.org / University of Bristol, UK
Co-author: Ruza Ivanovic, Leeds University, UK;
Lauren Gregoire, Leeds University, UK;
Peter Hopcroft, University of Bristol, UK;
The latest climate models for CMIP6/PMIP4 will generally be higher resolution than previous generations and increased resolution will often be suggested as the cause for any resulting changes. However, in the majority of cases many other aspects of the model will also have been changed and it can be difficult to rigorously attribute the changes to a specific increase of resolution.
We have therefore investigated the role of resolution in simulating past climate change through a series of simulation using one particular model, HadAM3. The advantage of using this model is that high resolution versions of this model were extensively used for weather forecasting so the model physics is optimised for both low (climate) and high (weather) resolution versions.
We have performed a series of atmosphere-only LGM and mid-Holocene PMIP simulations with a range of resolutions from those typical of PMIP2 and PMIP3 models (3.75 x 2.5 degrees) to much higher resolution (maximum resolution of 0.56 x 0.38) which is relatively high resolution even for CMIP6.
The results show that increased resolution improves the simulation of modern precipitation patterns by better representing the detailed orographic and coastal processes, but that palaeo changes in large scale precipitation are relatively robust and insensitive to resolution scale. The exception to this is at the LGM, when the flow direction changes (causing a shift in rain shadows etc.) and when land area expands with the reduced sea level.
The effects of resolution on the changes in extreme events will also be discussed. Furthermore, we will present analyses of the simulations with respect to “the wet gets wetter” paradigm. Initial work suggests that this is not especially effective at explaining the modelled changes.