Quantifying and understanding natural climate variability on long time scales is key to distinguishing natural and anthropogenic variability during the 20th and 21st century, and to predict the future climate. Climate variability in the pre-instrumental period can be estimated either from climate proxy data, e.g. tree rings, ice cores or stalagmites, or from numerical simulations. Palaeoclimate simulations have been performed so far mainly by prescribing climate forcings such as solar radiation and atmospheric composition. However, as a consequence of internal climate variability, the temporal evolution of the climate states is not completely determined by the forcings. In principle this problem can be addressed by combining empirical information from proxy data with numerical simulations (for an overview see Widmann et al. 2010). This approach is used operationally in meteorology and is known as data assimilation (DA), but adapting it to palaeoclimatic applications is challenging.
This project aims at improving climate reconstructions for the last millennium by refining a DA approach that has been implemented at the University of Birmingham in collaboration with the Max Planck Institute for Meteorology (MPI-Met) in Hamburg. We use ensemble simulations with the MPI-ESM General Circulation Model and select ensemble members that are closest to palaeoclimate proxy data to obtain a simulation that is both consistent with the model physics and with the empirical knowledge. This approach has been already successful with Earth System Models of Intermediate Complexity (Goosse et al. 2006). The increase in computing power now allows using it with complex GCMs.
The ensemble selection DA has been implemented using the MPI-MET Earth system model MPI-ESM assimilating Northern Hemisphere continental mean temperatures. These temperatures were taken from a recently published set of continental temperature reconstructions (PAGES 2K 2013). We have shown that although in general the assimilation simulations follow these target temperatures well, there is a lack of information propagation to smaller spatial scales (Matsikaris et al. 2015a; Matsikaris et al. 2015b) and that one of the reasons for this is that continental temperatures are not ideal for constraining the atmospheric circulation and thus regional temperatures (Matsikaris et al. 2015b).
Gaining information on variables that are not directly assimilated is one of the main reasons for performing DA, and thus our results show that there is a clear need to further develop the DA methods. We will use statistical analysis (e.g. Maximum Covariance Analysis) of gridded instrumental records and of long climate simulations to determine the optimal locations of temperatures to be used as input for DA. These networks will then be systematically tested in DA simulations. To distinguish between different sources of errors we will assimilate instrumental observations for the last 100 years, as well as proxy-based reconstructions extending back several hundred years.
The new supercomputer at MPI will allow us to perform DA with a higher resolution version of the MPI-ESM model (1.8 deg instead of 3.7 deg in previous simulations), which can be expected to lead to a better propagation of information from large to small scales.
Training and Skills
The project gives the chance to work on cutting-edge problems in palaeoclimate modelling using supercomputers, state-of-the-art climate models, and advanced statistical methods. Training in climate modelling, data assimilation and statistical methods will be provided at the University of Birmingham. The student has also the opportunity to attend related lectures on the MSc program ‘Applied Meteorology and Climatology’. Co-supervision by and visits to MPI-Met will strongly contribute to training in climate modelling. The student will also strongly benefit from strong links of both supervisors to international projects such as PAGES2k and PMIP.
CENTA students will benefit from 45 days training throughout their PhD including a 10 day placement. In the first year, students will be trained as a single cohort on environmental science, research methods and core skills. Throughout the PhD, training will progress from core skills sets to master classes specific to the student's projects and themes.
Year 1: Statistical analysis of observed and simulated temperature and pressure to identify optimal locations across the Northern and Southern Hemisphere for temperatures to constrain regional climates.
Learn how to run MPI-ESM in DA mode.
Year 2: Assimilate observed temperatures from optimised network. Compare simulated regional temperatures in assimilation run with regional observations. Compare skill with expected skill from statistical analysis of coupled regional and large-scale patterns.
Year 3: Assimilate proxy-based temperature reconstructions from optimised network for the last few hundred years. Compare simulted regional temperatures for the 20th century with regional observations. Analyse climate variability in the full multi-century simulation and compare with other available simulations.
Partners and collaboration (including CASE)
The School of Geography, Earth and Environmental Sciences at the University of Birmingham includes a strong group on meteorology and climate science, with expertise in regional and global climate modelling, palaeoclimate, dynamical meteorology, statistical climatology, and downscaling. The MPI-Met is one of the world-leading climate modelling centers.
Applicants should have a background in a related field such as climatology, geosciences, or physics or mathematics. Sound mathematical and statistical skills, as well as programming experience are essential. Working experience with UNIX, FORTRAN, and climate modelling would be beneficial.
For further details please contact M. Widmann (firstname.lastname@example.org)