About the Project
We are seeking for an enthusiastic, PhD student to work within the international project “Scale-aware sea-ice project – SASIP” (https://sasip-climate.github.io/) funded and supported by the Schmidt Futures foundation (https://schmidtfutures.com/). SASIP aims at developing a truly innovative, scale-aware continuum sea ice model for climate research; one that faithfully represents sea ice dynamics and thermodynamics and that is physically sound, data-adaptive, highly parallelized and computationally efficient. The University of Reading leads SASIP work package 4 (WP4) devoted to complement the model development efforts of the other WPs by providing data constrained sea ice analyses using data assimilation (DA) and machine learning (ML). Data assimilation is the science of combining observations of an environmental system with numerical models of that system in a statistically and dynamically sound way. The postholder will work on Task 3 of WP4 (Task 4.3) ML and hybrid DA-ML for parameterization design and parameter estimation.
In particular, the student will focus on the thermodynamics of the sea-ice and on the capabilities of advanced hybrid DA-ML methods to emulate these processes (Brajard et al., 2020) and/or to generate data-driven parametrization of unresolved scale phenomena (Brajard et al., 2021). The student will use a 1-dimensional sea-ice model that incorporates an advanced description of the thermodynamic processes in the sea-ice. Nevertheless, these models still show errors, in particular due to the complex representation of the surface albedo. The student will investigate combined DA-ML methods whereby DA is used to correct for model error while ML to improve the model. Ultimately the PhD work is expected to contribute to devising appropriate parametrizations, or to incorporate a data-driven emulator, of the thermodynamics processes in the new version of the sea-ice model neXtSIM (Rampal et al., 2016) under development in SASIP.
This recruitment is one of the three at University of Reading within SASIP, the other two being two Postdocs working on SASIP Task 4.1 (DA for sea-ice model using discontinuous Galerkin methods) and Task 4.2 (DA for parameter estimation). The PhD student and the two Postdocs will interact very closely and constitute the core of the SAPIP working group at University of Reading.
The PhD research will be conducted together with other members of SASIP, in particular Laurent Bertino, Julien Brajard and Einar Olason at NERSC (Bergen, Norway) and Marc Bocquet at ENPC (Paris, France) and the student will have the opportunity to spend research visits to their premises.
At the University of Reading, the postholder will be based at the Dept of Meteorology and will be part of the Data Assimilation Research Centre (DARC, https://research.reading.ac.uk/metdarc/) enjoying an international, diverse and stimulating research environment. The postholder will be also affiliated at the National Centre for Earth Observation (NCEO, https://www.nceo.ac.uk/) a distributed centre of over 100 scientists in the UK. Data assimilation is used throughout the NCEO and the role of the DA group at the University of Reading is to support this work by the development of new DA techniques. The postholder will thus also benefit from the ample offer of training offered locally at University of Reading and nationally within NCEO. There will also be the possibility to teach on the data assimilation training courses that NCEO offers together with the ECMWF.
Eligibility requirements:- We expect you to have either a 1st or upper 2nd class degree, or a master's with Distinction or Merit, in environmental science, applied mathematics, physics or computer sciences.
This opportunity is open to candidates worldwide but only covers Tuition Fees at the UKRI rate. A successful international candidate would need to fund or secure sponsorship for the difference in the fees (approximately £16,330/year).
• Brajard, J., A. Carrassi, M. Bocquet, and L. Bertino, 2021. Combining data assimilation and machine learning to infer unresolved scale parametrization. Phil. Trans. of the Roy. Soc. A, 379 (2194)
•Brajard, J., Carrassi, A., Bocquet, M. and Bertino, L., 2020. Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: a case study with the Lorenz 96 model. Journal of Computational Science, 44, p.101171.
•Pierre Rampal, Sylvain Bouillon, Einar Ólason, and Mathieu Morlighem, 2016. nextsim: a new lagrangian sea ice model. Cryosphere, 10(3). doi: 10.5194/tc-10-1055-2016
Based on your current searches we recommend the following search filters.
Based on your current search criteria we thought you might be interested in these.