Milan Klöwer

klower milan image 2
2017
Approximate computing in weather and climate forecasting: Information content diagnostics for reduced precision
milan.kloewer@physics.ox.ac.uk

Academic Profile

Education

•BSc Physics of the Earth System, University of Kiel, 2010-13
•Marine Physics, University of Western Brittany, France, 2013-14 (year 1 of MSc)
•MSc Course on the Arctic Atmospheric Boundary Layer, University Centre in Svalbard, 2016
•MSc Climate Physics, University of Kiel, 2014-2017
•PhD student in Climate Physics, University of Oxford, 2017-today

Positions
•Post-doctoral research assistant, University of Oxford, 2017

Research Visits
•Alfred-Wegener-Institute (AWI), Bremerhaven, Germany, 2012 and 2013
•University of Oxford, UK, 2015

Oceanographic research cruises
•Meteor M130. Tropical Atlantic from Mindelo, Cabo Verde to Recife, Brazil, 2016

Conference organization
•Stumeta, 4-8 May 2016 in Kiel, Germany. Conference for meteorology students. stumeta2016.de

Current Research

I came to Oxford to explore the potential of reduced precision in weather and climate predictions. My DPhil project aims to understand the bit-wise information content in geophysical fluid simulations with respect to predictability and climate. I am keen to develop diagnostics to identify the necessary precision for future weather and climate models. This is based on the idea, that a majority of bits in a 64-bit double precision floating point number does not contain real information in the case of geophysical systems. Using double precision contradicts the uncertainty that arises from initial conditions, model formulation, coarse resolution, parametrizations and scenarios. An appropriate number of bits for floating point numbers has the potential to drastically reduce the necessary computation and communication between processors on large super computers and can be traded for a higher resolution and complexity of weather and climate models. Consequently, more skillful predictions for weather and climate are allowed.

Publications

Klöwer M, M Latif, H Ding, RJ Greatbatch and W Park, 2014. Atlantic meridional overturning circulation and the prediction of North Atlantic sea surface temperature, Earth and Planetary Science Letters, 406, [dx.doi.org/10.1016/j.epsl.2014.09.001]

Klöwer M, T Jung, G König-Langlo and T Semmler, 2013. Aspects of weather parameters at Neumayer station, Antarctica, and their representation in reanalysis and climate model data, Meteorologische Zeitschrift, 22(6), [dx.doi.org/10.1127/0941-2948/2013/0505]

Klöwer M, PD Düben and TN Palmer, 2019. Posits as an alternative to floats for weather and climate models, In Conference for Next Generation Arithmetic 2019 (CoNGA'19), ACM, 10.1145/3316279.3316281

Klöwer M, MF Jansen, M Claus, RJ Greatbatch and S Thomsen, 2018. Energy budget-based backscatter in a shallow water model of a double gyre basin, Ocean Modelling, 132, 10.1016/j.ocemod.2018.09.006

Contact information

Department of Physics

Personal web page

Orchid ID