March 2015
Message 86

[Periods|Index by:DateThreadSubjectAuthor|Date:PreviousNext|Thread:(Previous)(Next)|List Information]

[Met-jobs] 12-month post-doctoral position: assimilation of in-situ marine surface observations, impact on weather prediction systems

Cc Doerenbecher Alex <>, MAHFOUF Jean-françois <>
Date Mon, 30 Mar 2015 11:06:13 +0200 (CEST)

Atmospheric research scientist.
E-SURFMAR impact study (EUMETNET)

Title (topic) : Characterization of the importance for numerical weather prediction of in-situ surface observing systems in the North Atlantic ocean (CHIDOSAN).
(in French: Caractérisation et hiérarchisation de l’importance pour la PNT des dispositifs d’observation in-situ à la surface océanique en Atlantique Nord.

Type of position : Atmospheric research scientist.
Nature of the position : Senior scientist – salary: from 2100€ to 3100€ per month,
                                  according to candidate’s experience and profile.
Affiliation : National Centre for Meteorological Research (CNRM),Météo-France, Toulouse, FRANCE.
Location :  CNRM-GAME on Météo-France site in Toulouse.
Duration and period : 12 months, non-renewable.
                                The contract should start as soon as possible and not later than August 2015.
Application deadline : Begin 1st April to 31st May 2015.

The scientist will carry out a detailed impact study on observing systems (and part of systems) that are deployed at the marine surface in the North-Atlantic. Here, “observing system”
means an ensemble of observing devices that share common characteristics. For example, drifting buoys, moored buoys or a group of weather station equipped commercial ships.

Context of the study
The EIG EUMETNET (EIG EUMETNET, Secretariat: Avenue Circulaire 3 1180 Uccle, Belgium), that funds this study raised questions about the efficiency of the
North-Atlantic marine surface in-situ observations in constraining the numerical weather prediction errors over Europe. This efficiency refers to the capability to reduce forecast errors
through better initial conditions thanks to data assimilation procedures.
It is known that surface pressure measurements are of particular importance when they are collected by drifting buoys. The influence of other observed parameters such as temperature,
humidity or wind is not known precisely from buoys or ships. This has not been thoroughly evaluated yet.
The EUMETNET Observation Programme includes E-SURFMAR, an operational service dedicated to observing systems at the marine surface.

The EIG EUMETNET prepared a document (hereafter EUMETNET E-SURFMAR study specification) that described both the operating mode and the approaches to be implemented
in that impact study. EUMETNET wonders about the effect of observation density, the complementarity of systems, the sub-regions of the oceanic basin as well as the influence of the
weather types (seasons or regimes) in the control of forecast errors.
For the numerical approaches, this study will use principally linear techniques, to estimate the impact of observations on forecast quality. This way of doing is deemed cheaper than
classical OSEs and allows to assess a much larger number of meteorological situations.

• Forecast Sensitivity to Observations (FSO) is a linear technique that use adjoint model to compute the sensitivity of some aspect of the forecast (defined as a function of that
forecast) to the observations that have been assimilated to produce the forecast initial state. A by-product of these computations is a linear estimate of the influence of each
observation in triggering a change of the function value. If the function is defined as the forecast error, i.e. the difference with respect to a reference state, the FSO will help
diagnosing which observation did contribute to the decrease of the forecast error.

• Observation System Experiments (OSEs) (or observation denial experiment) consist of a more classical ”twin experiments” approach. The influence of the observing system
of interest is evaluated through the difference between the two experiments that are identical except for the observing system that is missing in one of the two. OSE are
typical non-linear approach.

The (weather) case selection and the definition of sub-groups of observations will done by the employee in order to generate sample of data that correspond to experimental specifications given by EUMETNET. Tools already exist, but they may require moderate adaptation to exactly match what’s necessary to carry out the work. For example, a few developments to properly manage observation identifiers as well as observing network density will have to be carried out. For the selection of meteorological events, weather regime assignment methods will be used.
Some of the most striking results of this FSO study may be assessed, through an OSE approach, but only in a second stage. OSEs are not the priority. They are longer and more expensive to undertake, but bring complementary results.

Start of the work
FSO tools applied to the global ARPEGE NWP system exist at the CNRM/GMAP group (Groupe de Modélisation pour l’Assimilation et la Prévision).
These tools allow to compute linear estimates of the contribution of each observation to the forecast quality.
The management team will compute FSO results on as much weather cases as possible, prior to the arrival of the employee. This will allow focusing the work on the exploitation of the FSO results. 

The main deliverable consists of a detailed report on the results obtained with respect to the questions asked by EUMETNET. This report should be produced as soon as possible,
even prior to the end of the contact if possible. Indeed, it has been estimated that the FSO study could be completed in less than 12 months, the remaining time would be dedicated to
complementary OSEs.

• Bibliography
• Selecting subsets of dates and subsets of observations being representative of the E-SURFMAR study specific contexts
• Build coherent samples of sufficient size to obtain representative results
• Write down a comprehensive report on the work done and the results.

Required skills
• Background in unix/linux work environment;
• Good knowledge of Fortran, C or R programming languages and methods is a requirement;
• Good knowledge of documentation writing tools;
• Good knowledge of technical English for scientific meteorology (written and oral).

Desired skills :
• Basic knowledge in physics and geophysics.
• Basic knowledge in statistics
• Knowledge in data assimilation and numerical weather prediction would be an advantage.

Contact :
Jean-Franc ̧ois Mahfouf (CNRM)    +33 (0)
Alexis Doerenbecher (CNRM)    +33 (0)

Bibliography :
• Cardinali C. 2009. Monitoring the observation impact on the short-range forecast. Q.
J. R. Meteolol. Soc. 135, 239-250

Go to: Periods · List Information · Index by: Date (or Reverse Date), Thread, Subject or Author.