You are here: Applications > Projects > Data Assimilation
MERCATOR, operational oceanography
The Groupement d'Intérêt Public (GIP) Mercator
Océan is the first user of the PALM coupler. Mercator took part
in the funding of the coupler development to meet the commitment of the
setup and implementation of modular and evolutive ocean forecast and
data assimilation suites. The PALM coupler is used for R&D
configurations and in operations for all the data assimilation suites.
The MERCATOR system receives observational satellite or in situ data. These observations
are "assimilated" by the analysis and forecast model. The observational
data assimilation extends the validity of the ocean description and
forecast up to 14 days.
ADOMOCA, atmospheric chemistry
The french scientific community working on atmospheric chemistry has
decided to form a working team in "data assimilation" organised in
three main axes, with specific functions. The final goal is the setup
and the exploitation of two data assimilation suites, driven by PALM.
The three axes are: the "expertise centres", the "assimilation suites"
and the "external links". The "expertise centres" axis is focused on
the observation operators, the direct and adjoint models, the
assimilation algorithms and primarily the PALM coupler, backbone of the
whole project. The "assimilation suites" belong to two classes: the
suites driven by PALM, using PALM interfaced operators and models and
producing analyses in a R&D or pre-operational mode and the
research assimilation suites, collecting all the assimilation
algorithms previously implemented in the different laboratories. The
"external links" take care of the dissemination of the operational
(GMES) or research (ASSET, SAF O3) results, mainly trhough
the ETHER database and access facilities.
ADONIS, neutronics
PALM has been used for the ADONIS research project, jointly carried by
EDF
and CERFACS. This project showed the improvements coming
by data assimilation techniques for the knoledwge of the physics of
nuclear plants kernels. Data assimilation relies on measurement
directly taken in the reactors and on a neutronics model. PALM has been
used for the setup of two applications: one was focused on the
optimal interpolation of a physical fields using variational
assimilation techniques and the other was focused on the tuning of the
numerical parameters of a neutronics model developed by EDF. Moreoveor,
PhD level studies are carried on at EDF on the evaluation of different
assimilation algorithms without the constraint of using an industrial
code.
SAFRAN-ISBA-MODCOU, hydro-meteorology
Safran-Isba-Modcou (SIM) is an hydro-meteorological chain composed of
a meteorological analysis (Safran), a soil-vegetation-atmosphere-transfer model (Isba) which simulates the energy and water fluxes,
and an hydrological model (Modcou) which simulates the evolving of river streamflows and some French aquifer layers. SIM is running
on the all of France, with a 8kmx8km grid.
Two ensemble streamflow prediction systems based on SIM have been set up, the first one using the ECMWF EPS (10-day forecasts),
the second one using the short-term Météo-France EPS, the PEARP.
One of the solutions to improve this system is to improve the initial state of the streamflows. Currently, the initial streamflows are provided
by a SIM simulation based on the analysed meteorological data. They differ from observed streamflows in a certain range. In order to improve
the initial streamflow of the system, and so the streamflows predictions, assimilations technics will be set up to force the SIM streamflows
to get closer from the real streamflows.
ATHYS, MASCARET, flood forecasting
This activity is the object of the collaboration of the Service
Central
d'Hydrologie et d'Aide aux Prévisions des Inondations (SCHAPI)
and
CERFACS, aiming to explore the potential of improvement of flood
forecasting by the use of data assimilation. We use a classical BLUE
(Best Linear Unbiased Estimate) assimilation algorithm to tune some
relevant parameters of the direct models. A case study uses the the
rain-discharge model of the ATHYS platform to simulate the flash floods
of the Gardon d'Anduze river. A second one uses the fluvial model
MASCARET for the real-time simulation of the course of the Adour river
and of its tributaries. The flood forecast scores obtained with this
technique show a considerable improvement if compared to other
classical approaches in hydrology and hydraulics.