AVENUE Project
Variational-Ensemble Data Assimilation
Project period : September 2014 – September 2017
Coordinators : Ehouarn Simon (INPT-IRIT), Selime Gürol (CERFACS)
Partners: CERFACS, INPT-IRIT, MétéoFrance-CNRM, CNES-GET
Objectives
- Stochastic modeling of uncertainties (Initial conditions, models, observations)
- Large-scale non-linear ensemble estimation
- HPC algorithms for data assimlation
- Applications in meteorology, oceanography and earth’s gravity field
Methodologies
- Development of a variational-ensemble methodologies
- Probabilistic convergence study for large samples
- Time-parallel methodologies
- Covariance filtering
Application : Meteorology
- Météo-France model: global (ARPEGE) and regional (AROME)
- Strong operational constraints (HPC)
- Covariance matrix localisation
Application: Ocean
- Low resolution global model (NEMO) – (1 deg.) and high resolution (1/4 deg)
- Hybrid formulation of covariances
- Development of HPC algorithms
Application: geodesy
- Restitution of the earth’s gravity field on the basis of GRACE satellite measurements
- Inversion + spectral analysis in large scale
- Implementation of EnKF filters
