GmapDoc website
Slogan du site
Descriptif du site
2000 - Realization of a simplified, differentiable and realistic physical parametrization for incremental four-dimensional variational data assimilation
Article published on June 2000
last modification on 25 October 2005

A complete set of the simplified physical parametrizations has been developed for incremental 4D-Var assimilation system. It contains a simplified computation of radiative fluxes, vertical turbulent diffusion, orographic gravity waves, deep convection and stratiform precipitation fluxes. The validation of the whole direct simplified physical package has been done.

To include the described simplified physical parametrizations into assimilating model, their tangent linear and adjoint have been developed. The accuracy of the linearization of the simplified physical parametrizations has been investigated. The problem of additional regularization, i.e. smoothing modifications of some discontinuities in the parametrizations, arose during this validation and some smoothing modifications had to be developed.

To investigate the sensitivity of a 4D-Var assimilation system to physical processes, 4D-Var experiments using the adjoint model with simplified physical parametrizations have been performed. FASTEX (Fronts and Atlantic Storm Track Experiments) situations have been selected as test cases. The potential of including the physical processes in the assimilating model for the representation of the cyclogenesis has been demonstrated.

This document was written by Marta Janiskova from her PhD thesis in June 2000.
It presents :
 Position of the problem
 Standard formulation of 4D-Var
 Incremental approach of 4D-Var
 Importance of including physics in the assimilating model
 Surface processes
 Stratiform precipitation
 Vertical diffusion
 Solar radiation
 TL and AD models
 validation of TL model
 Importance of the regulation of the TL model
 regularization of the vertical diffusion scheme
 Regularization of the radiation & convection schemes
 4D-Var experiments
 One-week assimilation cycle

Physical parametrization for 4D-Var data assimilation 195.7 kb / PostScript