Toward the Assimilation of Images - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2011

Toward the Assimilation of Images


Since several decades the ocean and the atmosphere are permanently observed by satellites which provides many measurement in various channels, this information is used in modern operational data assimilation centers. Satellites provide another kind of information on the move of masses of air and water. This information, of lagrangian type, is of great importance for understanding and forecasting the geophysical processes at the present time the dynamics of images is used more in a qualitative way rather than is a quantitative one. The problem of the assimilation of images is to use this information and merge it with the information provided by numerical models and classical observations in order to retrieve the state of the flow. The main questions raised are: - What do we see? Is it compatible with solutions of numerical models? The lagrangian information is carried by discontinuities in the fields and therefore it will be necessary to extract the equivalent of discontinuities from the solution numerical models. - How to characterize the dynamics of the images? By a field of velocities extracted from images? By a field of Lyapunov exponents characterizing the dynamics of the flows? - How to compare and quantity the discrepancy between the numerical solutions of models and images? - How to control this discrepancy to get the best adjustment of the model to data and images? During the last four years the project ADDISA, funded by the French "Agence Nationale pour la Recherche" has been working on these questions, the talk will present a synthesis of the main.
No file

Dates and versions

hal-00769826 , version 1 (03-01-2013)


  • HAL Id : hal-00769826 , version 1


François-Xavier Le Dimet, Innocent Souopgui, Arthur Vidard. Toward the Assimilation of Images. AMS Annual Meeting, Jan 2011, Seattle, United States. ⟨hal-00769826⟩
98 View
0 Download


Gmail Facebook X LinkedIn More