Evaluation of CNES RL05 by AI and prospects for a sixth re-iteration
- 1CNRS, CNES DTN/CD/GS, Toulouse Cedex 9, France (jean-michel.lemoine@cnes.fr)
- 2Stellar Space Studies, 5 Esp. Compans Caffarelli, 31000 Toulouse, France
- 3Magellium, 1 Rue Ariane, 31520 Ramonville-Saint-Agne, France
Since December 2019, CNES is producing its fifth release of GRACE and GRACE-FO time-variable gravity solutions. This presentation will provide an update on the evolution of processing since last year and on the prospect of a new reprocessing in 2025, based on the new FES2022 barotropic tide model that recently became available and on the update of the TUGO ocean dealiasing model by the LEGOS/CLS team.
Furthermore, in order to characterize as objectively as possible the quality of the time-variable gravity field solutions produced by the different groups involved in the GRACE/GRACE-FO processing, a software for detecting defects in the solutions by supervised artificial intelligence methods was developed for CNES by the company Magellium. A description of the algorithms used (Random Forest and Convolutional Neural Network) will be presented, as well as the performance indices obtained for the CNES solution and, we hope, for a set of other solutions.
How to cite: Lemoine, J.-M., Bourgogne, S., Bruinsma, S., Vaujour, T., Pfeffer, J., and Thenoz, C.: Evaluation of CNES RL05 by AI and prospects for a sixth re-iteration, GRACE/GRACE-FO Science Team Meeting, Potsdam, Germany, 8–10 Oct 2024, GSTM2024-84, https://doi.org/10.5194/gstm2024-84, 2024.
