Speaker
Description
The Earth’s radiation belts are a complex system which dynamics spans many orders of magnitude and impacts satellites instruments and electronics. Forecasting and reanalyzing the overall state of this population is of prime importance for Space Weather and Space Climate activities.
For years, ONERA has been developing a model of the Earth’s radiation belts dynamics based on the Fokker-Plank equation named Salammbô. Nevertheless, such physics-based models always suffer from uncertainties as diffusion coefficients rely on statistical analysis. In parallel, ONERA also disposes of a large database of in-situ measurements (IPODE). In order to benefit of these two complementary sources of information we have developed data assimilation technics as it is done in Oceanography. We thus use an Ensemble Kalman Filter (EnKF) to assimilate satellite observations into Salammbô code. This way we improve and supervise Salammbô’s accuracy, while removing bias thanks to this optimal combination.
The tricky part is that one needs to accurately estimate the a priori uncertainties of both the model and the observations to ensure optimality of the assimilation process. This study focuses on the estimation of observations representation uncertainties, which is a topic not currently covered in our field of research. This way, it becomes possible to improve both forecast and reanalysis of the dynamics of the Earth’s radiation belts.
To assimilate satellite data, we first need to convert their positions, known in geographic coordinates, to magnetic coordinates (frame of the modelling). Several models exist to do this conversion, and they all possess their own bias and uncertainty. In IPODE database, we use Olson-Pfitzer Quiet (OPQUIET) model, which has two major issues: first, it is only suited for quiet geomagnetic conditions, and second it is less accurate than the more recent models. During the assimilation of satellite data, and especially during storm conditions, OPQUIET creates a representation error that we seek to reduce. This contribution tackles the impact of the modelling of OPQUIET bias and uncertainty on radiation belts study and Salammbô-EnKF results. We used the magnetic model Tsyganenko 89 (T89) as a reference. T89 is suited for both quiet and storm conditions and uses only Kp as a proxy. Through statistical analysis of THEMIS satellites positions over 17 years (2008 – 2025) we have been able to model both the bias and the uncertainty of OPQUIET over the L coordinates.
We first present the method use to model OPQUIET uncertainty, and the model and its use in the EnKF. We then show a comparison made by a twin experiment, and in real conditions, of the impact of the different magnetic models over Salammbô-EnKF results and discuss these results. We show that our model of OPQUIET uncertainty and bias on L enables a more accurate EnKF assimilation, particularly under storm conditions, hence enhancing Salammbô-EnKF’s predictions for radiation belts dynamics in the inner and the outer magnetosphere.
| Do you plan to attend in-person or online? | In-person |
|---|
