Uncertainty quantification of the DTM2020 thermosphere model

Abstract

Aims: The semi-empirical Drag Temperature Models (DTM) calculate the Earth's upper atmosphere's temperature, density, and composition. They were applied mainly for spacecraft orbit computation. We developed an uncertainty tool that we implemented in the DTM2020 thermosphere model. The model is assessed and compared with the recently HASDM neutral density released publicly in 2020. Methods: The total neutral density dataset covers all high-resolution CHAMP, GRACE, GOCE, and SWARM data spanning almost two solar cycles. We constructed the uncertainty model using statistical binning analysis and least-square fitting techniques, allowing the development of a global sigma error model to function the main variabilities driving the thermosphere state. The model is represented mathematically by a nonlinear manifold approximation in a 6-D space parameter. Results: The results reveal that the altitude parameter presents the most notable error range during quiet and moderate magnetic activity (Kp <5). However, the most considerable uncertainty appears during severe or extreme geomagnetic activities. The comparison with density data provided by the SET HASDM database highlights some coherent features on the mechanisms occurring in the thermosphere. Moreover, it confirms the tool's relevance to provide a qualitative database of neutral densities uncertainties values deduced from the DTM2020 model.