Database of single-particle autocorrelation functions for machine-learning based incoherent scatter spectral models

At the Jicamarca Radio Observatory, estimations of ionospheric parameters are performed employing inversion techniques based on simplified incoherent scatter (IS) spectral models. However, the applicability of these models is restricted to some particular regimes. On the other hand, the use of more advanced models is highly computationally demanding making impractical their application to routine estimation of parameters. To overcome this, machine learning techniques will be used to develop a fast and efficient model for ionospheric parameter estimation. For the development of this model, single-particle autocorrelation functions (ACF) for ions and electrons in a magnetized plasma configuration at a specific set of aspect angles are needed. This poster focuses on building a database of these ACFs for its use in developing the IS model. To compute these ACFs, a large set of simulated particle trajectories following a stochastic process is required. First, a stochastic differential equation (SDE) numerical solver with the best weak order of convergence for computing the single-particle ACFs was determined. This was done considering a non-collisional oxygen plasma in thermal equilibrium. The simulated trajectories obtained using the chosen solver were then used to compute the single-particle ACFs applying Fourier transform approach. All the simulations were implemented using the JULIA programming language, which offers a dedicated library for solving SDEs. The obtained database will also be a valuable resource for training different machine-learning based incoherent scatter models aimed at improving ionospheric parameter estimation.