A comparison study of extrapolation models and empirical relations in forecasting solar wind

TL;DR

This study compares different extrapolation and empirical models for forecasting solar wind properties, evaluating their accuracy and ability to predict magnetic field polarity at L1, to improve space weather prediction tools.

Contribution

It introduces and assesses two solar wind extrapolation models, including a physics-based hydrodynamic model, for improved background solar wind forecasting.

Findings

Physics-based model predicts thermal properties accurately.

High correlation coefficients (0.73-0.81) achieved in predictions.

Model standard deviation matches OMNI data for solar wind speed.

Abstract

Coronal mass ejections (CMEs) and high speed solar streams serve as perturbations to the background solar wind that have major implications in space weather dynamics. Therefore, a robust framework for accurate predictions of the background wind properties is a fundamental step towards the development of any space weather prediction toolbox. In this pilot study, we focus on the implementation and comparison of various models that are critical for a steady state, solar wind forecasting framework. Specifically, we perform case studies on Carrington rotations 2053, 2082 and 2104, and compare the performance of magnetic field extrapolation models in conjunction with velocity empirical formulations to predict solar wind properties at Lagrangian point L1. Two different models to extrapolate the solar wind from the coronal domain to the inner-heliospheric domain are presented, namely, (a) Kinematics based (Heliospheric Upwind eXtrapolation [HUX]) model and (b) Physics based model. The physics based model solves a set of conservative equations of hydrodynamics using the PLUTO code and can additionally predict the thermal properties of solar wind. The assessment in predicting solar wind parameters of the different models is quantified through statistical measures. We further extend this developed framework to also assess the polarity of inter-planetary magnetic field at L1. Our best models for the case of CR2053 gives a very high correlation coefficient ($\sim$ 0.73-0.81) and has an root mean square error of ($\sim$ 75-90 kms$^{-1}$). Additionally, the physics based model has a standard deviation comparable with that obtained from the hourly OMNI solar wind data and also produces a considerable match with observed solar wind proton temperatures measured at L1 from the same database.