Assessment of solar wind driven ionospheric storm forecasts: The case of the Solar Wind driven autoregression model for Ionospheric Forecast (SWIF)

This paper reports the results of the most recent validation tests on the performance of the Solar Wind driven autoregression model for Ionospheric Forecast (SWIF). SWIF uses the Interplanetary Magnetic Field (IMF) observations at L1 point as a proxy for upcoming ionospheric storm time disturbances, issuing an alert in case certain quantitative criteria are met. These criteria apply to the IMF total magnitude and the IMF-Bz component. The validation tests address the SWIF alert detection efficiency in the period 2014-2020 and were designed to complement previous efforts and allow the comprehensive discussion of SWIF's alert ability over one and a half solar cycle. The model's efficiency is quantified through probability of detection, false alarm rate and the success ratio for SWIF's users. The results are further discussed in terms of the intensity and the interplanetary causes of the storms to reveal high forecasting ability for specific storm scenarios (i.e. intense storms that are usually driven by coronal mass ejections), but also certain limitations in the efficiency of SWIF's alert criteria in capturing storm related interplanetary structures at L1 point: mainly structures that are not related to coronal mass ejections and typically, drive storms of moderate intensity. Additional results obtained by superposed epoch analysis of IMF key components indicate the potential relevance of the IMF-By component in ionospheric forecasting applications. The consideration of the IMF-By in SWIF's formulation can drive future improvements in its forecasting efficiency in all different faces of solar wind forcing. The findings of this work may also help the better understanding of the processes that control the coupling of the solar wind-magnetosphere-ionosphere.(c) 2022 COSPAR. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/ by-nc-nd/4.0/).