Assessing Dst prediction models for forecasting the geoeffectiveness of ICME structures

We compare solar wind parameters and storm conditions for two geomagnetic storms recorded during solar cycles 24 and 25 – 17 March 2013 and 23 April 2023. We select these events since they are isolated coronal mass ejection (CME) events with a two-step storm profile, which occurs because both the sheath and magnetic cloud of the CME are geoeffective. We predict the storm profile by implementing five Dst forecast models, Burton et al. (1975), Fenrich & Luhmann (1998), O’Brien & McPherron (2000), Wang et al. (2003), and Temerin & Li (2006) – on the in situ observations obtained at L1. We study the SYM/H profiles in the sheath and cloud regions by comparing the actual SYM/H index recorded in the OMNI database with predictions from the models by evaluating the difference in the recorded and predicted SYM/H minima and the time difference between the storm minima. Our study demonstrates that among the five models evaluated, the Temerin & Li (2006) model excels in predicting the overall storm profile and peak time of geomagnetic storms. In contrast, Fenrich & Luhmann (1998) provides a more precise forecast of the intensity in the storm’s peak. In this study, we defined a new metric, CI, the ‘storm coupling integral’, by integrating a solar wind coupling function over a time interval to quantify its geoeffectiveness. In addition to indicating the geoeffectiveness, CI can be used as a tool to assess and improve the overall performance of forecast models.