Forecasting solar wind structures and shock arrival times using an ensemble of models

Abstract

Forecasting the time of arrival at Earth of interplanetary shocks following solar metric type II activity is an important first step in the establishment of an operational space weather prediction system. The quality of the forecasts is of utmost importance. The performances of the shock time of arrival (STOA) and interplanetary shock propagation models (ISPM) were previously evaluated by Smith et al. [2000] for 36 solar events. Here we use 173 solar events between February 1997 and October 2000 to set thresholds for the Hakamada–Akasofu–Fry version 2 (HAFv.2) model and then present the results of a comparison of the performance of this model to the STOA and ISPM solar wind models. Each model predicts shock arrival time (SAT) at the Earth using real‐time metric type II radio frequency drifts and coincident X‐ray and optical data for input and L1 satellite observations for verification. Our evaluation of input parameters to the models showed that the accuracy of the solar metric type II radio burst observations as a measure of the initial shock velocity was compromised for those events at greater than 20° solar longitude from central meridian. The HAF model also calculates the interplanetary shock propagation imbedded in a realistic solar wind structure through which the shocks travel and interact. Standard meteorological forecast metrics are used. A variety of statistical comparisons among the three models show them to be practically equivalent in forecasting SAT. Although the HAF kinematic model performance compares favorably with ISPM and STOA, it appears to be no better at predicting SAT than ISPM or STOA. HAFv.2 takes the inhomogeneous, ambient solar wind structure into account and thereby provides a means of sorting event‐driven shock arrivals from corotating interaction region (CIR) passage.