A simple method for solar energetic particle event dose forecasting

Bayesian, non-linear regression models or artificial neural networks are used to make predictions of dose and dose rate time profiles using calculated dose and/or dose rates soon after event onset. Both methods match a new event to similar historical events before making predictions for the new events. The currently developed Bayesian method categorizes a new event based on calculated dose rates up to 5 h (categorization window) after event onset. Categories are determined using ranges of dose rates from previously observed SEP events. These categories provide a range of predicted asymptotic dose for the new event. The model then goes on to make predictions of dose and dose rate time profiles out to 120 h beyond event onset. We know of no physical significance to our 5 h categorization window. In this paper, we focus on the efficacy of a simple method for SEP event asymptotic dose forecasting. Instead of making temporal predictions of dose and dose rate, we investigate making predictions of ranges of asymptotic dose using only dose rates at times prior to 5 h after event onset. A range of doses may provide sufficient information to make operational decisions such as taking emergency shelter or commencing/canceling extra-vehicular operations. Specifically, predicted ranges of doses that are found to be insignificant for the effect of interest would be ignored or put on a watch list while predicted ranges of greater significance would be used in the operational decision making progress.