Development of Daily Maximum Flare-Flux Forecast Models for Strong Solar Flares

Abstract

We have developed a set of daily maximum flare-flux forecast models for strong flares (M- and X-class) using multiple linear regression (MLR) and artificial neural network (ANN) methods. Our input parameters are solar-activity data from January 1996 to December 2013 such as sunspot area, X-ray maximum, and weighted total flare flux of the previous day, as well as mean flare rates of McIntosh sunspot group (Zpc) and Mount Wilson magnetic classifications. For a training dataset, we used 61 events each of C-, M-, and X-class from January 1996 to December 2004. For a testing dataset, we used all events from January 2005 to November 2013. A comparison between our maximum flare-flux models and NOAA model based on true skill statistics (TSS) shows that the MLR model for X-class and the average of all flares ($\mathrm{M}{+}\mathrm{X}$-class) are much better than the NOAA model. According to the hitting fraction (HF), which is defined as a fraction of events satisfying the condition that the absolute differences of predicted and observed flare flux on a logarithm scale are smaller than or equal to 0.5, our models successfully forecast the maximum flare flux of about two-thirds of the events for strong flares. Since all input parameters for our models are easily available, the models can be operated steadily and automatically on a daily basis for space-weather services.