Comparison of solar flare forecasting methods

Solar flares produce X-rays which can have an almost immediate effect on the near-Earth environment, making it crucial to forecast flares in order to mitigate their negative effects. The energy available for flares is believed to be stored in the magnetic fields of active regions, and released by magnetic reconnection in the corona. Recently, the number of published approaches to flare forecasting using photospheric magnetic field observations has proliferated, with widely varying claims about how well each works. Because of the different analysis techniques and data sets used, it is essentially impossible to compare the results from the literature. This problem is exacerbated by the low event rates typical of large solar flares. The challenges of forecasting when event rates are low have long been recognized in the meteorology community, but have yet to be fully acknowledged by the space weather community. During the interagency workshop on ``all clear'' forecasts held in Boulder, CO in 2009, the performance of a number of existing algorithms was compared on a common data set, with consistent definitions of what constitutes an event. We summarize the importance of making such systematic comparisons, and of using standard verification statistics to determine what actually constitutes a good prediction scheme.

This work was funded by NASA LWS TRT contract NNH09CE72C.