Atmosphere (Jan 2022)
Automatic Detection of Sfe: A Step Forward
Abstract
Solar flare effects (Sfe) are magnetic variations caused by solar flare events. They only show up in the illuminated hemisphere. Their detection is a difficult task because they do not have a definite pattern and, additionally, they must be separated from other magnetic perturbations. However, we attempted to automatize these detections by using two different strategies. The first strategy takes advantage of one of the Sfe characteristics, as they usually have a rapid rise, followed by a smooth decay, which typically produces a crochet-like shape in the magnetograms. Thus, we created several morphological models for each magnetic component. Then, we identified a definite Sfe time interval by setting the conditions for various parameters, such as the correlations of the measured data with the models, or the model similarities among the different components. In the second stage of this strategy, we observed clusters of time intervals. Each of these clusters were attributed to a timespan of event possibility. We found the statistical optimal value of the correlation parameters by using the ROC curve method and Youden index. The second strategy was based on some of the properties of Sfe ionospheric electric currents, such as their spherical symmetry around the vortex. Here, the algorithm calculated the derivative of the data in order to avoid contamination of the daily variation Sq, and, by means of trigonometric formulas, computed the magnetic radial component relative to the Sfe current vortex (the focus). It then created an Sfe index with this data. A prior assumption of the focus position in a preceding work is no longer needed since we made a wide patrol of the space area to find it. Through a progressive thresholding process, we found its statistical optimal value (0.4 nT min−1) again by using the ROC curve method and Youden index. For both of the strategies, we have made a large calculation of Sfe detection (for the period of 2000–2020), which included 33 Sfe. Finally, we combined the results of both methods—which in fact are complementary—and obtained a unified list that gave a higher hit ratio than those that were obtained separately. This unified method gave promising results towards the possibility of Sfe automatic detection.
Keywords
