Evaluating Pointing Strategies for Future Solar Flare Missions

Abstract

Solar flares are events of intense scientific interest. Although certain solar conditions are known to be associated with flare activity, the exact location and timing of an individual flare on the Sun cannot as yet be predicted with certainty. Missions whose science objectives depend on observing solar flares must often make difficult decisions on where to target their observations if they do not observe the full solar disk. Yet, there is little analysis in the literature that might guide these mission operations to maximize their opportunities to observe flares. In this study, we analyze and simulate the performance of different observation strategies using historical flare and active region data from 2011 to 2014. We test a number of different target selection strategies based on active region complexity and recent flare activity, each of which is examined under a range of operational assumptions. In each case, we investigate various metrics such as the number of flares observed, the size of flares observed, and operational considerations such as the number of instrument repoints required. Overall, target selection methods based on recent flare activity showed the best overall performance but required more repointings than other methods. The mission responsiveness to new information is identified as another strong factor determining flare observation performance. It is also shown that target selection methods based on active region complexities show a significant pointing bias toward the western solar hemisphere. As expected, the number of flares observed grows quickly with field-of-view size until the approximate size of an active region is reached, but further improvements beyond the active region size are much more incremental. These results provide valuable performance estimates for a future mission focused on solar flares and inform the requirements that would ensure mission success.

Appendix: Flare Productivity of McIntosh Classifications

There are multiple estimates of the flare productivity of different McIntosh active region classifications. Many of these flare rates are based on relatively small number statistics and are subject to significant uncertainty. In the main body of the paper, we adopt the M-class flare rates published by Bloomfield et al. (2012). However, it is useful to compare the results of the simulation using alternative sources. Combined M- and X-class flare rates were previously published by Bornmann and Shaw (1994). Additionally, Bornmann and Shaw (1994) suggested that flare productivity of any active region could be estimated by averaging the overall productivity of each of the three McIntosh parameter values. One motivation for this approach is the ability to estimate the productivity of a region that has not been observed or has very few observations.

Figure 12a shows the flare productivity values published by Bloomfield et al. (2012) (blue) compared with the estimates by Bornmann and Shaw (1994) (orange). We can see that the overall trends in these datasets are consistent, with a few prominent outliers of note. Both datasets agree that the five most productive regions are those designated EKC, FKC, EHC, FSI, and FKI, with differences in the exact productivity values and ordering.

Figure 12

Top: The flare productivity estimates for each McIntosh active region classification according to Bloomfield et al. (2012) (blue) and Bornmann and Shaw (1994) (orange). For the Bloomfield et al. (2012) estimates, we take the values calculated for M-class solar flares, whereas the Bornmann and Shaw (1994) values combine M- and X-class solar flares. Bottom: The rank order of each of the McIntosh active region classifications in terms of their flare productivity. Rank 1 is the most flare-productive classification.

Figure 12b illustrates the comparison of these datasets via their rank ordering, where rank 1 is the most flare-productive region. This reinforces that both datasets agree on the overall trend of productivities, with localized differences in the ordering. The most notable deviation is the classification FAO; in the original Bornmann and Shaw (1994), this classification was assigned a productivity of 0.0; however, in Bloomfield et al. (2012), it has an estimated productivity of 0.21 flares/day, the 25th most productive classification. The FHO classification was also assigned a productivity of 0.0 by Bornmann and Shaw (1994) but has a productivity of 0.05 in Bloomfield et al. (2012). The other most notable discrepancy is that the FKI and EKI McIntosh classifications are both considered much more productive in the Bornmann and Shaw (1994) dataset compared with the Bloomfield et al. (2012) dataset, both in flare productivity value and in their rank order.

Figure 13 shows the impact of adopting these different methods of ranking McIntosh classifications on the mission performance. This simulation used the 12-hour delay scenario (see Section 4.1). From this it is clear that the difference is small; using the Bloomfield et al. (2012) flare productivities, a mean of 39.7% of $>\text{M1}$ flares was observed, whereas when the Bornmann and Shaw (1994) data were adopted, the mean percentage of flares observed was 39.6%. Interestingly, the “averaging” method proposed by Bornmann and Shaw (1994) performs slightly better than the explicit data-based approaches, observing a mean of 41.3% of available flares. However, this approach is more challenging to scientifically justify, and we therefore choose not to adopt it in the main body of this work.

Figure 13

The flare observing performance of the mission using the McIntosh target method and three different ways of estimating the flare productivity of McIntosh classifications. Left panel: Performance of the mission using flare productivities calculated using the Bloomfield et al. (2012) M-class flare productivities. This is used in the main body of the paper. Center: Mission performance using the flare productivity estimates from Bornmann and Shaw (1994). Right panel: Mission performance obtained by averaging the overall productivity of each individual McIntosh parameter, as originally suggested by Bornmann and Shaw (1994).