FAI and GOES eclipses

Nugget
Number:	508
1st Author:	Hugh HUDSON
2nd Author:	Janet MACHOL
Published:	October 20, 2025
Next Nugget:	TBD
Previous Nugget:	The EUV Late Phase

Introduction: Beware of (rare) FAI failures!

The "Flare Anticipation Index" (Ref. [1]) is a resource for early detection of solar flares. In a preliminary form (based on GOES near-real-time data, and with no anticipation of flare location, it is available here). The FAI has almost no false positives, but there is one situation in which this can happen, as illustrated in Figure 1.

Figure 1: On the left FAI is doing its typical right thing: as a flare approaches, flags are set according to the anticipated flare magnitude (green, amber, red for C, M, X). The dotted and dash-dotted lines show GOES C- and M-class flux levels, respectively. The right panel shows how the algorithm gets confused by eclipses.

The figure shows that the FAI algorithm falsely interpreted an eclipse recovery as a flare. This interesting behavior leads us to study how this works.

Two kinds of GOES eclipse

The GOES satellites are in geostationary orbit above the Earth's equator, and so have much the same geometry as ground-based observers for eclipses of the Sun as occur when the Moon gets between the Sun and the observer. But at geosynchronous orbit, the Earth can also interfere; such eclipses occur on a daily basis in Spring and Fall, and are much more impactful - from this orbit, the Earth subtends a large solid angle, and so the GOES eclipses can last for more than an hour of totality.

Figure 2 shows the GOES XRS time series - fluxes of soft X-rays in the standard bands 1-8 Å and 0.5-4 Å - on a day in which eclipses of both sorts happened. They produce strikingly different effects: the huge angular size of the Earth, as seen from geosynchronous orbit, results in a long umbra with total eclipse, flanked by two-minute intervals of penumbra. The "lunar intrusion" in this case is like a partial solar eclipse as seen on Earth's surface.

Figure 2: Left, GOES-16 soft X-ray time series for a two-hour interval containing an Earth intrusion. The penumbra makes the shadow fuzzy, so it is not as good a knife-edge as that of a true eclipses (the Moon between GOES and Sun). Right, two hours of data with a proper eclipse produced by the Moon sliding across the Sun. Here blue and red are the two GOES passbands, 1-8 Å and 0.5-4 Å respectively. During total eclipse the data show only background counts from non-solar sources, and so are not irradiances.

The images in Figure 3 show how the "true" eclipse works, via EUV images obtained by the SUVI imager on the same GOES-16 spacecraft (image times [15:31:09,15:39:09,16:03:09,16:23:09] UT, 171 Å passband). Note that in this case the N polar region of the Sun is un-occulted as the Moon moves across; this probably explains the non-zero minimum signal seen in the 1-8 Å channel. The lack of a corresponding detection in the 0.5-4 Å channel implies a low plasma temperature (below 4 MK, not exactly arctic, but it is a polar cap).

Figure 3: A series of EUV images obtained by the SUVI telescope on GOES-16 during a "true" eclipse. This series of images spans 50 minutes and does not achieve totality, with the N polar region remaining visible at all times.

Conclusion

The source of the FAI "false positive" signals, such as the one in Figure 1, lies somewhere in the behavior of the penumbral region of an Earth occultation of the GOES spacecraft. There are always two GOES spacecraft in operation (Ref. [2]), so this is not a fundamental problem, just an artifact of the real-time data restriction to only one of them.

More interestingly, the high time resolution (1 s) of the GOES data suggest that using the limb of the Moon as a sharp knife-edge scanner could produce solar X-ray imaging information at angular scales well below an arc second. Note the rich structure in the right-hand panel of Figure 2, which corresponds to the sharp (though rugged) limb of the Moon scanning across the X-ray structure related to the images of Figure 3. For absolute image information one would need to consult the [LOLA] database of correspondingly precise lunar limb shape (Ref. [3]).

References

[1] "Anticipating Solar Flares"

[2] "Chapter 19 - GOES-R Series Solar X-ray and Ultraviolet Irradiance"

[3] "A new lunar digital elevation model from the Lunar Orbiter Laser Altimeter and SELENE Terrain Camera"