Do the archival Keck HIRES spectra show e.g. H-alpha variability?

[gully]

There are 22 Keck HIRES epochs from Zhou et al. 2017:

	MJD	decker	iodine	"usable" orders
0	55016.5	B5	True	49
1	55042.4	B5	True	49
2	55043.5	B5	True	47
3	55044.4	B5	True	49
4	55048.4	B5	True	27
5	55107.2	B5	True	49
6	55107.2	B5	False	47
7	55251.5	C2	False	29
8	55696.3	C2	True	31
9	55696.4	C2	True	28
10	55697.3	C2	True	34
11	55698.4	C2	True	22
12	55699.3	C2	True	31
13	55700.4	C2	True	27
14	55704.3	C2	True	32
15	55705.4	C2	True	33
16	55706.3	C2	True	34
17	55707.3	C2	True	34
18	55853.2	C2	True	34
19	55944.7	C2	True	37
20	55996.5	C2	True	24
21	57935.3	B5	False	27

Unfortunately the spectra have some Issues:

1. 19 out of 22 were taken with the Iodine Cell

The Iodine cell should only corrupt spectra between about 5000 - 6500 Angstroms, where the Iodine has the greatest absorption. In practice something else seems to be corrupting the spectra outside of this range...

2. The C2 decker causes order overlap

Decker	H(")	W(")	R
B5	3.5	0.861	48,000
C2	14.0	0.861	48,000

The C2 decker has a 14 arcsecond long slit. This choice appears adequate for long wavelengths, but the short wavelength echellogram must pile-up causing order overlap and corrupting the bluest orders. Unfortunately that means CaII H&K lines are corrupted in 16 out of 22 of the spectra. But the 8 spectra that were taken in B5 Decker appear kosher. So we could restrict our attention to those 8 visits.

3. H alpha appears unavailable in the Keck Observatory Archive for many of the Iodine-cell spectra.

In principle the H-alpha region should be retrievable even in the presence of the Iodine cell. In practice the data reduction process balks at the Iodine lines and most H-alpha spectra are therefore absent.

4. The wavelength calibration appears coarse

Overlaying spectra shows a smearing out of the lines, rather than the nice sharp overlay we'd expect. I don't think the effect is astrophysical, since the 0.3 km/s peak-to-valley RV variation (Table 3 of Zhou et al. 2017) is about one-hundredth of the vsini. We're seeing variations at the 10 km/s peak-to-valley range (we have applied a barycentric correction and avoided telluric lines in our cross-correlation analysis), so these must be an artifact of the automate KOA pipeline.

Altogether we're in a bit of a bind-- extracting the insights we had hoped for will take more work than we want. Maybe the easiest thing we can do for now is focus on the B5 decker spectra for the CaII H&K analysis...

[gully]

Ok, here are the Keck HIRES spectra around the $\text{H}\alpha$ and CaII H & K lines. We compare to Keck HIRES spectra from a similar F star as characterized by Yee et al. 2017 (Specmatch-EMP). The HAT-P-67 spectra are consistent with a rotationally modulated version of the template spectra.

We see no conspicuous visit-to-visit variability nor do we see persistent chromospheric emission. Any conceivable features appear consistent with the background noise level. We are only showing those spectra that appear uncorrupted, based on the guidance above.

All in all, these diagnostics support the idea that the star appears relatively inactive, and that planetary atmosphere lines appear to manifest only in the Helium 10830 triplet lines, at least at this signal-to-noise level.