RR Ly to test/validate Galactic extinction

June 1 2022.

SDSS stripe 82 RR LY: https://vizier.cds.unistra.fr/viz-bin/VizieR-4 

Example SDSS RR Ly in the North

Sept 30, 2022, CWS notes

RR Lyrae are population II stars, roughly spherically distributed in the MW. Their apparent magnitude at the distance of the Large Magellanic Cloud (LMC, d=50 kpc while our distance from MW center is about 7.5 kpc) is 18th magnitude, so ATLAS detects them well outside MW dust lanes. 

Two possible approaches: 1) use RR LY colors to determine reddening, 2) use their apparent magnitudes and GAIA distances to determine attenuation directly. We can do both. 

References: 

Oct 2, 2022: 

Downloaded the entire ATLAS table4.dat catalog from Vizier. It's fixed-column format and machine-learning classifier output that is relevant for RR Ly is 

1715-1733 F19.17 ---       P(PULSE)        [0/1] Machine classifier probability

                                             that this star is in the PULSE 

                                             category

Full description of columns given here:  ATLAS variable catalog description

ATLAS file is fixed-format field width. SQLlite wants a comma-separated file. Here is an awk script that replaces multiple spaces with a comma:

awk 'OFS="," {$2=$2};1' table4.dat > table4.csv

did cut-and-paste operations from catalog description to make a file, colnames.txt, that has names of fields, one per row. Converted this into a single CSV row with

paste -d ',' -s colnames.txt > colnames.csv 

then 

cat table4.csv >> table4colnames.csv

This should make a comma separated file that can be imported into SQL lite or the equivalent. 

Uh oh. There seems to be an inconsistency in one of the DB fields. Specifically, 

595- 605 F11.6  d         fp-domper-o     [-0.000008/1499.992432]? Period
                                             corresponding to fp-powerterm-o
       606 A1     ---     f_fp-domper-o     [I] I for infinity
this add-on field in column 606 is problematic. 

Workaround is to delete the field name f_fp-domper-o and make fp-domer-o a text field. 

Deleted that as a field name, and remade CSV catalog. New CSV ATLAS variable catalog is called ATLASvars.csv. 

Database in SLQllite has 4.8M rows. 

select count(*) from ATLASvars where CLASS='PULSE' gives 25,888 

From ATLAS variable star paper: 

So we want objects with 0.2 < Period < 1.5 to and class = PULSE. 

GRRR need to enclose any column name that has a hyphen within double quotes. 

select count(*) from ATLASvars where (CLASS='PULSE' ) AND  ("fp-period" > 0.2) AND ("fp-period" < 1.5);

gives 23,372 stars. 

Add in a SNR cut and see what that does to object counts

select count(*) from ATLASvars where (CLASS='PULSE' ) AND  ("fp-period" > 0.2) AND ("fp-period" < 1.5) AND ("df-meansig" > 10);

Good. First, let's trim the table

create table RRLy as select * from ATLASvars where (CLASS='PULSE' ) AND  ("fp-period" > 0.2) AND ("fp-period" < 1.5)

Culled exported data file with PULSE variables with periods between 0.2 and 1.5 days (link): RRLy.csv

head -1 RRLy.csv |  awk -F "," '{print $2,$3,$63,$64,$65,$66,$71,$72'}

RAdeg DEdeg fp-min-c fp-max-c fp-min-o fp-max-o fp-const-c fp-const-o

O band constant magnitude term: 

Absolute magnitude is +0.75. So distance is given by (m-M) = -  5 + 5 * log10(D/1 pc), so log10(D/1 pc) = ((m  - 0.75) +5)/5.

Convert to histogram of distances. For scale, distance from here to Galactic center is about 7.5 kpc. 

The ATLAS data have these advantages: 1) hundreds to thousands of data points, 2) with all-sky well-calibrated photometry and 3) dynamic range in magnitudes well-matches to this problem. 

Steps from here: 

1. RR Ly 'magnitude' depends on metallicity, distance, age, and extinction. We should try to obtain independent measurements of the first three of these. 
   1. metallicity can come from APOGEE spectra, or the phase relationship between RR Ly light curve Fourier components.
   2. distances can come from GAIA.  Median constant term for Fourier fit magnitudes is 16th in o band. GAIA parallax uncertainty at 116th is about 0.08 marcsec, or 80 microarcsec.
      Distance for RR LY at 16th is 22.4 kpc. 1 mas of parallax uncertainty corresponds to 1 kpc distance uncert. 

2. Download the ATLAS light curves for the RR Lyrae in the data file above, from both differential photometry and unsubtracted.

3. Use the periods determined from Heinze paper to do period-folded analysis and obtain potentially updated Fourier parameters and other attributes using full time series. 

3.25 need to decide on whether we will use light curves of intensity or magnitudes to determine a mean magnitude. Is there a standard practice for this? 

3.5 Think about whether we need to correct to solar system barycenter timing to account for light travel time effects across Earth's orbit. See below for some estimates on this. Answer is yes. 

4. Assemble a combined data file with Schlafly et al extinction values, GAIA distances, and any well-calibrated IR photometry (WISE, 2MASS, etc). 

4.5 Compare to the 55 stars in the Neely et al paper "Standard Galactic field RR Lyrae II...", linked above, to determine o-c color at minimum flux for RRc and RRab type stars. 

5. Determine a metallicity-corrected and period-corrected apparent magnitude for the ATLAS stars. See plot below for period-to-color conversion. 

6. Use Gaia distances to determine difference between distance modulus (m-M) and distance. 

6.5 Analyze type ab and type c stars separately. Look for beats between fundamental and first harmonic Fourier components to eliminate multimode pulsators.  

7. Correlate the difference between these quantities and catalogued extinction values. 

8. Beware RR Ly that lie near Galactic center, since they're embedded in dust clouds. We should make an extinction cut (0.2, 0.3 mag?). That cut should be informed by the desire to make extinction corrections with mmag residuals.

9. Rather than just using periods and Fourier phase relationships and averaged magnitudes,  could we use the color excursions over the course of an oscillation? Is there more information there we aren't using?  

So for an RR Ly of fixed absolute magnitude, and known parallax, we should be able to predict its apparent magnitude to within 5 millimag. 

Converting from reddening to extinction with Rv=3.1 would require measuring a color shift three times smaller than this, or around 1.6 millimag. A magnitude difference, to determine a color,  has an uncertainty sqrt(2) times larger than each individual measurement, which would require 1.6/sart(2) = 1.1 mmag per band. How tight are the colors we have? Plot (c-o) vs. o:

Nope, likely dominated by extinction. 

Period-metalllicity-color-luminosity relationship. 

What's the extent of RR Ly in color-magnitude diagram? There is a relationship between period and color of star? Long-ago paper from John Huchra has this plot of how period relates to color 

Neely et al paper "Standard Galactic field RR Lyrae II..." suggests that (V-I) colors at minimum flux are known to around 0.02 mag. If that's the reddening uncertainty then how well can we infer attenuation? 

Using  IPAC extinction calculator: 

Here is a file with extinction calculations at higher attenuation, for many passbands: extinction_calculator.tsv

We see that if E(V-I)=0.032, then A_v=2.18*E(V-I). (note it's unclear what SED was assumed for this extinction estimate)

So if the intrinsic (V-I) color uncertainty of an RR Ly at minimum light is around 0.02 mag (we should be able to do better, however) and this dominates the determination of reddening, then the V-band extinction is uncertain by 2.18 times as much, or 0.04 mag. 
Unless we tighten up the minimum-light color, that would seem to be a floor in extinction determination. 

What about using magnitudes rather than colors? If color can be determined to 0.02 mag, then the precision (not accuracy) in each band is around 0.014. This implies that using magnitudes rather than colors should be about 4X more incisive for attenuation determination. 
It's only because we know Gaia distances that this is plausible. 

Extinction in Rubin deep drilling fields? 

What are the nominal extinction values to Rubin deep drilling fields? Extinction from NED IPAC calculator at https://ned.ipac.caltech.edu/extinction_calculator?in_csys=Equatorial&in_equinox=J2000.0&obs_epoch=2000.0&ra=10h00m24s&dec=%2B02d10m55s

Light travel time effects. Oct 9, 2022. 

Depending on the time-of-year an RR Ly is observed, there is a phase perturbation in the light curve due to solar system light travel time. Taking 16 minutes as maximum perturbation in arrival time, and a shortest RR Ly period of 0.2 days = 288 minutes, the phase perturbation is 360 deg * 16/288 = 20 degrees. That does sound well worth correcting so yes we should shift timing to solar system barycenter. A reference for this is here; Eastman_2010_PASP_122_935.pdf. An example python snippet that does the work is here: https://gist.github.com/mwcraig/38469d291643d0ae04dc

A possible Aux Tel observing program? 

RR Ly have short periods, like 0.5 day. We could use Aux tel to map out changes in spectrum over the course of an oscillation. Good target is star DN Aqr, ATLAS source 

RA = 

DEC = 

Galactic coordinates: 

Extinction along this line of sight is 0.058 in LSST r band. 

J2000 coords: 23 19 17.2012122276 -24 12 58.821938891

V=11.3, there are directions with no contamination stars:

12 arcmin box: