Spectroscopy is carried out using 1 of two codes depending on the instrument.

1. Spector is a package for reducing long-slit IDS data
2. For echelle spectra we use the CERES package from Rafael Brahm. Update these docs to include CAFE/FIES info?

Before running Spector:

1. chmod -R ugo+rwx * so Spector has permission to edit the files
2. Remove all files that should not be included in reduction i.e. bad flats, data from other central wavelengths or gratings etc. Place all this data in a folder called e.g. 'junk'. Use *.int logfile to help weed out unwanted files
3. Adjust the settings in the EDIT HERE section below

Assumptions:

1. This code was written for a particular observing strategy - Acquire, Arc, Science, Arc, repeat. Therefore the script is looking for this structure in the data when matching arcs to spectra. This can be easily changed by modifying the relevant section

What Spector does:

1. Prepares the working environment, making copies of original data
2. Strips multi extension fits to single extension
3. Renames files for easier viewing:
   1. i_* = integration (spectra)
   2. a_* = arc
   3. b_* = bias
   4. f_* = flat
4. Debiases and flat field corrects the data using CCDPROC in ASTROPY
5. Trims the spectra based on INT's heavily vignetted region
6. Calcuates mid exposure datetime in BJD_TDB, HJD and JD formats
7. Tidies up calibration frames into 'calibs' folder
8. Interactive spectral analysis with KPNOSLIT in IRAF
9. Normalises the spectra
10. Tidies up working directory after reductions are finished
11. Removes all intermediate data products to save disc space
12. Logs spectra information to the databse

What Spector does not do:

1. Flux calibration

What you will end up with:

1. A set of reduced, wavelength calibrated, normalised 1D spectra.

After reduction and extraction with Spector the files should be combined into one location for RV analysis. The files from multiple nights are linked by their SWASP\_ID. First populate the database column with each spectrum's SWASP\_ID using:

python database/setSwaspIds.py

Manually add any missing entries to the database and/or resolve naming or typos at the telescope. Add the ```--update`` flag to update the database

Next update the Q1 and Q2 keywords so they next observations can be planned:

python database/updateRvQuadStatus.py

use the --commit flag after checking the quads are ok. This will commit them to the database

Once all the SWASP_ID keywords have been updated and quads have been populated, gather all the spectra into per-object folders for analysis using:

python gatherSpectraFor_iSpec.py

This gets the unique list of targets from the database and puts all the spectra together in one folder per object ready for analysis with iSpec. Add the --copy flag to copy the data to the combined area.

Blends are spectra that contain two or more traces. The traces have to be asigned to the correct swasp_id. To do this we use the blends/correctBlands.py script. This plots each spectrum for a given object along with a finding chart set to the correct PA to identify the traces. This is a manual step. You must:

1. Run blends/correctBlends.py --filter\_blends --ds9
2. Make file1 containing the following on each line, in order: a. swasp_id (this is the main target ID from the header) a. image_id a. t1_swasp_id a. t2_swasp_id a. t3_swasp_id
3. Add any new blended objects to file2 containing: a. swasp_id a. pm_ra_mas_y a. pm_dec_mas_y a. period a. epoch a. V mag a. spectral_type a. status a. phase_coverage (boolean, 0 | 1) a. q1 (boolean, 0 | 1) a. q2 (boolean, 0 | 1) a. n_nites a. Then ingest those new objects into the eblm\_parameters mysql table using the ingestBlendCompanions.py script.

File1 is then used to split up the mulitspec files into per object single 1D spectra. From there we can run iSpec to get RVs as normal.

After a map file has been created (file1 above) we must cycle through it object by object, splitting the spectra in steps. This is because PyRAF does not like to change directorys for some reason. So, if the current object to split has 5 spectra, run:

$> python correctBlendIds.py --split\_blends --split\_apply --split\_length 5

Without --split\_apply the split will only be simulated. Where --filter\_blends was used above to display the objects and aid in disentangling the blends, --split\_blends is used to then apply disentanglement from the blend map. --split\_length is required or nothing happens.

Afetr splitting, the original spectra with more than 1 trace have analyse=-1 so we know not to look at them directly. They are also removed from each per-object folder, leaving only the extracted 1D spectra there. A new row is added to the database table for each 1D spectrum that is split out. They have some information from the original table row, and more information specific to that object (e.g. the right swasp_id).

undoBlendCollection.py splits the multispec files from all_blended back into their per object folders. The rows containing split files must also be removed manually from the database to undo everything.

The database/checkAllSpectra.py script will compare what is in the all_spectra folder and what is in the database and report any differences. It is good to sanity check there are no complaints before starting the RVs.

To use the right CCF mask with iSpec we need to estimate the spectral type of the target. This can be done by getting the Teff_jh and Teff_vk from Hunter and then using the Teff-to-Spectral_Type translation to estimate the type based on the average of the two temperatures. This is not super accurate but should be enough to distinguish F from K or M stars.

Loop over all the objects with the iSpec_RVs.py script. This will cross correlate spectra with the first one to determine the RVs. A second cross correlation is also done with the first spectrum as a double check.

By default iSpec_RVs.py will loop over all the objects that have spectra that have not had their RV measured. You can force it to analyse a given object by specifiying --swasp_id.

The barycentric correction is not the most accurate, but this is ok for initial analysis. The BCV should be calculated accurately before any detailed analysis.

This is now done by updateRvQuadStatus.py above. Run this after each reduction run to always have the correct current state of the world for the project

create table eblm_ids (
image_id                      varchar(40) not null primary key,
swasp_id                      varchar(40),
object_name                   varchar(20) not null,
night                         date,
utmiddle                      datetime,
hjd_mid                       double,
bjd_mid                       double,
jd_mid                        double,
n_traces                      int(2),
sky_pa                        double,
ccf_height                    double,
ccf_fwhm                      double,
atomic_velocity               double,
atomic_velocity_err           double,
telluric_velocity             double,
telluric_velocity_err         double,
barycentric_velocity_iSpec    double,
barycentric_velocity_exo      double,
comment                       varchar(40),
analyse                       tinyint(1)
);

James McCormac