4. Example 1-B: Dithered Point Source Longslit - Using the “Reduce” class

A reduction can be initiated from the command line as shown in Example 1-A: Dithered Point Source Longslit - Using the “reduce” command line and it can also be done programmatically as we will show here. The classes and modules of the RecipeSystem can be accessed directly for those who want to write Python programs to drive their reduction. In this example we replicate the command line reduction from Example 1-A, this time using the Python interface instead of the command line. Of course what is shown here could be packaged in modules for greater automation.

4.1. The dataset

If you have not already, download and unpack the tutorial’s data package. Refer to Downloading the tutorial datasets for the links and simple instructions.

The dataset specific to this example is described in:

Here is a copy of the table for quick reference.


S20171022S0087,89 (515 nm)
S20171022S0095,97 (530 nm)

Science biases


Science flats

S20171022S0088 (515 nm)
S20171022S0096 (530 nm)

Science arcs

S20171022S0092 (515 nm)
S20171022S0099 (530 nm)

Standard (LTT2415)

S20170826S0160 (515 nm)

Standard biases


Standard flats

S20170826S0161 (515 nm)

Standard arc

S20170826S0162 (515 nm)

4.2. Setting up

First, navigate to your work directory in the unpacked data package.

cd <path>/gmosls_tutorial/playground

The first steps are to import libraries, set up the calibration manager, and set the logger.

4.2.1. Importing libraries

1import glob
3import astrodata
4import gemini_instruments
5from recipe_system.reduction.coreReduce import Reduce
6from recipe_system import cal_service
7from gempy.adlibrary import dataselect

The dataselect module will be used to create file lists for the darks, the flats and the science observations. The cal_service package is our interface to the local calibration database. Finally, the Reduce class is used to set up and run the data reduction.

4.2.2. Setting up the logger

We recommend using the DRAGONS logger. (See also Double messaging issue.)

8from gempy.utils import logutils

4.2.3. Set up the Local Calibration Manager

DRAGONS comes with a local calibration manager and a local, light weight database that uses the same calibration association rules as the Gemini Observatory Archive. This allows the Reduce instance to make requests for matching processed calibrations when needed to reduce a dataset.

Let’s set up the local calibration manager for this session.

In ~/.geminidr/, edit the configuration file rsys.cfg as follow:

standalone = True
database_dir = <where_the_data_package_is>/gmosls_tutorial/playground

This tells the system where to put the calibration database, the database that will keep track of the processed calibration we are going to send to it.


The tilde (~) in the path above refers to your home directory. Also, mind the dot in .geminidr.

The calibration database is initialized and the calibration service is configured like this:

10caldb = cal_service.CalibrationService()

The calibration service is now ready to use. If you need more details, check the “caldb” documentation in the Recipe System User Manual.

4.3. Create file lists

The next step is to create input file lists. The module dataselect helps with that. It uses Astrodata tags and descriptors to select the files and store the filenames to a Python list that can then be fed to the Reduce class. (See the Astrodata User Manual for information about Astrodata and for a list of descriptors.)

The first list we create is a list of all the files in the playdata directory.

15all_files = glob.glob('../playdata/*.fits')

We will search that list for files with specific characteristics. We use the all_files list as an input to the function dataselect.select_data() . The function’s signature is:

select_data(inputs, tags=[], xtags=[], expression='True')

We show several usage examples below.

4.3.1. Two lists for the biases

We have two sets for biases: one for the science observation, one for the spectrophotometric standard observation. The science observations and the spectrophotometric standard observations were obtained using different regions-of-interest (ROI). So we will need two master biases, one “Full Frame” for the science and one “Central Spectrum” for the standard.

To inspect data for specific descriptors, and to figure out how to build our dataselect expression, we can loop through the biases and print the value for the descriptor of interest, here detector_roi_setting.

17all_biases = dataselect.select_data(all_files, ['BIAS'])
18for bias in all_biases:
19    ad = astrodata.open(bias)
20    print(bias, '  ', ad.detector_roi_setting())
../playdata/S20170825S0347.fits    Central Spectrum
../playdata/S20170825S0348.fits    Central Spectrum
../playdata/S20170825S0349.fits    Central Spectrum
../playdata/S20170825S0350.fits    Central Spectrum
../playdata/S20170825S0351.fits    Central Spectrum
../playdata/S20170826S0224.fits    Central Spectrum
../playdata/S20170826S0225.fits    Central Spectrum
../playdata/S20170826S0226.fits    Central Spectrum
../playdata/S20170826S0227.fits    Central Spectrum
../playdata/S20170826S0228.fits    Central Spectrum
../playdata/S20171021S0265.fits    Full Frame
../playdata/S20171021S0266.fits    Full Frame
../playdata/S20171021S0267.fits    Full Frame
../playdata/S20171021S0268.fits    Full Frame
../playdata/S20171021S0269.fits    Full Frame
../playdata/S20171023S0032.fits    Full Frame
../playdata/S20171023S0033.fits    Full Frame
../playdata/S20171023S0034.fits    Full Frame
../playdata/S20171023S0035.fits    Full Frame
../playdata/S20171023S0036.fits    Full Frame

We can clearly see the two groups of biases above. Let’s split them into two lists.

21biasstd = dataselect.select_data(
22    all_files,
23    ['BIAS'],
24    [],
25    dataselect.expr_parser('detector_roi_setting=="Central Spectrum"')
28biassci = dataselect.select_data(
29    all_files,
30    ['BIAS'],
31    [],
32    dataselect.expr_parser('detector_roi_setting=="Full Frame"')


All expressions need to be processed with dataselect.expr_parser.

4.3.2. A list for the flats

The GMOS longslit flats are not normally stacked. The default recipe does not stack the flats. This allows us to use only one list of the flats. Each will be reduced individually, never interacting with the others.

34flats = dataselect.select_data(all_files, ['FLAT'])

4.3.3. A list for the arcs

The GMOS longslit arcs are not normally stacked. The default recipe does not stack the arcs. This allows us to use only one list of arcs. Each will be reduce individually, never interacting with the others.

35arcs = dataselect.select_data(all_files, ['ARC'])

4.3.4. A list for the spectrophotometric standard star

If a spectrophotometric standard is recognized as such by DRAGONS, it will receive the Astrodata tag STANDARD. To be recognized, the name of the star must be in a lookup table. All spectrophotometric standards normally used at Gemini are in that table.

36stdstar = dataselect.select_data(all_files, ['STANDARD'])

4.3.5. A list for the science observation

The science observations are what is left, anything that is not a calibration or assigned the tag CAL.

First, let’s have a look at the list of objects.

37all_science = dataselect.select_data(all_files, [], ['CAL'])
38for sci in all_science:
39    ad = astrodata.open(sci)
40    print(sci, '  ', ad.object())

On line 37, remember that the second argument contains the tags to include (tags) and the third argument is the list of tags to exclude (xtags).

../playdata/S20171022S0087.fits    J2145+0031
../playdata/S20171022S0089.fits    J2145+0031
../playdata/S20171022S0095.fits    J2145+0031
../playdata/S20171022S0097.fits    J2145+0031

In this case we only have one target. If we had more than one, we would need several lists and we could use the object descriptor in an expression. We will do that here to show how it would be done. To be clear, the dataselect.expr_parser argument is not necessary in this specific case.

41scitarget = dataselect.select_data(
42    all_files,
43    [],
44    ['CAL'],
45    dataselect.expr_parser('object=="J2145+0031"')

4.4. Master Bias

We create the master biases with the Reduce class. We will run it twice, once of each of the two raw bias lists, then add the master biases produced to the local calibration manager with the caldb instance. The output is written to disk and its name is stored in the Reduce instance. The calibration service expects the name of a file on disk.

47reduce_biasstd = Reduce()
48reduce_biassci = Reduce()

The two master biases are: S20170825S0347_bias.fits and S20171021S0265_bias.fits.


The file name of the output processed bias is the file name of the first file in the list with _bias appended as a suffix. This is the general naming scheme used by the Recipe System.

4.5. Master Flat Field

GMOS longslit flat fields are normally obtained at night along with the observation sequence to match the telescope and instrument flexure. The matching flat nearest in time to the target observation is used to flat field the target. The central wavelength, filter, grating, binning, gain, and read speed must match.

Because of the flexure, GMOS longslit flat field are not stacked. Each is reduced and used individually. The default recipe takes that into account.

We can send all the flats, regardless of characteristics, to Reduce and each will be reduce individually. When a calibration is needed, in this case, a master bias, the best match will be obtained automatically from the local calibration manager.

56reduce_flats = Reduce()
58reduce_flats.mode = 'ql'
61for f in reduce_flats.output_filenames:
62    caldb.add_cal(f)


GMOS longslit reduction is currently available only for quicklook reduction. The science quality recipes do not exist, hence the use of the ql mode to activate the “quicklook” recipes.

4.6. Processed Arc - Wavelength Solution

GMOS longslit arc can be obtained at night with the observation sequence, if requested by the program, but are often obtained at the end of the night instead. In this example, the arcs have been obtained at night, as part of the sequence.

Like the spectroscopic flats, they are not stacked which means that they can be sent to reduce all to together and will be reduced individually.

The wavelength solution is automatically calculated and the algorithm has been found to be quite reliable. There might be cases where it fails; inspect the *_mosaic.pdf plot and the RMS of determineWavelengthSolution in the logs to confirm a good solution.

63reduce_arcs = Reduce()
65reduce_arcs.mode = 'ql'
68for f in reduce_arcs.output_filenames:
69    caldb.add_cal(f)


Failures of the wavelength solution calculation are not easy to fix in quicklook mode. It might be better to simply not use the arc at all and rely on the approximate solution instead. When the science quality package is released, there will be interactive tools to fix a bad solution. Remember, this version only offers quicklook reduction for GMOS longslit.

4.7. Processed Standard - Sensitivity Function

The GMOS longslit spectrophotometric standards are normally taken when there is a hole in the queue schedule, often when the weather is not good enough for science observations. One standard per configuration, per program is the norm. If you dither along the dispersion axis, most likely only one of the positions will have been used for the spectrophotometric standard. This is normal for baseline calibrations at Gemini. The standard is used to calculate the sensitivity function. It has been shown that a difference of 10 or so nanometers does not significantly impact the spectrophotometric calibration.

The reduction of the standard will be using a master bias, a master flat, and a processed arc. If those have been added to the local calibration manager, they will be picked up automatically.

70reduce_std = Reduce()
72reduce_std.mode = 'ql'

To inspect the spectrum:

76from gempy.adlibrary import plotting
77import matplotlib.pyplot as plt
79ad = astrodata.open(reduce_std.output_filenames[0])
81plotting.dgsplot_matplotlib(ad, 1)

To learn how to plot a 1-D spectrum with matplotlib using the WCS from a Python script, see Tips and Tricks Plot a 1-D spectrum.

The sensitivity function is stored within the processed standard spectrum. To learn how to plot it, see Tips and Tricks Inspect the sensitivity function.

4.8. Science Observations

The science target is a DB white dwarf candidate. The sequence has four images that were dithered spatially and along the dispersion axis. DRAGONS will register the four images in both directions, align and stack them before extracting the 1-D spectrum.


In this observation, there is only one source to extract. If there were multiple sources in slits, regardless of whether they are of interest to the program, DRAGONS will locate them, trace them, and extract them automatically. Each extracted spectrum is stored in an individual extension in the output multi-extension FITS file.

This is what one raw image looks like.

raw science image

With the master bias, the master flat, the processed arcs (one for each of the grating position, aka central wavelength), and the processed standard in the local calibration manager, to reduce the science observations and extract the 1-D spectrum, one only needs to do as follows.

83reduce_science = Reduce()
85reduce_science.mode = 'ql'

This produces a 2-D spectrum (S20171022S0087_2D.fits) which has been bias corrected, flat fielded, QE-corrected, wavelength-calibrated, corrected for distortion, sky-subtracted, and stacked. It also produces the 1-D spectrum (S20171022S0087_1D.fits) extracted from that 2-D spectrum. The 1-D spectrum is flux calibrated with the sensitivity function from the spectrophotometric standard. The 1-D spectra are stored as 1-D FITS images in extensions of the output Multi-Extension FITS file.

This is what the 2-D spectrum looks like.

87display = Reduce()
88display.files = ['S20171022S0087_2D.fits']
89display.recipename = 'display'
2D stacked spectrum

The apertures found are list in the log for the findApertures just before the call to traceApertures. Information about the apertures are also available in the header of each extracted spectrum: XTRACTED, XTRACTLO, XTRACTHI, for aperture center, lower limit, and upper limit, respectively.

This is what the 1-D flux-calibrated spectrum of our sole target looks like.

91from gempy.adlibrary import plotting
92import matplotlib.pyplot as plt
94ad = astrodata.open(reduce_science.output_filenames[0])
96plotting.dgsplot_matplotlib(ad, 1)
1D spectrum

To learn how to plot a 1-D spectrum with matplotlib using the WCS from a Python script, see Tips and Tricks Plot a 1-D spectrum.

If you need an ascii representation of the spectum, you can use the primitive write1DSpectra to extract the values from the FITS file.

 98writeascii = Reduce()
 99writeascii.files = ['S20171022S0087_1D.fits']
100writeascii.recipename = 'write1DSpectra'

The primitive outputs in the various formats offered by astropy.Table. To see the list, use showpars from the command line.

showpars S20171022S0087_1D.fits write1DSpectra

To use a different format, set the format parameters.

102writeascii = Reduce()
103writeascii.files = ['S20171022S0087_1D.fits']
104writeascii.recipename = 'write1DSpectra'
105writeascii.uparms = [('format', 'ascii.ecsv'), ('extension', 'ecsv')]