Project DR-03

Stacking of dithered / jittered / ”stared” spectra

The project addresses the stacking of spectra from the same observing block (OB, i.e. from the same observing sequence) or from different OBs (i.e. different observing sequences). The data combination should handle cosmic rays removal and match PSFs due to varying seeing conditions (e.g. Alard & Lupton 1998). UVES stared observations are an example of application. Generally all spectrograph data would benefit from stacking.
 

References 

• Robertson, J.G., 1986, Optimal Extraction of Single-Object Spectra from Observations with Two-Dimensional Detectors, PASP 98, 1220, Section III
• Alard & Lupton, 1998, A Method for Optimal Image Subtraction, ApJ, 503, 325
  
• A routine that efficiently combines spectra is HRS_MERGE, initially written for the combination of HST GHRS spectra: http://www.astro.washington.edu/docs/idl/cgi-bin/getpro/library43.html?HRS_MERGE

Applicable instruments and modes

In general this project applies to all ESO optical and infrared spectrographs. However, the stacking of dithered or jittered stared spectra project should put more emphasis on high-resolution spectroscopic data (e.g. UVES and X-Shooter). For instance, the combination of UVES spectra obtained with different instrument settings (involving different resolutions and seeing conditions) is not straightforward. This is particularly true in regions of overlap where the spectral slopes can be different depending on the accuracy of the flux calibration, or if no relative flux calibration is applied, as is common practice in absorption-line spectroscopy.

Required archive, calibration or observation data

As with imaging, a lot of observers split observations of a given target among different OBs. Data required for this project is are available in the ESO archive. The GOODS data, which has been optimally reduced by hand, provides a good benchmark.

 
Required algorithm developments

A number of pipelines provide algorithms for the stacking of data acquired within a template sequence. However, no method is currently provided for stacking data taken in different OBs, or possibly on different nights. Different algorithms and stacking strategies will be required depending on the relative brightness of the object and its background (first aligning/stacking and then extracting, or extracting first and stacking).

The main tasks required by this project are the following:

• Define an interface to the pipeline products, namely, the set of keywords and data structure that must be fulfilled for a pipeline product to be adequate for the task. One can assume that the input data for this task are pipeline reduced data, basically free of instrumental signature at the single-chip level.
• The task DR-03 will provide tools for the stacking of one-dimensional extracted spectra, and for the stacking of 2-dimensional spectra when the source can be identified. First priority shall be given to the 1D stacking.
• The stacking of 1-dimensional extracted spectra will involve regridding to the same wavelength grid, and co-adding the spectra with variance propagation. The corresponding CPL recipes will be instrument independent, and be integrated in a Reflex workflow.
• The stacking of 2-dimensional spectra performs the spectra combination into a mosaic, and copes with: 

• overlaps, multiple targets,
• global fringing correction,
• background matching,
• varying resolution in the case of wide slits,
• rejection of data based on seeing, image quality, or signal-to-noise criteria,
• propagation of errors

• Investigate the handling of crowded fields, and propose solutions to ensure the correct identification of targets in different exposures, possibly moved along the slit. The tool shall also support the interactive determination of the offsets.
• Implement the recipes in CPL, provide user documentation and Reflex workflows.

Required software developments

All software will be developed following the guidelines of the VLT-SPE-ESO-19000-1618 document. The software delivery will include in particular the CPL-based recipes, test reports, user documentation and Reflex workflows.

Validation

There is currently no package available against which the new recipes can be evaluated (TBC). The validation will mostly consist of comparing the results with interactive reduction, e.g. for the GOODS data or UVES stared observation data.