Focus, Guide and Alignment system for
DESI
Han Soul Lee
Florida Atlantic University Wayne State University Physics REU 2014 SLAC National Laboratory
Advisor: Dr. Kevin Reil
Dark Energy Spectroscopic Instrument(DESI), which is planned to be installed on the Kitt Peak Observatory, Tuscon, Arizona in 2018 is a project seeking to map the 3D model of the universe.The model is created through obtaining precise measurement the red shifts and Baryon acoustic oscillations from the astronomical object.Once the light is received to the instrument's fiber-tips,the spectrum is analyzed and the redshift by calculated. By obtaining the object's distance from earth, the instrument allows to model three dimensional model of the universe.This research was conducted for guide, focus and alignment system for DESI.
1. Background
The GFA (Guide, Focus and Alignment system) for DESI is consist of 10 cameras. The DESI focal surface consists of 10 “petals” each with 500 fiber tips and one GFA camera. Once the light enters the fiber, the spectrum is analyzed and obtains the wavelength which calculates redshift.  The camera will be running at ambient temperature, where the dark current noise is expected to be increased sharply with the temperature. Unfortunately, the CCD camera with the desired e2v sensor did not arrived in time. Rather an alternative CCD camera produced by Santa Barbara Instrument Group (SBIG) camera was used for test performance instead.  In addition to dark current studies, stellar density analysis images were taken from Blanco telescope in Chile. The Blanco telescope is an identical to Mayall telescope in Tuscon, Arizona which DESI will be mounted. A single exposure was taken at 8
different galactic latitudes (each exposure has62  sensors). Two of the  sensors (number two and sixty one) are known to have problems, and thus data from those sensors were excluded from the analysis process. To confirm Mayall telescope's ability to obtain accurate measures of  astronomical objects, exposures from Blanco's were used to compare the number of the star counts per galactic magnitudes to existing NOMAD catalog.
2. Procedure
Since60sensors took individual exposures for each parts of the galactic location, each location required total combined result of 60 images. In order to extract only stars from each images, Source Extractor (shortened as SExtractor) was used as a main program for the research. Once the sextractor had returned the output catalog, the catalog was once again filtered to leave only objects with higher possibilities of being stars. Using the updated catalog, a region file was created. The region file was then applied onto the applicable fits file to ds9. Each image was carefully examined through ds9 to confirm the most accurate results. Once the stars were extracted from each exposure, a MAG_AUTO histogram was created to compare the pattern with other exposures in the same extensions. A final histogram that total all the stars in each RA and DEC was later created. Ultimately, the number of stars per magnitude in each galactic latitude which was compared to the results found in NOMAD catalog.
3. Process
The dark current from the SBIG camera was measured from taking dark exposures in different temperature. The ranged from -12 to 20 celsius. Five minute exposures were taken for each temperature.  From the exposure, average pixel values
were measured and compared.
instrument的复数Sextractor version 2.8.6 was installed for stellar analysis purposes, which was later updated to version 2.19.7. Shell script was written to run sextractor on the individual images consecutively. The sextracted results were returned as a catalog for each individual images, which contained detected objects. The catalog also contained personalized parameter values, such as MAG_AUTO, Ellipticity and FLAG. The catalogs often included objects with variety of errors such as saturated and blended stars. The biggest challenge though the process was to omit the saturated stars from the catalog.
Figure 1. A black and round object on the top left corner is a saturated star
A black spot on the top left corner in figure 1 demonstrates an example of a saturated star. Saturation, caused by the over flow of photon in the in particular area appears as a black donuts with bright rim. Each of the images contained different saturation level which pixels arises saturation value.In order to assign correct saturation level for each image,sextractor required default.sex file with each personalize
d saturation level.To personalize default.sex script,which contained different required parameter values for sextraction, pyfits's header function was used to reach the individual saturation level in the header. Once the  saturated stars were omitted from the catalog, another issue was raised where the program started to include parts of the rim as an elongated individual objects. From figure 1, ellipses around the saturated star is located around the spot on the top left corner. On the
bottom right corner an elongated object (suspected to be a galaxy) is also included in the catalog as well. Stars are round gaseous objects. Since it appears as a round object, returned ellipticity was used to filter the objects with lower possibility to be stars. Equation for ellipticity follows:
Where A is the major axis and B is the minor axis of an ellipse. As B increases to the value of A, the limit of e reaches zero. Thus rounder the object, ellipticity closer to zero. After filtering the images for different values of ellipticity,0.25 returned the best results.
Figure 2. Illustrates an ellipse with 0.25 ellipticity
Figure 2, an eplise with ellipticity  0.25        Figure 3, list of flags value
In addition to the ellipticity, flags parameter, the value raised for any detection error was also used in filtering process. The filtration limit was set as 3. Any object with blending issue was eliminated from the analysis process. By eliminating objects with blending issue, saturation and elongation, approximate number of star count was obtained. Using the updated catalog, ds9 region file was created. The image of an exposure before and after filtration is shown below: