Identifying High Redshift Dropouts Jeffrey Gerber Physics REU, University of California at Davis, Davis, CA, 95616 Dr. Benjamin Cain and Dr. Marusa Bradac Cosmology Department, University of California at Davis, Davis, CA, 95616 (Dated: August 29, 2011) Abstract A high redshift dropout is an object located at a high redshift distance that shows up in red filters, but disappears in blue filters. In this experiment, images from the Hubble Space Telescope Wield Field Camera 3 were used to attempt to locate high redshift dropouts. The objects attempting to be found were at a redshift of z-7 or greater if possible. Z-7 dropouts were expected to show up in filters showing 1050 nanometer wavelength and greater, and then disappear in filters showing a wavelength of 850 nanometers and less. 1
I. INTRODUCTION High redshift dropouts are objects located about thirteen billion light years away (the edge of the known universe). They are classified as objects that are z-7 or greater with z being the measure of redshift. These objects are a great interest to cosmology because of the data that they could provide us about the properties of galaxies about 700 million years after the big bang. In order to see objects this far with more clarity, we used a galaxy cluster in order to take advantage of gravitational lensing. Gravitational lensing focuses the light as it approaches Earth and is captured by the Hubble Space Telescope. It has been proven to find objects in the same positions as they would be seen in the Deep Field Survey, but they have a greater clarity. 1 1. Redshift: Redshift is a version of the Doplar Effect that is commonly seen in astronomy. Redshift occurs because the universe is expanding. This expansion means that objects that we view are moving away from us as we observe them. As they move away, the light waves behind them gain a higher wavelength than their original emission wavelength. Because the light waves we see now have a greater wavelength (or lower frequency) than their original, the light appears redder than it was when it was emitted from the object. We can use this redshift value to tell how far away the object is from us. The higher the value for redshift than the farther away the object. Below is the equation for calculating redshift using the light s wavelength. It can also be found in a similar way using the frequency of the light. z = λ obsv λ emit λ emit (1) 2. Dropouts: A dropout is an object who emits a high wavelength of light, but nothing lower than a certain point. In our case, the z-7 objects did not emit light a wavelength below 850 nanometers because of the large amounts of hydrogen present at that distance from the Earth. The hydrogen absorbs all of the energy from the light below 850 nm, which 2
causes the object to dropout. Dropouts can be looked for by having multiple filters take images of the same object. Then the observer can compare the object in all of these filters to see whether or not the object existed and then disappeared. In order to look for higher redshift dropouts, one would simply expect them to dropout at a filter higher than 850 nm. For example, a z-8 dropout would not appear in a filter set to 1050 nm. 3. Gravitational Lensing: Gravitational lensing occurs when an object with a large amount of mass, such as a galaxy cluster, bends the light waves from objects as they pass through. This turns the cluster into a large magnifying glass in space that can be used to focus the light from objects behind it. Gravitational lensing allows for better clarity of images that are extremely far away. Because we were working on identifying images at such large distances, we used gravitational lensing to achieve better image data. II. DATA USED The images that were used to look for dropouts come from the Hubble Wield Field Camera 3 (wfc3), and both the IR and UVIS cameras were used. Redder filters were used to detect objects, and then bluer filters were used to see if the objects had dropped out or not. The red filters or detection filters were 1050, 1100, 1250, 1400, and 1600 nm. The blue filters were 435, 606, 625, 775, 815, and 850 nm. III. ANALYSIS OF IMAGES The images we were looking at from the wfc3 have anywhere from 1000 to 1500 objects in them as you can see from the.jpg s from the IR and UVIS cameras (figures 1 and 2, respectively). These are far too many objects for one person to scan through and see whether or not they dropout of a filter one by one. This means that most of my work went into automating the process of figuring out what objects were potential dropouts and what objects were not. The.fits images from the Hubble Telescope had already been collected. The first step was to analyze the images by identifying the objects in them. In order to do this for multiple 3
sets of images from multiple filters, a bash script was used to run Source Extractor multiple times. Source Extractor (Sextractor for short) is a program that scans a.fits image and creates a catalog of all the objects in the image. The catalogs can output different values for the object such as R.A and Dec., the x and y position in the image, the magnitude of the object, flux of the object, etc based on the parameters that the program is fed. 2 We mainly used Source Extractor to find the magnitude of an object in a filter and its R.A and Dec. Another helpful use for Source Extractor is that it can run two images at the same time, using one for detecting objects and another for measuring their magnitudes. This was helpful for detecting dropouts because Source Extractor could locate an object using a detection filter, and then measure that same spot in another filter and see if the object was still there. If the object did not show up, it would come across as a 99 magnitude measurement in the catalog created. The magnitude zeropoints that we used for measuring were from the Hubble ABmag zeropoints for the wfc3 cameras. The next step to cutting down the objects found to a list that one could easily go through by hand was to combine all of the catalogs that were created by the bash loop running through Sextractor. In order to do this, a c program was used that the bash script ran after running the images through Sextractor. The c program took input from multiple catalogs, and then printed them as one catalog for a detection filter with the magnitude and magnitude errors from every filter listed based on the position of the object. The program also corrected the magnitudes based on an aperture error that had been found earlier. Another adjustment that had to be made to the catalogs created was the problem that the images from different filters are different sizes. The IR filters cover much larger areas than the UVIS filters, which can be seen in the IR and UVIS.jpg s (figures 1 and 2). Some images also show different areas because of the fact that the Hubble Telescope is rotating in space as it takes images. This means that Sextractor could be identifying objects as missing because the image did not cover the same area rather than actually not show up in space. This problem was fixed by creating a region file that covered the area that all the images had in common in ds9, which was the program used to look at the.fits images. A perl program was then run on the catalogs that masked the objects that did not show up in the region file. An example of a region file can be seen in figure 3. The region file is the green line in all of the images. The green that can be seen around the star was used to exclude points picked up in that region that were caused by noise spikes from the star. 4
Now that everything was taken into account to adjust the catalogs appropriately, cuts needed to be made to reduce the list based on whether or not objects were possible dropouts. Objects were chosen as dropouts based on color plots that plotted J-H vs. z-j (z being 850 nm, J being 1100 nm, and H being 1600 nm). An example of these plots are shown in figure 4. The criteria that the objects had to meet were a color of z-j greater than 4 and a color of J-H less than 1. It was determined that anything less than 4 for z-j was not faint enough in the z band (850 nm) to be considered a dropout. The J-H criteria was set based on the fact that anything that faint in the J band compared to the H band was an object that was not being seen very well due to dust in the galaxy cluster used for the gravitational lensing. A c program was also written to input the masked catalogs and then output a catalog of possible dropouts based on whether the objects matched the color criteria or not. Once the lists of possible dropouts had been made, the objects had to be looked at one by one to identify whether they were an actual dropout or something that was not a dropout, but had made the color cuts anyway. Examples of these include objects that dimmed in blue filters, but did not dropout (as seen in figure 5), an object that did not show up in the 850 nm filter, but reappeared in the other blue filters (as seen in figure 6), or noise that Sextractor detected as an object that did not show up in blue filters. In order to make objects easier to distinguish from background noise, the images were shown in color, and region files were created that circled every point identified as a dropout. A bash script was used to run through the list of possible dropouts and then sort into catalogs of dropouts and not dropouts. IV. RESULTS Overall five sets of images were looked at from the wfc3, which were A383, A2261, MACS1149, MACS1206, and MACS2129. An average of about 200 objects from each set matched the color cuts. From these, about 20-30 objects were identified as definite z-7 dropouts. An example of a dropout from the MACS1149 data can be seen in figure 7. Some of the objects were identified as possible z-8 dropouts as they disappeared in the 1050 nm filter as well. Right now, some discrepancies were found between the dropout lists I created and ones created by Dr. Cain. These discrepancies will continue to be studied to see why they were caused. 5
V. FURTHER RESEARCH Further research that would be conducted on these objects would include closer observations of the objects to make sure that they are dropouts. An emission spectrum would be observed to ensure that the object does indeed dropout when it should. Z-7 dropouts could also show us information about how galaxies were when they were formed right after the Big Bang. The reason dropouts occur at this distance is that there are large amounts of hydrogen in that period of time. This hydrogen absorbs the blue light from objects causing them to dropout. Research on these objects is being conducted to try and discover what causes this large amount of hydrogen. In general, there is a lot of information that we can gain from studying objects this far in the universe. ACKNOWLEDGMENTS I d like to thank Dr. Benjamin Cain and Dr. Marusa Bradac for their help on conducting this research this past summer. I d also like to thank UC Davis for the opportunity to work on research, and the NSF for founding the REU. Also at a second address here if needed 1 Hall, Nicholas, et al. 2011 Using the Bullet Cluster as a Gravitational Telescope to Study z 7 Lyman Break Galaxies 2 Emmanuel Bertin, Astromatic.net http://www.astromatic.net/software/sextractor. 6
FIGURES FIG. 1. IR JPG Image 7
FIG. 2. UVIS JPG Image FIG. 3. Example of a Region File 8
80 All Detection Filters 60 40 20 Z-J 0 20 40 60 80 80 60 40 20 0 20 40 60 80 J-H FIG. 4. Example Color Plot FIG. 5. Example of an Object that did not Dropout 9
FIG. 6. Example of a Noise Spike FIG. 7. Example of a Dropout 10