I. Multi-waveband Galaxy Surveys and Measurement of Physical Parameters in Galaxies

I have been a core member of the teams designing and performing large multi-waveband galaxy surveys, using the greatest of the space- and ground-based observatories. These include: Great Observatories Origins Deep Survey (GOODS); Hubble Ultra Deep Field (HUDF); Cosmic Evolution Survey (COSMOS); Coma Cluster ACS Treasury Program and Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey  (CANDELS).  I have had a leading role in measuring photometric redshifts, stellar masses, rest-frame colors, dust extinction and star formation rates for galaxies in these surveys. I also lead the spectroscopic follow-up observations for galaxies in these fields, using the DEIMOS instrument on the Keck Telescopes. In the following, I summarize my recent projects.


I.1). Photometric Redshift Measurement for Deep Galaxy Surveys

Fig1.Comparison between the photometric redshifts estimated from the photometric data in GOODS-South galaxies and their spectroscopic counterparts. The accuracy in the estimated photometric redshifts is: (zphot – zspec)/(1+zspec) = 0.03, with an outlier fraction of 2%.


For many years, I led measurement of photometric redshifts to galaxies in the GOODS and COSMOS fields. The first science papers from the GOODS and COSMOS data were based, to a large extent, on my photometric redshifts (Giavalisco et al 2004; Scoville et al 2007). Using the available spectroscopic redshifts, we calibrated the photometric redshifts and measured them to an accuracy of (zphot – zspec)/(1+zspec) = 0.03 for GOODS (Dahlen, Mobasher et al 2010) and COSMOS (Ilbert et al 2010)-(Fig 1). The work is currently in progress to extend this to other galaxy surveys with available near-infrared (WFC3) data from the Hubble Space Telescope (HST)- ie. the CANDELS fields.


I.2). Comprehensive Test of Photometric Redshift and Stellar Mass


Fig2a. A selected example of the test of reliability of different photometric redshift methods. The same photometric data are used for all tests, with the results compared with the spectroscopic redshifts (from Dahlen, Mobasher et al 2012)
Fig2b. Selected example of the comparison between stellar mass estimates from different methods in the mock catalogue with the input (expected) mass (from Mobasher, Dahlen et al 2012)

I led (with Dr. Tomas Dahlen) a comprehensive test of different methods used to measure photometric redshifts and stellar masses of galaxies. The aim of this exercise was to understand why different techniques using exactly the same data resulted in different  photometric redshifts and stellar masses? To accomplish this, we provided photometric data for a sample of galaxies with known spectroscopic redshifts, to 13 different experienced groups and asked them to predict both the photometric redshift and stellar mass values for each of the galaxies in the sample. The spectroscopic redshifts were not provided and the test was completely blind. While one could assess the accuracy of the photometric redshifts by comparing them with their spectroscopic counterparts, this is not possible for fainter galaxies (for which it is hard to obtain spectroscopic redshifts) and for their stellar mass (which cannot be obtained by direct observations). Therefore, a mock catalogue was also generated with known photometry, input redshifts and stellar masses for individual galaxies. The results from different groups were provided to us, where they were compared with their expected spectroscopic and input mock values (Figs 2a & 2b). This is the most comprehensive test of its kind, with the results for the photometric redshift (Dahlen, Mobasher et al 2012) and stellar mass (Mobasher et al 2014) analyzed.



I.3). Hubble Legacy Archival Program

Fig3. Depth vs. area for some of the HST/ACS galaxy surveys in fields and clusters. Self-consistent multi-wavelength galaxy catalogues will be generated, containing all the physical parameters measured for galaxies in these fields.

I am the PI on a successful Legacy HST Archival proposal to generate self-consistent multi-waveband photometric catalogues for galaxies in all the surveys in field and clusters, performed by the Advanced Camera for Surveys (ACS) on-board the HST (Figure 3). For each survey, the multi-waveband imaging data from different ground- and space-based observatories are scaled to that with the highest resolution (i.e. ACS), using the Template FITting (TFIT) technique (Laidler et al 2007). While retaining the high spatial resolution (and hence, avoid blending), this allows self-consistent photometry across a range of wavelengths from UV to mid-infrared. For each galaxy in these catalogs, the physical parameters are measured, using their SEDs. The final product is self-consistent photometric catalogs containing all the ground- and space-based data, reduced to the same scale, with photometric redshifts, stellar masses, rest-frame colors, star formation rates and extinctions. This will be an extremely useful data set for the community to use as, for the first time, the data taken from different surveys can be combined and analyzed.


I.4). Near-infrared Survey of the GOODS-North Field 

We have completed a near-infrared (J & Ks band) survey of the GOODS-North field, using the available imaging data from the MOIRCS instrument on 8.2m Subaru telescope and WIRCam instrument on the 3.6m Canada-France-Hawaii Telescope (CFHT)- (with my Post-Doc Nimish Hathi). This fills a serious wavelength gap in this field. We combined the MOIRCS and WIRCam data and generated the J & Ks band images, covering the entire GOODS-North field (~169 arcmin2) to an AB magnitude limit of ~25 mag (3sigma)- (Hathi, Mobasher et al 2012). We reduced and stacked CFHT and WIRCam data. This is the deepest K-band data currently available for the GOODS-N field. Combined with the deep WFC3 images of this field (from the CANDELS project), we now have unparalleled multi-waveband data used to measure the photometric redshifts for galaxies in this field and to search for Lyman Break Galaxies (LBGs) at z > 6.5.


II). The Science

In this section I present a short description of my main research projects. These are mostly based on the data catalogues I contributed to generate, as discussed in section I. I only discuss the projects in which I have had a leading role.

(II.1). Search for a New Populations of Galaxies

In the Cold Dark Matter (CDM) scenario for formation of galaxies, the present-day galaxies are expected to form by mergers of smaller sub-units throughout the age of the Universe. If so, we do not expect to find a large population of massive and evolved systems at high redshifts (i.e. large look-back times). To test this, we developed a technique, using the strength of Balmer Break features (lrest= 3536 A), which is directly correlated with the age of the stellar population, to select evolved galaxies-ie. galaxies with strong Balmer features  (Wiklind et al 2008). By identifying these features, shifted to near-IR wavebands (> 1.2 mm), we select evolved galaxies at high redshifts (3 < z < 7). The subset of these galaxies with bright rest-frame infrared luminosities, are expected to also be massive, as rest-frame infrared light of galaxies is directly proportional to their stellar mass. Using these (color) criteria, we selected a population of massive and evolved galaxies at 3 < z < 5 and 5 < z < 7 (Balmer Break Galaxies: BBGs) in both GOODS and COSMOS fields (Figure 4). A high number density for the BBGs runs counter to expectations from the CDM scenario, and suggests much more rapid galaxy growth at high-z (Nayyeri, Mobasher et al 2014).


(II.2). Nature of High Redshift Galaxies

Fig4. Near-infrared color-color diagram used to select the BBGs. The lines are model predictions for post-starburst (black), starburst (blue) and elliptical (red) galaxies in the redshift range 3 < z < 5. The grey area is the expected location of massive evolved (post-starburst) systems in this redshift range.

We have identified a large and complete population of the Lyman Break Galaxies (LBG) and Lyman Alpha Emitters (LAE) in GOODS and COSMOS fields at 3 < z < 6.5(Fig 5). The star formation rate, stellar mass and dust content in these galaxies are measured using their SEDs and are used to study the nature of these systems and differences between  populations of high-z galaxies (LBGs, LAEs, BBGs) and their dependence on different selection techniques (Capak, Mobasher et al 2011; Ouchi, Mobasher et al 2009). The LGBs and LAEs appear to be more typical and less massive than the BBGs (by selection). Extensive spectroscopic observations of these galaxies have been performed with Keck/DEIMOS.

Fig5. Shows the multi-waveband images of z > 6 candidates in the COSMOS field.








(II.3). SpectroscopicConfirmation of a Galaxy at z ~ 7.2

In an extensive and deep survey of the GOODS-N field, led by Massami Ouchi (University of Tokyo), we identified a number of candidates at z > 6.5 (Ouchi, Mobasher et al 2009). Follow-up spectroscopy of these sources with Keck/DEIMOS confirmed the redshifts for many of these galaxies to be at z > 6.8. In the course of this study we discovered one of the most distant galaxy with a confirmed spectroscopic redshift of z=7.2 (Ono, Ouchi, Mobasher, Dickinson et al 2011). After a total of 18 hours exposure on this galaxy, using Keck/DEIMOS, we confirmed its spectroscopic redshift (Fig 6). This provides a first step in quantifying the escape of Lyman alpha photons from galaxies and the role of such systems in the re-ionization of the Universe.

Fig6. - Shows the Keck/DEIMOS spectrum of a high-z candidate with confirmed spectroscopic redshift of z=7.21. This is one of the highest redshift galaxies confirmed with spectroscopic observations (Mobasher et al 2012).


(II.4). Spectroscopic Observations of High-z Galaxies

I am leading a program of spectroscopy of high-z candidates in the CANDELS fields (with S. Faber, D. Koo and M. Dickinson), using Keck/DEIMOS. The targets consist of the BBGs, LBGs and LAEs for which we have successfully confirmed spectroscopic redshifts (Fig 7). Comparison between the spectra of the LBG and BBG populations, using the stacked spectra at a given redshift, reveals differences in the nature of these galaxies. For the massive and evolved population (BBGs) at 3 < z < 7, changes with redshift in their number density, constrains the hierarchical models for formation of galaxies (Nayyeri, Mobasher et al 2014). The spectroscopic redshifts are used to calibrate photometric redshifts (Dahlen, Mobasher et al 2010), find the fraction of the LBGs with lyman alpha emission and the evolution of Lyman alpha spectral shape (ie. Equivalent Width and asymmetry) with redshift (Hemmati et al 2013).


Fig7. Examples of the Keck/DEIMOS spectra for high-z galaxies confirmed to be at z=6.965 and 6.844 (From Ono, Ouchi, Mobasher et al 2011).


(II.5). Large Scale Structure in COSMOS


Fig8. - The diagram shows an example of a sigma map for a cluster at z=0.738, identified in the COSMOS field, using the technique discussed in B. Darvish et al (2012). There are over 100 structures (with different richness degrees) in the COSMOS field.


Using the photometric data in the COSMOS field, we have identified real groups and clusters in this field (with Nick Scoville and graduate student Behnam Darvish). A “density” parameter is assigned to each galaxy based on the number of its closest neighbors, its photometric redshift and redshift probability distribution (the probability the photometric redshift has a given value). A catalogue of groups and clusters for the entire COSMOS field is generated at redshifts 0.2 < z < 1.2 (Fig 8.) and agree very well with their X-ray maps. We are studying the evolution with redshift of the rest-frame color-magnitude relation, the red sequence and mass-SFR in these structures and their environmental dependence.





(II.6). Spectroscopic Confirmation of the Most Distant Cluster

Fig9. - Shows the image of a proto-cluster at z~5.29. The area corresponds to 22.5” x 22.5” on the sky (0.865 Mpc co-moving or 0.137 Mpc proper distance at z=5.298). The six optically bright objects with spectroscopic redshifts are shown. The galaxy labeled starburst is the optical counterpart of a sub-mm detected source (from Capak et al 2011)


I lead a Keck/DEIMOS program to perform spectroscopy of candidates in proto-clusters identified at the highest redshifts (z=4.5 and 5.7) in the COSMOS field (with P. Capak). The clusters were discovered using broad- and narrow-band photometric data (Capak et al 2011)- (Figure 9). We have confirmed cluster membership for ~23 galaxies at z~5.7, performing Keck/DEIMOS spectroscopy. Using the combined photometric and spectroscopic data, we study the effect of the environment on formation of galaxies when the Universe was ~1 Gyr old. Furthermore, for the first time, we extend the study of the evolution of the density-morphology (spectral type) and color-magnitude relations to z=5.7.



(II.7) Supernovae Search at z > 1.5

Fig10. Shows a high-z supernova at z =1.5. The left panel is the F160W filter image taken at the discovery epoch. The right panel shows the supernova image with the host galaxy image subtracted (Rodney et al 2012).


As a part of the CANDELS project, we have undertaken a comprehensive search for supernovae Type Ia and core collapse at z > 1.5, using the infrared camera (WFC3) on the HST (a program led by Adam Riess and Steve Rodney at Johns Hopkins University and STScI). This extends our previous search in optical wavelengths with HST/ACS (Strolger et al 2004) to z > 1.5 by performing the search at infrared (1.2-1.6 micron) wavelengths. I have been heavily involved in measuring photometric redshifts and other observable parameters (extinction, spectral type, stellar mass) for the supernovae host galaxies detected in this study. These data are used to measure the Supernovae rate and its change with redshift (Dahlen et al 2012) and to constrain evolution of dark energy with look-back time. Through this program, we discovered the highest redshift type Ia SNe at z=1.5 (Rodney et al 2012) and z=1.91 (Jones et al 2012)- (Figure 10).


(II.8). Supernovae Host Galaxies

We study properties of supernovae host galaxies at z > 1.5, discovered in our high-z supernovae search (section II.7), using both photometric and spectroscopic data. Study of the dependence of supernovae properties to their host galaxy allows us to have a better understanding of the systematic effects when using type Ia supernovae as standard candles. Using emission line intensities and SED fits to the photometry, we estimate the star formation rate, stellar mass, metallicity and spectroscopic emission line strengths diagnostic of physical properties of the host galaxy, allowing us to identify the type of the host galaxies and whether they contain active nuclei (Frederiksen et al 2012).


(II.9) Lyman alpha Escape Fraction

Fig10. Left: Estimated Ly alpha escape fraction is plotted vs. extinction. This shows that while extinction inhibits the escape of Ly alpha photons, there are other factors that govern Ly alpha escape, such as HI covering fraction and gas kinematics. Right: The Ly alpha escape fraction is plotted vs. the stellar mass. There is a slight trend between the stellar mass and the escape fraction with higher mass systems having lower escape fractions (Form Mallery, Mobasher et al 2012)


Using Keck/DEIMOS spectroscopic data in the COSMOS field, we measured the equivalent widths (EWs) and escape fractions of Lyman alpha emission lines for broad-band and narrow-band selected LBGs and Lyman Alpha Emitters (LAEs) at redshifts z=4.2, 4.8 and 5.7 (Mallery, Mobasher et al 2012). We found no evolution of the Lyman alpha Equivalent Widths with redshift for both the LBGs and LAEs. We also find that the Lyman alpha escape fractions for the LAEs is, on average, higher and has a larger variation compared to the LBGs, with sources with larger extinction showing smaller escape fractions (Figure 10). This indicates dust extinction to be the most important factor affecting the escape of Lyman alpha photons while, at low extinctions, other factors such as neutral Hydrogen covering factor and gas kinematics can be as effective at inhibiting the escape of Lyman alpha photons. We find a slight dependence of the escape fraction on the stellar mass of galaxies.




(II.10). Evolution of the SFR-Mass relation with Redshift

For a mass selected sample of galaxies at 0.2 < zphot < 1.2 in the COSMOS field, we measured the star formation rate from their rest-frame 2800 A flux (corrected for extinction) and studied the relation between the SFR and stellar mass in galaxies and its evolution with redshift (Figure 11)- (Mobasher et al 2009). We find the mean SFR to be a strong function of the stellar mass at any given redshift, with massive systems (log(M/Msun) > 10.5) contributing less (by a factor of ~5) to the total star formation rate density. We also find that massive systems have had their major star formation activity at earlier epochs (z > 2) than the low mass galaxies. We are continuing with this program using data in CANDELS fields with accurate estimates for the star formation rates and stellar mass for galaxies. 


Fig11. The diagram shows changes in the star formation density (SFRD) with redshift in different stellar mass intervals to z ~ 2. There is a clear trend between the SFRD and redshift for galaxies of all masses. The relation flattens at higher redshifts (from Mobasher et al 2009).

(II.11) Comparison Between Different Star Formation diagnostics

Why do the Star Formation Rates (SFRs) measured for the same galaxy, using different star formation diagnostics, differ? And what is the cause of this difference? The multi-waveband photometric and spectroscopic data for galaxies in different fields (CANDELS, COSMOS, GOODS) allow measurement of their SFRs using different diagnostics. The available star formation indicators include: rest-frame UV flux, far-infrared (8-1000 micron), radio (1.4 GHz), Spitzer IRAC (8 micron) and MIPS (24 micron) flux and spectroscopic emission lines ([OII], Ha). For an unbiased sample of mass-selected galaxies, we measure the SFR for individual galaxies, using these diagnostics (with Naveen Reddy). We then compare the SFRs measured independently and study the difference between these measurements and the cause of the deviation.


(II.12). MOSfire Deep Evolution Field (MOSDEF)

A new multi-object infrared spectrograph (MOSFIRE) has been commissioned on the Keck Telescope, allowing measurement of rest-frame optical spectra of faint and distant galaxies.  I am a co-PI on a large program aiming to perform spectroscopy of two thousand galaxies in the range 1.4 < z < 3.8 (MOSDEF) to measure their star formation histories, dust extinction, metallicity, ISM physical conditions and stellar and gas dynamics. We have been awarded 45 nights on the Keck Telescope to perform this study. My collaborators on this project are: A. Shapley (UCLA), A. Coil (UCSD), M. Kriek (UCB), B. Siana (UCR) and N. Reddy (UCR).