HI DEEP FIELD AND GALAXY GROWTH
While the simulations predict that gas accretion has evolved significantly for larger galaxies changing from cold to hot mode accretion over the last 4.6 Gyr, observations show that the star formation rate has dropped by a factor of 3 over that range (Hopkins & Beacom 2006), with the larger galaxies forming most of their stars earlier and the smaller galaxies still forming stars today. Observationally, the low mass galaxies appear to collect large reservoirs of gas and have not formed stars efficiently over their lifetime as compared to more efficient massive galaxies, or alternatively they have acquired their reservoirs late, because that is when cold mode accretion dominates. CHILES will image in HI hundreds of galaxies over this redshift range and connect HI morphologies with the star formation histories. We note here that at z=0 galaxies with stellar masses less than 10^9 M_sun have more mass in HI than stars, while in contrast galaxies like the Milky Way have only a small fraction of their mass in HI. Consequently HI content is relatively insensitive to stellar mass, suggesting that CHILES will probe galaxies with a wide range of stellar mass that cover both regimes over most of the redshift range. It is precisely the alternative scenarios presented above explaining the large gas reservoirs around small galaxies, that the deep field will be able to test, because we can, for the first time, trace the gas reservoir with look-back time and halo mass over a substantial look-back time. The survey will contribute in many different ways:
The HI content, morphology and kinematics of individual galaxies:
Here CHILES is likely to make its most important contribution. It will provide HI gas masses, morphology and kinematics for a wide range of stellar masses and environments. Cold mode accretion models make very specific predictions how cold mode accretion depends on redshift, galaxy mass and environment (i.e. Keres et al. 2005). For example, a Milky Way size galaxy will switch from being cold mode accretion dominated to hot mode accretion dominated at a redshift of about 0.5. For smaller galaxies this redshift is lower and we may be able to trace that transition as function of redshift and mass. The prediction is that at z=0 small galaxies in voids should still show evidence for cold mode accretion. High resolution HI imaging of nearby void galaxies shows kinematics consistent with infall in some galaxies (Kreckel et al. 2012; Stanonik et al. 2009). The best example of a galaxy in the local universe that is still assembling is IC10, and the evidence was found from HI (Wilcots & Miller 1998). The exquisite resolution of the VLA-B array (spatially and in velocity) is necessary to do detailed kinematic studies of the larger galaxies and compare those with models of infall. We will make an inventory of kinematically disturbed HI as function of stellar mass, environment and redshift, and use the HST images to distinguish between galaxies with and without stellar disturbance. Thus we will be able to assess importance of mergers (disturbed optical and gaseous morphology) versus gaseous accretion (disturbed gas kinematics). At the highest redshift, CHILES will only be sensitive to the larger HI masses. Here we can test how the cold gas reservoir of the high stellar mass galaxies evolves with redshift.
The HI mass function:
The HI mass function describes the space density of galaxies as function of their HI mass. Its evolution with redshift will eventually be useful to constrain models of hierarchical galaxy formation. Although in the local universe this function is well defined on a global scale (Martin et al. 2010), even locally not much is known about its environmental dependence. Interestingly, although many measurements of the HI mass function have been made in the past, it was the larger volume probed by ALFALFA that showed that there was a significant population at the high end of the HI mass function that was missed in the previous surveys (Martin et al. 2010). It is expected that if there is any evolution at all with redshift, it is the high mass end that will change. Our survey is well designed to probe a possible change at the highest redshifts. This can then be directly compared to predicted changes in accretion mode. Since the LSS structure in the COSMOS field is well defined over the entire redshift range we will be able to derive mean HI mass functions for different environments. We can do this by averaging all the data for specific LSS structures, such as voids, walls, groups etc.
The Cosmic Neutral Gas Density:
Observations of Omega_HI over redshift suggest that it has been relatively constant from z=3 to z=0.5 and then it appears to drop by a factor 2 to the locally measured and most value by Martin et al. (2010). Our survey will contribute several points between z=0.5 and 0 and help elucidate why it goes down. Is the universe running out of gas, or is more of the gas ionized?
Environmental suppression of star formation:
It has been well documented that disk galaxies in the centers of clusters lose (parts of) their gas and have reduced star formation, but, there is increasing evidence that this phenomenon already starts in the group environment (McGee et al. 2009, and references therein; Hess and Wilcots, 2013, in prep). We will probe a huge range of environments over the entire redshift range. The detailed morphology and kinematics of the HI will allow us to assess what the dominant cause of this so called pre-processing is.
Dark haloes and baryons:
The Tully-Fisher relation connects rotational velocity (total mass) to total luminosity (baryon mass). Although the original relation was established using HI profile widths, its evolution with redshift is actively studied using optical emission lines. CHILES will be the first survey that can make a consistent comparison, using the same probe, between the relation at z=0 and z=0.4.
ANCILLARY SCIENCE
Continuum Study:
This data set will yield the deepest L-band continuum image ever made. Preliminary reduction of our pilot study data produced a continuum image with 4.2 microJy/beam rms noise, within 50% of theoretical noise predictions. With a factor of 16 more integration time and four times the bandwidth, the full HI deep field will produce a continuum image with sub-microJy sensitivity. This will enable the unprecedented detection of ULIRGs with SFR=100 M_sun/yr and the brighter LBGs out to z = 3.3. The image will be a wonderful supplementary data set for the ongoing VLA Large program, "Deep 3GHz EVLA-COSMOS Survey" by V. Smolcic et al., which is aimed at studying dust-unbiased star formation activities at high angular resolution and searching for faint radio AGN to examine their importance in the evolution of host galaxies and their environment (e.g., "radio mode" feedback). With their sensitivity of 2 microJy they are expected to find 1000-3000 sources in our deep field area. Our sub-microJy sensitivity at L band will allow for detection of all the 3GHz sources and allows for much deeper spectral index analysis.
Transients:
These data will also enable the deepest survey for radio transients to date. The HI Deep Field data will be collected across ~200 epochs, each yielding a deep (5.5 microJy/beam) continuum image, accessing a brand new region of radio transient parameter space - a very interesting region, where we expect detections of tidal disruption events, radio supernovae, orphan gamma-ray bursts, and neutron star-neutron star mergers (Frail et al. 2012). The COSMOS field is often targeted by transient surveys at other wavelengths - for example, the Pan-STARRS1 medium deep survey has detected 12 supernovae in the HI Deep Field over the past three years. Such complementary multi-wavelength data ensure that radio transients will be understood, not merely detected.
HI absorption:
is a powerful probe of cold neutral gas, with a distance independent sensitivity, but depending on the strength of the continuum sources. Our field, which was selected to have no strong continuum sources, has one 10 mJy source and about 50 1 mJy sources. The column density sensitivity toward the 1 mJy sources is 3 x 10^19 cm^-2 at 5 sigma, comparable to our sensitivity to HI emission at low z. Probing of these 50 lines of sight for HI absorption offers complementary information about cold gas distribution in and around galaxies and the associated large scale structure within the same redshift range probed by the 300 HI emitters we expect to detect.