Clustering at high redshift: Precise constraints from a deep, wide-area survey

We present constraints on the evolution of large-scale structure from a cat alog of 710,000 galaxies with I-AB less than or equal to 24 derived from a KPNO 4 m CCD imaging survey of a contiguous 4 degrees x 4 degrees region. T he advantage of using large contiguous surveys for measuring clustering pro perties on even modest angular scales is substantial: the effects of cosmic scatter are strongly suppressed. We provide highly accurate measurements o f the two-point angular correlation function, omega(theta), as a function o f magnitude on scales up to 1.degrees 5. The amplitude of omega(theta) decl ines by a factor of similar to 10 over the range 16 less than or equal to I less than or equal to 20 but only by a factor of 2-3 over the range 20 < I less than or equal to 23. For a redshift dependence of the spatial correla tion function, xi(r), parameterized as xi(r, z) = (r/r(0))(-gamma)(1 + z)(- (3+epsilon)), we find r(0) = 5.2 +/- 0.4 h(-1) Mpc, and epsilon greater tha n or similar to 0 for I less than or equal to 20. This is in good agreement with the results from local redshift surveys. At I > 20, our best-fit valu es shift toward lower r(0) and more negative epsilon. A strong covariance b etween r(0) and epsilon prevents us from rejecting epsilon > 0 even at fain t magnitudes, but if epsilon > 1, we strongly reject r(0) less than or simi lar to 4 h(-1) Mpc (comoving). The above expression for xi(r, z) and our da ta give a correlation length of r(0)(z = 0.5) approximate to 3.0 +/- 0.4 h( -1) Mpc, about a factor of 2 larger than the correlation length at z = 0.5 derived from the Canada-France Redshift Survey (CFRS). The small volume sam pled by the CFRS and other deep redshift probes, however, makes these spati al surveys strongly susceptible to cosmic scatter and will tend to bias the ir derived correlation lengths toward the low end. Our results are consiste nt with redshift distributions in which similar to 30%-50% of the galaxies at I = 23 lie at z > 1. The best-fit power-law slope of the correlation fun ction remains independent of I magnitude for I less than or equal to 22. At fainter limits, there is a suggestive trend toward flatter slopes that occ urs at fluxes consistent with similar trends seen by Neuschaffer & Windhors t and Campos and coworkers. The galaxy counts span 11 magnitudes and provid e an accurate calibration of the galaxy surface density. We find evidence f or mild galaxy evolution: about 1 mag of brightening or a doubling of the d ensity by I = 23 relative to an Omega(0) = 1 no-evolution model, about 0.5 mag of brightening or a factor of 1.5 increase in surface density relative to an open model. Our galaxy counts agree well with those from the Hubble D eep Field survey and, thus, argue against a significant inclusion of subgal actic components in the latter census for I < 24.