Near-infrared galaxy counts to J and K∼24 as a function of image size

We have used the Keck 10 m telescope to count objects as a function of image size in two high Galactic latitude fields covering 1.5 arcmin(2) and reaching 50% completeness depths of K = 24 and J = 24.5 for stellar sources. Our counts extend similar to 1 mag deeper in K than those of surveys with other telescopes; complement other Keck surveys in the K-band that provide counts at comparable or shallower depths but that have not utilized image structure; and extend by several magnitudes the J-band counts from brighter surveys using smaller telescopes that cover larger areas. We fmd the surface density of objects at K = 23 to be higher than previously found (similar to 500,000 mag(-1) deg(-2)), but at K < 22 to be consistent with other surveys. The slope of the K-band counts (d log A/dm = 0.36) is similar to others near this depth as well as to our own J-band counts (0.35). Counts in the J- and K-bands are both in excess of our empirical no-evolution models for an open universe, with the largest excess observed in J. The counts are a factor of 2 higher than mild-evolution models at J and K similar to 23. The slope of the model counts is insensitive to the assumed geometry even in the near-infrared primarily because the model counts are dominated by Iow-luminosity (<0.1L*) objects at modest redshift (z < 1) with small apparent sizes (half-light radii less than or equal to 0." 4, i.e., <4 h(50)(-1) kpc). The numbers of observed counts rise most steeply for these smaller objects, which dominate the counts fainter than K = 22.3 and J = 23.3. However, the greatest excess relative to no-evolution models occurs for the apparently larger objects, which have a median J-K of similar to 1.5. At these depths, the size and colors of such objects correspond equally well to luminous (greater than or equal to 0.1L*) blue galaxies at 1 < z < 4, or progressively more diffuse, blue, low-luminosity (0.001-0.1L*) galaxies at z < 1. The majority of these sources are too faint for spectroscopic measurement. Based on optical colors, we can rule out the possibility that the excess is caused by very low luminosity (<0.0001L*) red galaxies at z < 0.25. We also find a deficit of galaxies with red J-K colors corresponding to nonevolving, luminous, early-type (i.e., "red envelope") galaxies at 1 < z < 3. Even assuming that the deficit is caused by their appearance as blue galaxies, they could account for only 10%-30% of the excess of large, blue galaxies. The nature and redshift distribution of excess large and small galaxy populations at K = 24 and J = 24.5 remain indeterminate from these data alone.