Fitting a Galactic model to an all-sky survey

We use star counts from the APS Catalog of the POSS I to develop a Galactic model optimized for large, statistically significant data sets uniformly distributed over the sky. The power of the APS Catalog for Galactic structure studies comes from its large sky coverage, individual photometric calibrations, adequate scanning resolution, statistically significant sample sizes, color information, and, most importantly, multidirectional sampling. The APS Catalog is an extremely useful exploratory data set but requires new methodologies to maximize its usefulness. We have selected an 88-field subset, 16 deg² each, from the catalog for a program of magnitude-limited star counts (12 ≤ O [blue] ≤ 20) within the completeness limit of the survey and in a realm where star-galaxy classification has minimal effects on the results. We have developed a simple three-component (disk, halo, thick disk) model optimized for efficiently and objectively analyzing star-count information. Our model not only produces model counts for our multidirectional data, but also returns a " goodness of fit " statistic. We use a genetic algorithm, a robust optimization technique well suited for this large multidirectional and multiple-parameter study, to optimize the fit and derive a self-consistent set of global parameters to model the Galaxy. With this global fit, we can identify significant deviations from symmetry in the Galaxy's large-scale distribution of stars. The results from 12 independent executions or trials yielded consistent results. All of the model fits produced a flattened inner halo with c/a ≈ 0.5-0.6. The radial scale length of the disk, ≈ 3.5 kpc, is higher than found in recent infrared surveys but agrees with older optical studies. The density normalizations to the plane for the thick disk and halo are consistent with previous work. Our model results all yielded a scale height for the thick disk of ≈900 pc. The most surprising results from these global fits are relatively high values (>4 kpc) for the de Vaucouleurs radius for the halo and the radial scale length of the thick disk. The radial scale length for the thick disk is significantly larger than that for the old/thin disk and, if confirmed with additional work, may imply an independent origin for the thick disk. We also present evidence that the scale height and normalization of the thick disk may be variable with direction.