In this paper we perform a comprehensive study of the main sources of random and systematic errors in stellar mass measurement for galaxies using their Spectral Energy Distributions (SEDs). We use mock galaxy catalogs with simulated multi-waveband ph
otometry (from U-band to mid-infrared) and known redshift, stellar mass, age and extinction for individual galaxies. Given different parameters affecting stellar mass measurement (photometric S/N ratios, SED fitting errors, systematic effects, the inherent degeneracies and correlated errors), we formulated different simulated galaxy catalogs to quantify these effects individually. We studied the sensitivity of stellar mass estimates to the codes/methods used, population synthesis models, star formation histories, nebular emission line contributions, photometric uncertainties, extinction and age. For each simulated galaxy, the difference between the input stellar masses and those estimated using different simulation catalogs, $Deltalog(M)$, was calculated and used to identify the most fundamental parameters affecting stellar masses. We measured different components of the error budget, with the results listed as follows: (1). no significant bias was found among different codes/methods, with all having comparable scatter; (2). A source of error is found to be due to photometric uncertainties and low resolution in age and extinction grids; (3). The median of stellar masses among different methods provides a stable measure of the mass associated with any given galaxy; (4). The deviations in stellar mass strongly correlate with those in age, with a weaker correlation with extinction; (5). the scatter in the stellar masses due to free parameters are quantified, with the sensitivity of the stellar mass to both the population synthesis codes and inclusion of nebular emission lines studied.
For a mass-selected sample of 66544 galaxies with photometric redshifts from the Cosmic Evolution Survey (COSMOS), we examine the evolution of star formation activity as a function of stellar mass in galaxies. We estimate the cosmic star formation ra
tes (SFR) over the range 0.2 < z < 1.2, using the rest-frame 2800 A flux (corrected for extinction). We find the mean SFR to be a strong function of the galactic stellar mass at any given redshift, with massive systems (log (M/M(Sun)) > 10.5) contributing less (by a factor of ~ 5) to the total star formation rate density (SFRD). Combining data from the COSMOS and Gemini Deep Deep Survey (GDDS), we extend the SFRD-z relation as a function of stellar mass to z~2. For massive galaxies, we find a steep increase in the SFRD-z relation to z~2; for the less massive systems, the SFRD which also increases from z=0 to 1, levels off at z~1. This implies that the massive systems have had their major star formation activity at earlier epochs (z > 2) than the lower mass galaxies. We study changes in the SFRDs as a function of both redshift and stellar mass for galaxies of different spectral types. We find that the slope of the SFRD-z relation for different spectral type of galaxies is a strong function of their stellar mass. For low and intermediate mass systems, the main contribution to the cosmic SFRD comes from the star-forming galaxies while, for more massive systems, the evolved galaxies are the most dominant population.
