Our ability to project future climate change, especially trends in costly precipitation extremes, hinges upon whether coupled GCMs capture processes that affect precipitation characteristics. A comparison of such characteristics was made between the NCAR CAM3.1 simulation and observations derived from the NASA Tropical Rainfall Measuring Mission (TRMM) satellite. It reveals that despite a fairly good simulation of the annual mean rain rate, CAM rains about 10-50% more often than the real world and fails to capture heavy rainfall associated with deep convective systems. Comparison of regionally aggregated probability density functions (PDFs) of the rain rate shows that the model compensates for the lack of heavy precipitation through raining more frequently within the light rain category, which leads to an annual rainfall amount close to what is observed. Through the method of Bayesian Stochastic Inversion (BSI), model uncertainties associated with key parameters in the cumulus convection schemes are quantified in terms of their posterior probability distributions. A set of optimal values of those parameters are identified based on their ability to produce seasonal mean atmospheric states that are consistent to observations. Influences of these parameter values on the model simulated precipitation characteristics are examined for both the present-day climate and a projected warmer world in the coming decades.