Integrated precipitable water vapor (PW) estimated from ground GPS has for many years been assimilated into operational weather models, and under a NASA AIST project, our collaboration of JPL, SIO, NOAA Weather Forecasting Offices in southern California, and NOAA's Earth System Research Laboratory demonstrated that GPS PW estimates enhance forecaster situational awareness during North American Monsoon events. However, during a rigorous investigation of operational near real-time processing, we discovered some interesting discrepancies between ESRL GAMIT 30-minute and JPL GIPSY 5-minute zenith delay and PW solutions. Of particular concern is the possibility that processing artifacts could lend themselves to misinterpretation in a subjective forecasting setting. Exploring this more deeply, we observed that PW timeseries determined with various GPS software packages and mapping functions have variations that manifest themselves on seasonal timescales. We present a study of differences in Zenith Total Delay (ZTD) estimated using different mapping functions and a priori troposphere information in an otherwise uniform processing scheme for stations located in various climate settings.