Abstract. Water vapor tomography is a promising technique for reconstructing the 4-D moisture field, which is important to the weather forecasting and nowcasting as well as to the numerical weather prediction. A near real-time 4-D water vapor tomographic system is developed in this study. GPS slant water vapor (SWV) observations are derived by a sliding time window strategy using double-difference model and predicted orbits. Besides GPS SWV, surface water vapor measurements are also assimilated as real time observations into the tomographic system in order to improve the distribution of observations in the lowest layers of tomographic grid. A 1-year term experiment in Hong Kong was carried out. The feasibility of the GPS data processing strategy is demonstrated by the good agreement between the time series of GPS-derived Precipitable Water Vapor (PWV) and radio-sounding-derived PWV with a bias of 0.04 mm and a root-mean-square error (RMSE) of 1.75 mm. Using surface humidity observations in the tomographic system, the bias and RMSE between tomography and radiosonde data are decreased by half in the ground level, but such improved effects weaken gradually with the rise of altitude until becoming adverse above 4000 m. The overall bias is decreased from 0.17 to 0.13 g m⁻³ and RMSE is reduced from 1.43 to 1.28 g m⁻³. By taking the correlation coefficient and RMSE between tomography and radiosonde individual profile as the statistical measures, quality of individual profile is also improved as the success rate of tomographic solution is increased from 44.44 to 63.82% while the failure rate is reduced from 55.56 to 36.18%.