The precipitation process is central to the hydrological and energy cycle in the atmosphere. The spatial and temporal variability associated with rainfall not only has a tremendous influence on the agriculture, drought and flood but also has a controlling effect on large-scale circulation of atmosphere and ocean. In spite of its scientific and socioeconomic importance, our quantitative knowledge of precipitation variability is still unsatisfactory. Different techniques in estimating precipitation rate shown significant differences in various areas. On the other hand, the precipitation is also one of the most difficult process to model and predict. Cloud microphysics and the patteren of temperature, humidity and wind control the intensity, scales and timing of rainfall. The simplistic parameterization of precipitation process in the coarse grid climate model is likely the reason for the poor performance of precipitation simulations in the present climate models. Efforts to improve these models are hindered by the lack of reliable data. There are two main objectives of the present study. One is the intercomparison between various observed rainfall estimates. It provides the characteistics of the precipitation analyses from individual technique and the range of uncertainty for the present rainfall estimates based on single source or combined data. Here the focus will be the precipitation variability in different datasets over a few specific climate regimes. The other goal is to examine and validate the model simulation of regional precipitation variability using the aboved mentioned observational datasets. Different model runs are used to illustrate the sensitivity of model results to different horizontal resulutions and physical parameterizations.