In order to assesses accurately the characteristics of the land surface from different angles, and provides guidance to improve the simulation performance of land surface models, the framework of LSMS (Land Surface Modeling System) is constructed on the basis of scripting languages and Model-Data Fusion methods to provide researchers a land surface modeling environment from data processing to simulation analysis. The modeling system is an integration of observing data, land surface simulation, high-performance computing, data processing and analysis, and visualization techniques. Different land surface models were used in the framework to prove the utilization potentials of modeling system based on scripting language in high-performance computing environment and the role of various visualization schemes in the modeling system.