Based on the existing land-surface schemes and models,an improved Land-surface ProcessModel(LPM-ZD)has been developed.It has the following major characteristics:(1)Thecombination of physical equations and empirical analytical formulae are used to construct thegoverning equations of soil temperature and moisture.Higher resolution of model level andphysical equations are adopted for the upper soil layers,and for the lower soil layers,lowerresolution of model level is adopted and empirical analytical formulae are used.(2)In land surfacehydrological process,the sub-grid distribution of rainfall and its effects are taken into account.(3)A simple snow cover submodel has been used,which includes effects of snow cover on soilthermodynamics and hydrology,as well as albedo.By use of this model and three groups of point observation data,a series of“off-line”testshave been carried out.The simulation results indicate that land-surface process model has goodperformance and can well simulate diurnal and seasonal variation of land surface processes for manykinds of land surface covers(forest,grass,crops and desert)in different climate zone.The resultssimulated by the model are consistent with the observations.Later,by use of one group ofobservation data and the model,a series of sensitivity experiments have been done.It is shownthat the model is much sensitive to some parameters,such as initial soil moisture,vegetationphysical parameters as well as the proportion of the grid covered with rain.Therefore it is muchimportant for land-surface process model to define these parameters as accurately as possible. Based on the existing land-surface schemes and models, an improved Land-surface Process Model (LPM-ZD) has been developed. It has the following major characteristics: (1) The combination of physical equations and empirical analytical formula of soil temperature and moisture.Higher resolution of model level andphysical equations are for the upper soil layers, and for the lower soil layers, lower resolution of model level is adopted and empirical analytical formulae are used. (2) In land surfacehydrological process, the sub-grid distribution of rainfall and its effects are taken into account. (3) A simple snow cover submodel has been used, which includes effects of snow cover on soil thermodynamics and hydrology, as well as albedo. By use of this model and three groups of point observation data, a series of “off-line” testshave been carried out. The simulation results indicate that land-surface process model has good performance and can well simulate diur nal and seasonal variation of land surface processes for many kinds of land surface covers (forest, grass, crops and desert) in different climate zone. The result is simulated by the model are consistent with the observations. Lot, by use of one group of data and the model, a series of sensitivity experiments have been done. It is much less sensitive to some parameters, such as initial soil moisture, vegetationphysical parameters as well as the proportion of the grid covered with rain. Wherefore it is muchimportant for land- surface process model to define these parameters as accurately as possible.