The mean instantaneous runoff rate over a mesoscale region is considered to be the residue between the areal mean precipitation and the amount of water infiltrated into surface soil layer.A rainfall probability density function(PDF) derived from rainfall data in 1996 is used,in conjunction with the mathematic description and empirical expression of rain water infiltration physics,in order to accurately estimate the soil infiltration rate and distribution.A statistical-dynamic scheme of regional mean surface runoff is constructed.The runoff rate can be viewed as the difference between mean precipitation and infiltration on a regional basis,and the averaged infiltration can be treated independently over saturated and unsaturated areas.For the physics of land surface water cycle,infiltration is caused by a source of water supply associated with the properties of underlying surface.After rainfall,with part of the water transported into the soil surface layer,runoff occurs just due to the surplus from the water trapped by vegetation cover and soil infiltration.Hence,the key in calculating surface runoff lies dominantly in the estimation of water amount required for surface-layer soil.Thus,the expression for soil water flux is utilized to derive a formula for the infiltration.Similarly,from the PDF for soil moisture and precipitation,a formula of sub-grid regional mean runoff rate is obtained by considering heterogeneous soil water content and climatic rainfall forcing in 1996 as well as precipitation and soil data over the Yangtze delta region.Sensitivity experiments are also done to indentify affecting factors.Evidence suggests that the proposed scheme gives the runoff rates highly close to those from the Mosaic method,thereby demonstrating the high reliability and feasibility of the statistical-dynamic parameterization scheme. The mean instantaneous runoff rate over a mesoscale region is considered to be the residue between the areal mean precipitation and the amount of water infiltrated into the surface soil layer. A rainfall probability density function (PDF) derived from rainfall data in 1996 is used, in conjunction with the mathematic description and empirical expression of rain water infiltration physics, in order to precisely estimate the soil infiltration rate and distribution. A statistical-dynamic scheme of regional mean surface runoff is constructed. run rate rate be be viewed as the difference between mean precipitation and infiltration on a regional basis, and the averaged infiltration can be treated independently over saturated and unsaturated areas. For the physics of land surface water cycle, infiltration is caused by a source of water supply associated with the properties of the underlying surface. After rainfall, with part of the water transported into the soil surface layer, runoff occurs just due to the s urplus from the water trapped by vegetation cover and soil infiltration .ence, the key in calculating surface runoff lies dominantly in the estimation of water amount required for surface-layer soil.Thus, the expression for soil water flux is utilized to derive a formula for the infiltration.Similarly, from the PDF for soil moisture and precipitation, a formula of sub-grid regional mean runoff rate is obtained by considering heterogeneous soil water content and climatic rainfall forcing in 1996 as well as precipitation and soil data over the Yangtze delta region .Sensitivity experiments are also done to indentify affecting factors. Evidence suggests that the proposed scheme gives the runoff rates highly close to those from the Mosaic method, thereby demonstrating the high reliability and feasibility of the statistical-dynamic parameterization scheme.