Abstract: Calculation of surface layer turbulent fluxes is very important for atmosphere numerical models.How to parameterizc the turbulent fluxes is one of the key research questions in current atmosphere boundary layer study.The paper uses two common schemes (MM5 similarity non-iterative Scheme A and ETA similarity iterative Scheme B) in the Weather Research Forecast Model (WRF) to make offline test and intercomparison of the parameterization results with the PBI.eddy-correlation obscrvation.The aerodynamic roughness length (z0)obtained on the Mali boundary layer is determined before calculating the turbulent fluxes.The aerodynamic roughness length (z0)by fitting four different heights wind speed from the Ptil.tower data under neutral condition varies significantly with season due to the obvious changes in underlaying surface conditions during the whole year (horsebcan in winter half year and paddy in summer half year).The results show that the vegetation sparseness degrec has a significant effect on turbulent fluxes calculation in the Mali cropland.In the unstable condition,Schcmc B undcrcstimates the momentum flux, and Scheme A is superior to Scheme B.Conversely, in the stable condition,Scheme A generates lower values with Scheme B is superior to Scheme A.On the whole,both of them produce little error.Scheme B takes account of the difference between aerodynamic roughness length (z0)and heat roughness length (z0h).The sensible heat flux result of Scheme B is much better than that of Scheme A,especially for the bare soil or sparse vegetation underlaying in the unstable condition.In case of bare soil the calculated sensible heat flux by Scheme B is still larger than the observation.The result is much better, if the (Geng, et al,1998)'s method using the radiometric surface temperature is adopted.