Abstract: High accuracy surface modelling(HASM)constructed based on the fundamental theorem of surfaces is more accurate than the classical methods.But HASM must solve a big sparse linear systems.Gauss-Seidel(GS)can be considered as the first method for solving the linear systems.In order to decrease the computation costs and improve the accuracy of HASM,we employed a modified Gauss-Seidel(HASM-MGS)to solve the linear systems of HASM.Gauss synthetic surface was selected as the research object.We proved that HASM-MGS is more accurate than HASM-GS and the classical methods used in Matlab.The computation time of HASM-MGS is approximately proportional to the one power of the total number of grid cells,which can be considered as a big improvement in solving HASM systems.