A characteristic eddy decomposition is applied to extract the coherent structure of random turbulent flow field. The characteristics of Karhunen-Loeve (K-L) expansion and Fourier expansion are compared on the convegence of their expansions in representing inhomogeneous instantaneous turbulent flows. The model turbulence is generated by solving the Burgers` equation with random forcing. The coefficients of the Fourier expansion are determined by a Galerkin approach. When the Burgers` turbulent flow field is represented by the K-L expansion and the Fourier expansion, the RMS error increases with an increase of Reynolds number. The RMS error of the K-L expansion is always smaller than that of the Fourier expansion by a Galerkin approach. The results show the superiority of the K-L expansion, especially, for high Reynolds number flows.