The authors explore the resolving power of an inversion algorithm which estimates five parameters of the seafloor covariance function from a single swath of multibeam echosounding data. The resolving power is evaluated as a function of the swath length, the orientation of ship track with respect to topographic grain, and the response width of the sounding system. The analysis is conducted by inverting sets of synthetic data with known statistics. The mean and standard deviation of the inverted parameters can be directly compared with the input parameters and the standard errors output from the inversion. Experiments show that resolution of the covariance parameters is strongly dependent on the number of characteristic lengths which are sampled. Root-mean-square seafloor height can be estimated to within approximately 15%, and anisotropic orientation to within approximately 5% (for a strong lineation), using track lengths as short as three characteristic lengths. >