Considering the relationship between the measures of terrain correlation algorithms and the 2-D self-correlation coefficient of terrain, this paper proposes an improved Johnson method-to analyze the statistical properties of measures of matching algorithms on correlated terrain. First, based on the issue of Johnson method, we modify the independence assumption of consecutive elements in reference data and develop distribution functions of measures of MSD. Then, the suitability of Johnson method and our method are discussed, and the shortcomings of the former are pointed out. Finally, an application of our method to terrain correlation suitabitity, to estimate the false-fix probability in correlated non-matching positions is presented.