PARAMETRIC AND NONPARAMETRIC PROCEDURES FOR TESTING THE METRIC QUALITY OF A DIGITAL MAP

This paper proposes an original procedure for testing the metric quality of point coordinates contained in the data files of a digital map. The procedure starts by verifying the normal distribution of residuals of tie and control point coordinates for groups of models or for the entire (photogrammetric) block adjustment. If the normal distribution of the residuals is satisfied, a parametric test is applied to the coordinate differences of the checking points relating to two alternative survey methods. On the other hand, if the normal distribution of residuals is not accepted, simple non-parametric test statistics are proposed. Among these the Wilcoxon signed rank test and the sign test seem to be applicable. If the null hypothesis for the homogeneous distribution of the point coordinates obtained by two different survey methods is rejected, a ''shift model'' is proposed. Some numerical results highlight the capabilities and the limits of the procedure proposed.