To support future decisions on alternative strategies of updating soil survey information, we have studied the efficiency of four strategies, viz. revision, upgrading, revision plus upgrading and upgrading by two-phase sampling. Revision results in a new soil map. Upgrading gives statistical information about means and variances within mapping units of the soil map. The merits of these strategies were measured in terms of the increase in accuracies of the spatial estimates of the values at unvisited points and of the spatial means of a study area of several soil characteristics. Point values were estimated by assigning an estimate of the mean value to any point in a given mapping unit. In the first strategy, the estimator was derived from the representative profile description of that mapping unit. In the other strategies, the estimator was derived from the statistical sample. Although the mapping units of the revised map were more homogeneous for some characteristics, the point estimates using the values of the representative profile description were no more accurate. This was due to the bias of this estimator, which rules out the reduction in spatial variance. If revision was followed by sampling this bias could be eliminated. As a result, for some characteristics the point estimates became more accurate than those based on the original map. Estimates of the spatial means of the study area via upgrading by two-phase sampling were more accurate, for all characteristics, than estimates via revision plus upgrading. Using the estimates of the spatial variance within the mapping units of the original map for allocation was apparently more effective than reducing the variance itself.
