Geodatabase updating by new measurements; a Bayesian intermezzo

Abstract

The question the paper deals with refers to how it is possible to update existing geodatabases considering both their accuracies and those of the new measurements taken for their updating. Traditionally, maintaining geodatabases (or map bases) has been highly time consuming, costly, and sometimes difficult work, especially in urban and high-density areas. The most common procedure is to globally generate geodatabases every few years by photogrammetric techniques. On the opposite, the possibility of dynamically updating the landscape information from a maintained core spatial database can be considered as an appealing alternative to traditional map revision techniques. A kriging solution, based on the hypothesis that the vector field of the position error on a geodatabase is a homogeneous, isotropic intrinsic random field with constant mean and variogram depending only on the squared distance, known a priori from the relative accuracy of the map, is proposed. The method is a first approach to the problem, as far as at the moment it does not consider constraints to which points on the geodatabase must adapt to. That is the reason why it is presented as an intermezzo. © Società Italiana di Fotogrammetria e Topografia (SIFET) 2009.