Admissible relative errors of the parcel area while creating a land parcel map base using an orthoimage as a reference

Abstract

Currently the official land parcel map of Uruguay does not allow a suitable interoperability with other available geographic information. We are considering using a national high accuracy and high resolution orthomosaic in two ways: either to support the creation of a brand new land parcel map or to allow amending the existing one. In particular, the present work is part of a set of experiments which considers different techniques to amend the existing land parcel map. In order to perform a fair comparison it is necessary to have test cases produced with a reference procedure properly and objectively evaluated. The metric of success will be used as attainable goal for any other alternative. Using the orthomosaic in order to identify the apparent limits of parcels in the land parcel map we were able to modify its location, thus making the map interoperable with the image. In order to measure the accuracy of the fit, we used the relative difference between the area declared at the parcel map (taken as the reference value) respect to the one computed from the GIS. We found that the discrepancy between the areas computed from the initial and the final dataset w.r.t. the declared area decreases, but not up to expected level. Once filtered for outliers the remaining discrepancies follow a normal distribution, where its bias and standard deviation varies from (-0.65%, 11.24%) before editing to (‑1.72%, 8.59%) later. Despite the drastic change in planimetric accuracy the improvement of the area accuracy is marginal. Considering the admissibility criteria of the Uruguayan National Cadastral Office, for the legacy dataset only 39.4% parcel maps could be accepted, a value which improves up to 49.7% after the manual edition. It is still far from the goal of 100%, and the reasons for such pessimistic finding will be discussed.