The impact of rectification error on the analysis of landscape transformation based on archival maps

Archival topographic maps are the main source of information about the land cover and land use (LULC) structure, particularly with reference to times before the application of aerial and satellite remote sensing. Maps created in different periods differ substantially in regard to cartographic technique, scale and generalisation level. This brings problems of using them as a data source in landscape transformation analysis. If these problems are not correctly solved in the initial stage of the research, the obtained land use/cover change (LULCC) results may be biased by errors leading to incorrect conclusions. For the interpretation of landscape transformation in the aspect of ecological processes, a simple comparison of proportions of particular LULC classes in certain periods is not sufficient, because a given transformation type in one place might be compensated by an opposite change in another place. Thus, in order to investigate the actual LULCC dynamics, and thereby to get to know its influence on vegetation, biodiversity and other landscape elements, it is necessary to use methods allowing for a detailed analysis of changes between LULC classes in the given period. One of the most straightforward approaches is a transformation matrix. In order to apply a transformation matrix to cartographic materials from different times, they need to be rectified to a common coordinate system. Because of deformations of the topographic map contents due to the map scale, map projection, cartographic and print technique, the possible distortions during storage and utilization, as well as during the digital scanning of the map, different for each map series or even between single sheets of the same map series, a perfect georectification of a scanned topographic map is virtually impossible. Therefore, the transformation matrix contains the information about the factual transitions between LULC categories, as well as about the artificial ones, due to map rectification error. Spatial error assessment procedure is then necessary to extract the information about the real transformation that took place. In this paper a method for reducing the impact of rectification error on the LULCC analysis is presented, based on authors landscape transformation research, conducted in the agricultural landscape of the Odra valley, using digitized maps from 1930's-1940's and 1990's.