Differences in land cover interpretation in landscapes rich in cover gradients: reflections based on the montado of South Portugal

TitleDifferences in land cover interpretation in landscapes rich in cover gradients: reflections based on the montado of South Portugal
Publication TypeJournal Article
Year of Publication2007
AuthorsDoorn, A. M., & Correia T. Pinto
JournalAgroforestry Systems
Volume70
Pagination169-183
ISBN Number2200100035
Keywordsland cover classification, montado/dehesa, thematic cartography, visual aerial photo interpretation
Abstract

This paper contributes to the discussion on current issues in methodologies of mapping land cover in the agro-silvo-pastoral landscapes of the Mediterranean. These landscapes, characterized by intermixed land use and indeÞnite boundaries, require particular attention in applying the patch-corridor-matrix model when classifying patches and their delineation. In a case study area in southeast Portugal, mainly characterized by agro-silvo pastoral systems, the land cover for 1990 has been mapped. The paper discusses the consequences of the complexity of some Mediterranean land use systems for land cover mapping dealing with detailed landscape dynamics. Within this scope a land cover mapping project in a small case study area is compared with the mapping undertaken within a national land cover database. Both studies were carried out on the same scale and through visual interpretation of aerial photographs. Differences in land cover classiÞcation and allocation are explored using matrix with levels of agreement. Recommendations for future land cover mapping projects are: the application of fuzzy approaches to land cover mapping in agro-silvo-pastoral landscapes should be explored and land cover classiÞcations should be standardized in order to enhance consistency between databases. On the other hand, the fuzziness of the boundaries in this kind of landscapes is inherent to the system and should be accepted as such. The accompanying uncertainties should be taken into account when undertaking landscape analysis on the basis of land cover data.