Premium
A comparison of remotely sensed and pollen‐based approaches to mapping Europe's land cover
Author(s) -
Woodbridge Jessie,
Fyfe Ralph M.,
Roberts Neil
Publication year - 2014
Publication title -
journal of biogeography
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.7
H-Index - 158
eISSN - 1365-2699
pISSN - 0305-0270
DOI - 10.1111/jbi.12353
Subject(s) - pollen , land cover , vegetation (pathology) , scale (ratio) , physical geography , geography , remote sensing , cover (algebra) , vegetation cover , land use , cartography , ecology , biology , medicine , mechanical engineering , pathology , engineering
Abstract Aim Remote sensing coupled with direct observation allows recent changes in vegetation to be investigated but, in order to extend our understanding of land‐cover change further back in time, different proxies for vegetation are required. The pseudobiomization ( PBM ) approach has been developed to transform fossil pollen data into land‐cover classes ( LCC s) in order to reconstruct broad‐scale anthropogenic land‐use change through time. The aim of this study was to test and refine the PBM approach through application to an extensive modern pollen dataset and comparison with remotely sensed CORINE land‐cover maps for Europe. Location The study area comprised 2471 modern pollen sites from across Europe. Methods Pollen sites were assigned to one of eight LCC s using the pollen‐based PBM method, which draws upon biomization techniques to transform pollen data into records of land‐cover change. Five of the LCC s were ‘pure’ classes (e.g. broad‐leaf forest) and three of them were mixed vegetation. Remotely sensed CORINE land‐cover maps were used to assign LCC s to sites, and the results were compared with pollen‐assigned LCC s. Results The results revealed a good correspondence between the proportions of different LCC s that were registered using CORINE and pollen‐based PBM , when data were aggregated at a pan‐European scale. However, the match between the two datasets was much less close at a site‐specific level. The overall results were improved to c. 60% when the target was broadened to include similar as well as identical LCC s. The spatial correspondence was best across north‐central Europe and least good in south‐west Europe, the Mediterranean and northern Scandinavia. Main conclusions The ability of distinct data types to sense actual vegetation is limited by various sources of error; for example, both pollen and remote sensing vary in terms of spatial and temporal heterogeneity and taxonomic resolution. It is likely that many of the main sources of error are common to both methods, rather than being approach‐specific. We conclude that pollen‐based methods of intermediate complexity can be used as a proxy for broad‐scale land‐cover change across most of temperate Europe, but may be less reliable at a site‐specific scale.