A 10-year survey of trace metals in sediments using self-organizing maps
MetadataShow full item record
AuthorsBesada, V. (Victoria); Quelle, C. (Cristina); Andrade, J.M. (José Manuel); Gutiérrez, N. (Noemí); Gómez-Carracedo, M.P. (María Paz); Schultze, F. (Fernando)
Three-way unsupervised pattern recognition
Self-organizing maps (SOMs) (in particular, Matrix reOrganization Layout to Map Analytical Patterns (MOLMAP)) were used to unravel the main patterns in a three-way dataset after a preliminary unfolding of the cube. Eleven sites of the ría of Vigo (NW of Spain) were monitored during the last decade (from 2000 to 2010) to assess pollution trends in this area. Twelve trace metals (Hg, Pb, Cd, Cu, Zn, Cr, As, Li, Fe, Al, Ni and Mn), the total organic carbon and the percentage of fine particles were measured. Results from MOLMAP, the SOM-based approach, were compared to those of three established alternatives: parallel factor analysis, matrix-augmented principal component analysis and generalized Procrustes rotation, the latter two employing unfolding as well. MOLMAP showed the best capabilities to differentiate groups of samples. The spatial and temporal trends, as well as the analytical variables causing them, were almost the same for all methods, which confirms MOLMAP as a simple and ...