Multivariat dataanalys och gisutvärdering av metallsammansättningen i jord

This report was made to evaluate the multivariate data analysis as a tool to investigate the metal composition of soil with the use of principal component analysis (PCA), and in combination with GIS interpolation. Data with several variables and observations can be difficult to analyze with only descriptive diagrams. PCA is one way to reduce the complexity of several variables without losing too much information and thus increases the interpretation. The aim was to examine if the metal composition changes with depth and in different soils, as well as whether GIS can account for the spatial distribution in an area with results from the PCAs. Analysis results were obtained from SGU geochemical survey within Skellefteå municipality and data from Tyréns four sample places: Hedensbyn, Näsudden, Örviken and Sörböle. The studied metals were arsenic (As), lead (Pb), chromium (Cr), copper (Cu), cobalt (Co), cadmium (Cd), zinc (Zn) and nickel (Ni). PCA were performed with two different data analysis programs, SIMCA and PAST. Results showed that multivariate data analysis can show differences and similarities in metal composition in soil, both for depth and in different soils. Sample depth had the biggest impact on the metal composition. Interpolation of PCA results can show the spatial distribution of the metal composition in soil within an area. Thus is GIS a useful tool when interpreting multivariat data analysis results, and can be used to identify sources to contamination.