In this work, heavy metal related soil- and plant contamination of two floodplain areas near the Middle Elbe River was analysed. The potential of non-imaging and imaging hyperspectral data for the quantification of the total metal contents was investigated. For this, the spectral range between the visible light and the short wave infrared (0.35 - 2.50 μm) was considered and various multivariate statistical and machine learning algorithms were used. Based on soil spectra, reliable model outcomes were obtained for the majority of the analysed elements, with the best prediction results for Pb (R² = 0.89) and Cd (R² = 0.86). Interactions between spectral plant parameters and metal concentrations were weaker. Promising results were generated for Cd (R² = 0.75), As (R² = 0.61) and Pb (R² = 0.56). This thesis shows that hyperspectral information can be used to support the monitoring of floodplain areas and to quantify selected heavy metals with a high accuracy.