The current dataset was compiled to study sediment fingerprintings practices, i.e tracer selection and contribution modelling. Colorimetric properties analysed with a portable diffuse reflectance spectrophotometer (Konica Minolta CM-700d) and geochemical contents obtained with an energy dispersive X-ray fluorescence spectrometer (ED-XRF Epsilon 4) were analysed in potential source material that may supply sediment to coastal rivers draining the main Fukushima radioactive pollution plume (Japan). Three potential soil source materials (n = 56) were considered: cropland (n = 24), as non-decontaminated soil before the application of local decontamination policies: forest soils (n = 22) and subsurface material originating from channel bank collapse or landslides (n = 10). A sediment core was collected in the Mano Dam lake (Hayama lake) on the 6th June 2021 and was sectionned into 1-cm layers (n = 34). Laboratory mixtures (n = 27) were made to assess different contribution levels from the sources. In addition to colorimetric and geochemical properties, organic matter and stable isotopes were analysed by EA-IRMS for sources and sediments samples. The current dataset comprises three Excel files including the metadata description, the data itself and a file describing the composition of laboratory mixtures prepared to provide a dataset to calibrate/validate un-mixing models implemented to address this research question and analysed in the same conditions and using the same equipment as the source/target material. Recommended encoding format: latin1