Tracking the dispersion of contaminated sediments in Fukushima coastal catchments

Large quantities of radionuclides were released into the atmosphere by the Fukushima Dai-Ichi Nuclear Power Plant (FDNPP) after the earthquake-triggered tsunami devastated the eastern coast of Japan on March 11, 2011. Many of these radionuclides (20%) were deposited on soils of the Fukushima Prefecture. This PhD thesis develops original fingerprinting methods to track the dispersion of contaminated particles following this accident. The study focuses on 3 coastal catchments north of the FDNPP (Mano - 175 km², Niida - 270 km² and Ota - 75 km²) draining heavily contaminated areas of the radioactive plume. The Fukushima Prefecture is characterized by an erosive climate, with the occurrence of spring floods and summer typhoons. To study the dispersion of the radioactive contamination, soil samples and sediment drape deposits were collected during 6 sampling campaigns (every six months between November 2011 and May 2014; i.e., after the major flood events). Each sample was analyzed by gamma spectrometry to determine radionuclide activities, and several soil and sediment samples were also analyzed by neutron activation analysis to determine their geochemistry. First, the analysis of ¹³⁷Cs activity in 10 soil cores collected in paddy fields confirmed the limited migration of radiocesium with depth in the soils of the coastal catchments. More than 90% of the contamination was still concentrated in the uppermost 2 cm of the soils by November 2013. Particles contaminated were therefore available for mobilization and transport downstream by processes that govern soil erosion. Second, metastable silver-110 (^110mAg) was detected in most of the samples collected between November 2011 and November 2012, and our investigation showed that this radionuclide has a similar behavior as ¹³⁷Cs in soil and sediment. Consequently, we used ^110mAg to track the dispersion of the contamination as the ^110mAg/¹³⁷Cs activity ratio in soils of the Niida catchment showed significant differences between upstream and downstream locations. The use of a binary mixing model allowed the identification of a seasonal cycle of erosion and dispersion of particles. However, as ^110mAg has a short half-life (250 days), it rapidly decayed and could not be detected anymore by May 2013. To overcome its disappearance, the contribution of soils located on the mountainous plateaus to the sediment transiting the river in the coastal plains was quantified based on their ¹³⁷Cs signature. Binary mixing models were used, based on the distributions of ¹³⁷Cs in mountainous areas (≥ 20 kBq/m²) and in coastal plains (≤ 20 kBq/m²). The results demonstrated that the contribution of the mountainous area varied in the different catchments. In the Niida catchment where no dam has been built, the mountainous area supplies more sediment to the river (≅46%) than in the Mano catchment that has a dam (≅20%). These results show the impact of dams generating a sediment dis-connectivity. Finally, the soil map of the region was used in order to identify the soil types that may supply sediment to the rivers. The main soil types (Andosols, Cambisols and Fluvisols) were characterized by their geochemical composition, and Sc and Yb were identified as the most discriminant elements. The distributions of these elements in the three sources were used in a mixing model. Results show that Fluvisols are the main source supplying ≥70% of sediment to the rivers in both catchments. This soil type is mainly found in paddy fields, which confirms the enhanced erodibility of these cultivated areas. In the future, the dataset compiled could be used to improve soil erosion model operating at the catchment scale. Moreover, the impact of the ongoing decontamination works on the dispersion of contaminated sediments should be investigated. (author)