EGU2020-4548
https://doi.org/10.5194/egusphere-egu2020-4548
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Radiocesium wash-off from japanese rivers impacted by the Fukushima accident. Calibration and comparison of simple assessment models for exported dissolved and particulate fluxes by bayesian inference

Laurent Garcia-Sanchez1 and Seiji Hayashi2
Laurent Garcia-Sanchez and Seiji Hayashi
  • 1Institute for Radioprotection and Nuclear Safety (IRSN), Laboratory of Research on Radionuclide Transfers in Terrestrial Ecosystems, France (laurent.garcia-sanchez@irsn.fr)
  • 2National Institute for Environmental Studies (NIES), Fukushima branch, Japan (shayashi@nies.go.jp)

This study aimed at reconstructing the dynamics of radiocesium (134Cs, 137Cs) fluxes exported at the outlet of catchments impacted by the Fukushima Dai-ichi Nuclear Power Plant accident. 

River flowrate, load and dissolved/particulate radiocesium concentrations were simultaneously monitored during the period 2012-2016 in 4 coastal forested catchments of Uda (0.3, 97 km2), Mano (20 km2) and Ohta rivers (43 km2). Precipitation time series were derived from raingauges (JMA) and different satellite datasets (AMeDAS, GSmap, Aphrodite). Contamination maps were derived from MEXT airborne monitoring surveys (MEXT/JAEA).
 A class of stochastic models adapted from Mass Response Functions (originally introduced by Rinaldo & Marani, 1987), was implemented to reconstruct radiocesium fluxes. This theory, describing the coupled transport of water and contaminant particles with transit/holding time distributions, was proposed non only because it encompasses the classical assessment models, but also because it allows some improvements with other hypotheses about the catchment response, notably: transit time distribution of effective rainfall, exchanges between mobile and immobile phases (mass transfer rate, equilibrium concentration) and the macroscopic mass balance at the basin scale.
Classic hypotheses about the catchment response (corresponding to variants of removal coefficients and transfer function models) were tested by Bayesian inference. Inferences were conducted with routines provided by the R environment (R Development Core Team, 2013) and the package BayesianTools (Hartig et al., 2018). The oversampling of extreme events in the monitoring design was counterbalanced by assigning weights to the observations corresponding to the likelihood of the carrier flow rate (liquid or solid flow rate).
For catchments of Mano (43.64 km2) and Ohta (20.28 km2) rivers, observations covered both low-flow and high-flow periods. The short-term fluctuations of the wash-off catchment response were strongly transport-limited: dissolved 137Cs fluxes varied linearly with river flow rate (m3/s), whereas particulate 137Cs fluxes varied linearly with solid flow rate (kg/s) or nonlinearly with river flow rate (m3/s). The longer-term decline of radionuclide availability to wash-off was credible for dissolved wash-off, but was not plausible for solid wash-off, certainly due to the short period of observations of the monitorings. 
For assessment purposes, the removal coefficient approach appeared as a good option for both dissolved and particulate 137Cs wash-off. For Mano and Ohta catchments, median model predictions agreed with observations within a factor ranging from 1.47 to 1.66 for dissolved wash-off, and from 19.22 to 22.44 for particulate wash-off. The explicative power of the proposed models will need to be updated when more recent measurements are available and their predictive power needs to be confirmed on independent observations on other catchments.

How to cite: Garcia-Sanchez, L. and Hayashi, S.: Radiocesium wash-off from japanese rivers impacted by the Fukushima accident. Calibration and comparison of simple assessment models for exported dissolved and particulate fluxes by bayesian inference, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-4548, https://doi.org/10.5194/egusphere-egu2020-4548, 2020