Local atmospheric transport behaviors of representative radionuclides during the Fukushima accident: A 200-m-resolution cross-scale study from site to 20 km

Abstract

Modeling of local atmospheric radionuclide transport is essential for nuclear emergency response. However, very few studies of the Fukushima Dai-ichi nuclear power plant (FDNPP) accident have focused on this topic because of the complex meteorology and the cross-scale transport behaviors from the site to 20 km of the FDNPP. In this study, both the local meteorology and transport behaviors were investigated at high resolution (200 m) using ensembles of different meteorology and models. Four wind fields calculated from onsite observations and three regional-scale meteorological fields (i.e., the 1-km ECMWF, 3-km and 1-km NHM-LETKF), and two transport models: the RIMPUFF Lagrangian puff model and the SPRAY particle model were considered and combined with each other. These eight simulations and their ensemble mean were analyzed based on onsite observations of wind and gamma dose rates, and local-scale observations of 137Cs concentration. Results revealed that at the site, the Onsite wind field which captured the frequently changing wind, best reproduced the onsite gamma dose rates with the grid resolution of 200 m. At the local scale (up to 20 km), the observations present a smoother temporal change. The wind fields assimilated with Japanese domestic observations presented advantageous performance, and the 1-km NHM-LETKF achieved the best score of the factor of 5 metric of 0.49 for the simulated 137Cs concentration. The SPRAY coupled with the three-dimensional (3D) convolution method and RIMPUFF showed better performance in simulating the onsite gamma dose rate and the local-scale concentration, respectively. The ensemble mean achieved robust metrics, better simulated the baseline of onsite gamma dose rates, and reproduced more peaks of local-scale concentration at the expense of peak value deviation.