Sensitivity analysis on the wet deposition parameterization for 137Cs transport modeling following the Fukushima Daiichi Nuclear Power Plant accident

Wet deposition has been identified as a critical impactor for the modelling of ¹³⁷Cs in the Fukushima Daiichi Nuclear power plant (FDNPP) accident. However, it is difficult to simulate due to the involvement of close interaction between various complicated meteorological and physical processes during the wet deposition process. The limitation of measurement of the in-cloud and below-cloud scavenging also contribute to the uncertainty in wet deposition modeling, leading to the great variation of ¹³⁷Cs wet deposition parameterization. These variations can be amplified further by inaccurate meteorological input, making simulation of radionuclide transport sensitive to the choice of wet scavenging parameterization. Moreover, simulations can also be influenced by differences between radionuclide transport models, even if they adopt similar parameterization for wet scavenging. Although intensively investigated, wet deposition simulation is still subject to uncertainties of meteorological inputs and wet scavenging modeling, leading to biased ¹³⁷Cs transport prediction.

To improve modeling of ¹³⁷Cs transport, both in- and below-cloud wet scavenging schemes were integrated into the Weather Research and Forecasting-Chemistry (WRF-Chem) model, yielding online coupled modeling of meteorology and the two wet scavenging processes. Overall, 25 combinations of different in- and below-cloud scavenging schemes of ¹³⁷Cs, covering most wet scavenging schemes reported in the literature, were integrated into WRF-Chem. Additionally, two microphysics schemes were compared to improve the simulation of precipitation. These 25 models and the ensemble mean of 9 representative models were systematically compared with a previous below-cloud-only WRF-Chem model, using the cumulative deposition and atmospheric concentrations of ¹³⁷Cs measurements. The findings could elucidate the range of variation among these schemes both within and across the five in-cloud groups, reveal the behaviors and sensitivities of different schemes in different scenarios.

The results revealed that the Morrison's double moment cloud microphysics scheme improves the simulation of rainfall and deposition pattern. Furthermore, the integration of the in-cloud schemes in WRF-Chem substantially reduces the bias in the cumulative deposition simulation, especially in the Nakadori and Tochigi regions where light rain dominated. For atmospheric concentration of ¹³⁷Cs, those models with in-cloud schemes that consider cloud parameters showed better and more stable performance, among which Hertel-Bakla performed best for atmospheric concentration and Roselle-Apsimon performed best for both deposition and atmospheric concentration. In contrast, the in-cloud schemes that rely solely on rain intensity were found sensitive to the meteorological conditions and showed varied performance in relation to the plume events examined. The analysis based on the spatial pattern shows that the Roselle scheme, which considers cloud liquid water content and depth, can achieve a more balanced allocation of ¹³⁷Cs between the air and the ground in these two cases than that achieved by the empirical power function scheme Environ. The ensemble mean achieves satisfactory performance except for one plume event, but still outperforms most models. The range of variation of the 25 models covered most of the measurements, reflecting the reasonable capability of WRF-Chem for modeling ¹³⁷Cs transport.