Simplified data assimilation for simulating wet deposition distribution of radioactive materials in FDNPP accident

Abstract:

Due to the massive earthquakes and tsunami on March 11th 2011 in Eastern Japan, Fukushima Daiichi Nuclear Power Plant (FDNPP) was severely damaged. Radioactive materials were released and spread out by atmospheric advection-diffusion. Especially on March 21 - 23th when precipitation was observed, “hotspot” where the high concentration was detected locally. This area was formed in the metropolitan area in Kanto region. Thus, pollution at water treatment plants because of the deposition became a concern. Therefore, the reliable information of the hotspot is expected. Currently, atmospheric transport simulations by numerical models are developed for reproduction of the distribution. However, there are some uncertainties in the simulations. In the case of hotspot, accuracy of simulated precipitation have to be well considered because the hotspot seemed to be formed by wet deposition.

 We modified the stable isotope mode of Regional Spectral Model (IsoRSM) to enable to simulate the transport of the radioactive tracers, namely ¹³¹I and ¹³⁷Cs, by including the dry and wet deposition processes. As the simplified data assimilation, simulated precipitation was replaced with Radar-AMeDAS precipitation data (RAP). RAP was assimilated in the post-process, after running simulations, to redistribute wet deposition of ¹³⁷Cs. The ratio of ¹³⁷Cs deposited from the cumulative vertical column with precipitation in the domain was not changed, however its pattern was redistributed corresponding with RAP and simulated concentration. As a result, the redistributed wet deposition was within factor 10 to 2 compared with the fallout data in Kanto region, and further data assimilation would be contributed. In addition, we found that due to the arrival time of the plume in the morning on 21st and the border time of daily observation data of fallout, validation result might be worse even though hourly distributions are well simulated.