Intercomparison of numerical atmospheric dispersion prediction models for emergency response to emissions of radionuclides with limited source information in the Fukushima Dai-ichi nuclear power plant accident

Abstract

The utilization of numerical atmospheric dispersion prediction (NDP) models for assisting the emergency response to emission of radionuclides has been recommended by a working group of the Meteorological Society of Japan. This paper verifies the feasibility of the recommendation through NDP model intercomparison with limited emission source information for the case of the Fukushima Dai-ichi Nuclear Power Plant accident caused by the Great East Japan Earthquake in 2011. According to the recommendation of the working group, the NDP models are run under the assumption of a constant rate of emission during the whole forecast period. This is the worst-case scenario when limited source information is available. Generally, no information is provided on the temporal variability and strength of the emissions, while the source location is known. Surface air radionuclide forecasts are utilized for providing warnings of the risk of inhaling radioactive substances suspended in the low-level atmosphere, whereas column-integrated radionuclide forecasts are utilized for estimating the potential maximum wet deposition of radioactive materials on the ground due to precipitation. The NDP model short-range forecasts were validated with observational data for three locations, at the times when the most serious contamination events occurred at each of the three monitoring stations. The NDP models successfully predicted the risk of surface air contamination and/or ground surface contamination caused by wet deposition in these cases. Particularly, the NDP model forecasts allow us to disseminate warnings at effective lead times before exposure to radiation. The different NDP models gradually deviate their forecasts as the lead time progresses. The deviations may indicate the magnitude of forecast errors. Thus, the use of multi-model forecasts is of greater benefit than the single model forecasts, because forecast error information is suggested.