大気汚染物質の放出量推定における精度向上に関する研究

The purpose of this report is the improvement of the source intensity estimation based on a minimal residual method using observed and calculated data of concentration, radiation dose and so on.This source intensity estimation method uses short range observed data (e.g. a few km) and calculated results of Gaussian plume model that is suitable for short range dispersion calculation. In Fukushima Dai-ichi nuclear power plant accident, short range observed data had not been used for source intensity estimation, because it's difficult for the Emergency response model (named SPEEDI) to resolute spaces of calculation region with less than 1 km.The accuracy of source intensity estimation significantly decreases due to several uncertainties. The 4 uncertainties of the source intensity estimation were investigated and their improvements were verified.1) Uncertainty of wind direction was decreased by expanding observation time of concentration.2) Uncertainty of emission height was decreased by removing observation data measured near from source point.3) Terrain influence was reproduced in Gaussian plume model by modifying plume widths, effective source height and displacement of plume axis, based on wind tunnel experiment.4) Meandering effect of Atmospheric stability on lateral plume width was reproduced in Gaussian plume model by using the observed data on fluctuation of wind direction in the field under a stable atmospheric stability.According to “Meteorology guideline for nuclear power facilities safety analysis”, wind directions of meteorological observation data were categorized into 16 directions and concentration calculated from a mean value for few minutes. Consequently, the wind direction can include an error. The uncertainty caused by wind direction error was improved by using data on observation, and dispersion simulation that calculated a mean value for 1 hour. Because of lateral dispersion width is proportional to the variance of wind direction and increase with observation time, concentration distribution is smoothed for 1 hour, more than few minutes.When a release from a nuclear reactor building is imagined in a severe accident, an uncertainty of release height can be inherent. The ground-level concentration distribution does not depend on the release height over the downwind distance of more than 1000m, but they depend on it in vicinity of the source. The uncertainty due to release height was improved by not using data on observation and dispersion s