A perspective method of the Pasquill-Gifford atmospheric stability classes calculation based on Planetary Boundary Layer model

In case of an accident at a nuclear power plant (NPP) radioactive plume dispersion rate in the atmosphere determines irradiation dose levels for population and depends on the meteorological parameters (atmospheric stability, wind speed, precipitation), topography (underlying surface roughness length, relief) in the region of NPP site and release height. According to Russian and international normative-technical documents in force, irradiation doses forecast for public at initial stage of accidents at NPP shall be made for the most unfavorable weather conditions specific for the region of NPP site. An approach described in section 4 IAEA № 50-SG-S3 and USNRC "Atmospheric Dispersion Models for Potential Accident Consequence Assessments at Nuclear Power Plants" RG 1.145 is accepted for new NPP designs (Leningradskaya NPP-2, Baltic NPP, Byelorusskaya NPP as examples). This approach consists in statistical processing of meteorological data specific for NPP site and using the results for estimation of maximum release dispersion / deposition factors with probability of 99,5 and 95 % at given time intervals in further dose calculations for design basis accidents and beyond design basis accidents respectively. In order to calculate atmospheric dispersion factors for short-term releases the Gaussian dispersion model is traditionally used with approximation for diffusion parameters as functions of the distance from the source and the Pasquill-Gifford atmospheric stability classes given by Smith-Hosker and Briggs formulae. A problem arises regarding collection of initial data sufficient for calculating so high percentiles: - wind speed in the surface boundary layer; - temperature up to the height of ~ 200 m with a vertical step equal to a few meters for further estimation of the mixing layer height, upper and lower boundaries of temperature inversions; - lapse rate and vertical wind velocity gradient. Meteorological data of gradient measurements cannot be used for this purpose as there are not available for a period of 5-7 years at NPP-sites at design stage. It does not seem possible to use standard hydro-meteorological information in problem solving: - because the measurements are taken at a single level (temperature and humidity – at 2 m, wind speed and direction – at 10 m above the ground surface), and these measurement data are not directly applicable to investigation of the vertical structure of planetary boundary layer (PBL), necessary for further calculation using accepted Gaussian model of emergency release dispersion; - as obtaining a limited number of meteorological data combinations does not allow precise evaluation of design dispersion parameters with high realization probability (for instance, 99,5 and 95 % as required). This is explained by the discrecity of the stability parameter used in the Gaussian model, namely “stability class”, leading to formation of identical values on some fragments of the sequence (for example, after processing 23000 measurement data less than 100 different meteorological data combinations are obtained, and therefore design percentiles 99,5 and 95 % of dispersion parameters prove to be identical). In this paper a problem solution is suggested, that consists in the use of numeric PBL-model integrating long-term series of synchronous standard meteorological data available practically for any point on the planet, where an ordinary weather station exists, with re-analyses of atmospheric processes. In order to describe the vertical PBL structure based on measurement data at standard level above the ground surface and synchronous data from standard surfaces 850, 925 hPa, the method is realized using a physically substantiated numerical model, which takes correctly into account the diurnal variations. In principle, any other model describing correctly the diurnal variations may be applied. In this investigation an original model using closing scheme based on equations for the second moments of turbulent fluctuations is applied. For each day during the observation period of 8 - 10 years the Cauchy problem is solved. Some fictive initial conditions are imposed on the vertical temperature distribution constructed on the basis of re-analyses data. In order to calculate initial fields of wind velocity components and turbulence parameters, the stationary problem is solved. Then non-stationary differential equations for horizontally homogeneous PBL are integrated, using measurement data for wind velocity and temperature as lower boundary conditions and re-analyses data at isobaric surface 850 hPa (about 1,5 km) as upper boundary conditions. During the calculation for each hour of the real time all initial data necessary for further dispersion / deposition factor estimation are determined. Then, in conclusion, the dispersion / deposition factors are processed statistically to obtain maximum values of a given probability for given time intervals.