The paper deals with the development of mathematical model and numerical algorithms for solving the problem of transfer and diffusion of aerosol emissions in the atmospheric boundary layer. The model takes into account several significant parameters such as terrain relief, characteristics of underlying surface and weather-climatic factors. A series of numerical experiments were conducted based on the given model. The obtained results presented here show how these factors affect aerosol emissions spread in the atmosphere.

Manual preprocessing of raw data for air pollution modeling is often turns to quite painstaking job. In order to automate some basic routine of preprocessing the input information, we propose a method for consolidation of meteorological and spatial data provided by open web-services and repositories. The implemented algorithm can significantly reduce the time and costs for parameterization of the atmospheric boundary layer before solving the problems of evaluation and forecasting the environmental conditions of the industrial regions.

We report the verification and validation of the numerical core of an atmospheric flow and pollutant-dispersion module, in which the Reynolds-averaged Navier–Stokes equations with a k–ε closure and an advection–diffusion equation for pollutants are discretised by a conservative finite-volume method. Convective schemes, linear solvers and the pressure-correction algorithm were tested against manufactured solutions, a Poisson problem, a positivity problem and the lid-driven cavity. The street-canyon prototype itself was verified with a manufactured solution on its own grid, a turbulent-channel test of the wall functions and an empty-domain test of the inflow profiles; these tests revealed two implementation errors, which were corrected. The corrected model, with standard constants, was validated against the Karlsruhe street-canyon wind-tunnel data, and the computed wall concentrations agree with the measurements within a factor of two. A sensitivity study was carried out to assess the effect of the grid, the turbulent Schmidt number, the inflow turbulence, the convective scheme and the domain size on the results.