Air Pollution Modelling with Perturbational Downscaling

Abstract

A technique of downscaling based on the reformulation of the governing equations in the perturbational form with arbitrary amplitude of the perturbbations had been introduced in our previous works and successfully applied in meteorological modelling. In particular, preliminary results of testing such a technique, built-in into the mesoscale met model TVM, have indicated that it has certain advantages as compared with conventional methods of downscaling based on the brute-force damping of computational errors in the strip of cells located near the boundaries of the downscaled domain. The present paper is devoted to application of this technique in the atmospheric dispersion modelling. Because the case of the continuous point source located inside the downscaled domain seems to be most critical, the problem of numerical approximation of the source term on nested grids is considered first of all. In distinction from NWP problems, the source term in dispersion modelling is extremely sensitive to the spatial resolution, and its approximations in the whole computational domain and in the downscaled domain should be in concert one with another.

The efficiency of perturbational downscaling is tested first in the model case where the advection-diffusion equation (ADE) has an analytical solution. It is demonstrated that the perturbational downscaling in a given sub-domain provides the numerical solution of ADE in this sub-domain “of the same quality” as the direct numerical solution of ADE with the use of fine grid uniformly in the whole domain. Work is on-going towards the practical application of the perturbational down-scaling using the TVM model as a meteorological driver for the SILAM dispersion model. A specific case of an accidental release in Europe is considered as a test scenario where the sensitivity of the concentration field to the spatial resolution in NWP and dispersion models is high enough to justify nesting efforts. The ETEX first release case is used for the trial application as a prominent example.