Population exposure calculations for entire cities are applications with tough requirements on performance. To describe the rapid decrease in pollutant concentrations close to roads, it is necessary to describe the dispersion of pollutants with high spatial resolution.
To facilitate this type of calculations, a new feature is under development that allows the user to specify locally refined computational grids. This makes it possible to concentrate the computational resources on the areas where the gradients in pollutant concentrations are largest, for example close to roads and other major emission sources. The solution uses a quad-tree representation of the computational grid.
Development finishes in 2011
The development will be finished during late 2011. The initial development was funded by the Swedish EPA. Below is given an example of a dispersion modeling result using the new gridding technique.
Strategies to improve efficiency
Airviro already provides several ways to improve efficiency. Through parallelization the computations are divided into several smaller tasks to make use of modern multi-core processors or server clusters.
Nesting can, among others, be used to calculate background concentrations for a whole city while still describing the contribution from local sources with a high degree of detail.
A classic way of reducing computational times for large areas is the use of Climatologies instead of time-series simulations. Meteorological data are then processed to form a so called climatology (a library) of meteorological conditions observed in the area. Instead of running the dispersion model for all hours during a time-period, calculations are only made for the meteorological conditions in the climatology. This reduces the computational time dramatically.