Self-adaptive grids for noise mapping

Often, the quality of the results in a noise map is expressed as a comparison between measured and calculated noise levels at several sample locations spread over the map. Although, under some circumstances this could be considered as a valid approximation, it excludes from the noise mapping process two very important stages: interpolation and classification. The two tasks must be applied to obtain a full map, with contours and isolines that allow characterizing the noise levels at every location in it. The grid of receivers used for calculations has an influence in the final results. We could summarize that, the thinner the grid (more receivers), the better the results. As the receivers’ density increases, the accuracy of the map is improved, but the computational costs will also increase exponentially. Furthermore, most of the receivers do not contribute to the optimization of the map quality. The main objective of this paper is to present a method that substantially improves the quality of a map at its isolines by using self-adaptive grids. Self-adaptive grids are created using a smart iterative algorithm that analyses previous knowledge to concentrate the calculation effort in those areas where it will be more effective. Because of this, it can provide better results than higher resolutions grids, while using fewer receivers for calculation and interpolation. Self-adaptive grids improve the results obtained by any starting grid, no matter the resolution of the grid, the interpolation method (idw, spline, kriging,..), the number of receivers, or the way they have been located (random, triangulated or equal spaced). As further knowledge (extra receivers) is provided as in input to the interpolator, the results will be improved. Self-adaptive grids have been proven to be an excellent method especially for open spaces with few obstacles. The use of a coarse initial grids, allow the number of calculated receivers to be reduced enormously, while improving the accuracy of the map. The calculation time is drastically reduced, as the number of calculation receivers is reduced. Because of this reason, this method may be very useful for grid refinement in models that do not consider obstacles in propagation paths (for instance, Integrated Noise Mapping, INM, for airports noise mapping).