Improvement of forecast noise levels in confined spaces by means of geostatistical methods

The aim of this paper is to improve, through the application of geostatistical methods, the results provided by forecasting models for determining the noise level in confined spaces. After a brief glance at the legislative aspect of assessing noise risk in working environments, the importance is stressed of the space-time discretization of the work cycle for the purpose of the aforesaid assessment under the most general conditions, and hence of the need to have the sound pressure level map available for every time element. Attention was then shifted to the spatial reconstruction of the sound pressure levels by means of forecasting models. In particular, in a case prepared ad hoc, the discrepancy was evidenced, by means of a set of regular grid measurements, between forecasting and reality, characterized by a strong bias varying in size from one point to another. The external drift method, applied using a few measurements of the primary variable and the output of the model as the external drift, proved to be very effective in removing the bias.