Annual traffic noise levels estimation based on temporal stratification

This paper proposes a temporal sampling strategy that increases the accuracy of long-term noise level estimation and allows to establish the estimation error according to the number of sampled days. Days of the week are stratified into working days and weekend days. This research shows how to use measurements of L_eq on working days to estimate the corresponding values for weekend days. This is possible because working days have higher noise levels and less variability than weekend days. The improvement in accuracy allows for a reduction in the number of required sampled days compared to taking samples randomly, which would help to reduce the uncertainty in environmental noise assessment. As a reference, to obtain a 90% confidence interval of ±1 dB for L_day, the proposed sampling strategy reduces the required measurement days by more than 38%. For L_DEN, the reduction is close to 18% of the total number of days. The proposed strategy could be adapted to different environments by simply changing a few parameters.