Representativeness of the measurement sample in the determination of the long-term noise indicators of traffic noise

Abstract

This paper concerns the representativeness of a measurement sample for traffic noise measurements. On the basis of data coming from a fourteen-year-long constant noise monitoring in Kraków, there are statistically significant differences between noise levels of different categories across the year with regard to seasons, days of the week, months and working days vs. holidays and Sundays. These differences were proven to be substantive. When the measurement sample is drawn from a group of the days with an average sound level even slightly different from the global average, the coverage probability decreases by several to several dozen percent; the larger the sample size, the larger the decrease. Therefore, the commonly used methods, such as taking measurements for a few subsequent days or drawing the measuring days from a short time interval, result in a failure to meet the legal requirement for covering the actual value with a 95% uncertainty interval.