Comparison of confidence interval models for long-term road noise indicators with a semivariance-based approach

This study compares confidence interval models for long-term road traffic noise indicators L d , L e , L n , and L den , including classical approaches and he author's method for constructing asymmetric intervals based on semivariance.The analysis covers intervals derived from the arithmetic mean, logarithmic mean, energy mean, Taylor series approximation, and semivariance-based constructions.The comparison was performed using daily equivalent sound level data from continuous road noise monitoring in Gdańsk over a full calendar year.Method performance was evaluated through empirical coverage of the true annual indicator values for sample sizes ranging from 5 to 50 observations using repeated random sampling.The results indicate substantial differences among the analyzed methods.Semivariance-based intervals provided more stable coverage, particularly for small samples and distributions exhibiting asymmetry or heavy tails.These findings suggest that semivariance-based confidence intervals may offer a useful tool for uncertainty assessment of long-term environmental noise indicators.