A model study for determining the minimum sample size for a percentile estimation

In the field of environmental noise evaluation, there is the basic problem of determining how much observed data should be used for the calculation of the noise evaluation index Lχ(i.e., (100 - χ) percentile level of the cumulative distribution function) from the viewpoint of reliability. In this correspondence, we will consider this problem theoretically by use of the order statistics and give a method of determining the minimum sample size for the Lχestimation.