Sampling strategies for monitoring noise in the vicinity of airports

Abstract

This paper is the third in a series dealing with the development of temporal sampling strategies for estimation of mean noise levels in the vicinity of airports. It extends the previous analysis for westcoast, one-direction airports (due to prevailing winds) to eastcoast, multidirection airports (Boston Logan, Washington Dulles, and National). The results show that the data for many of the eastcoast airport sites are nonstationary in the mean level and the corresponding consecutive sampling requirements predicted by the Dynamic Data System (DDS) methodology are very large, at times exceeding 1/3 of a year. When the data are stationary, Monte Carlo simulations using the data produce sampling requirements comparable to the values obtained by the DDS methodology. However, the DDS methodology tends to overestimate sampling requirements for nonstationary data. The simulations demonstrate that nonconsecutive sampling strategies reduce the overall sampling requirements for nonstationary data. In general, the results reveal the following: (a) Westcoast (one-direction); ±50% precision—four weeks, any sampling strategy, ±35% precision—eight weeks, any sampling strategy. (b) Eastcoast (multidirection); ±60% precision—four weeks, one from each quarter, ±40% precision—eight weeks, one from each eighth.