**Published:** Jan 1979

Format |
Pages |
Price |
||

PDF Version (172K) | 11 | $25 | ADD TO CART | |

Complete Source PDF (5.2M) | 11 | $55 | ADD TO CART |

**Source: **STP692-EB

Statistical techniques allow the prediction of the error (expressed as a standard deviation about the measured energy equivalent sound level (*L*^{eq}) or time exceedance level (*L*^{x})) due to temporal sampling of a time-varying signal. To first approximation, *p*^{2}) is the standard deviation of pressure squared about its mean ⟨ *p*^{2} ⟩, and *N* is the number of independent data points in the sample. These predicted standard deviations are compared with experimentally determined values, obtained by computer sampling of large sets of actual field data recorded at a highway, an airport, a suburban road, and a rural residence. The predicted standard deviations, for the example given, of sampling at 30 s intervals for 1 h, range from 0.3 to 8 dB. These values agree closely with the experimentally determined values, and the small remaining discrepancies are explained. For “continuous” (or very frequent) sampling, an “effective number” of independent samples is used, computed from the autocorrelation in the time-varying signal.

**Keywords:**

noise, community noise, noise exposure, sampling (temporal), measurement errors, time-varying signals, statistics

**Author Information:**

Mansbach, PA *Physicists (acoustics), National Bureau of Standards, Washington, D.C.*

Corley, DM *Physicists (acoustics), National Bureau of Standards, Washington, D.C.*

**Paper ID:** STP35969S

**Committee/Subcommittee:** E33.03

**DOI:** 10.1520/STP35969S