Autocorrelation and Variability of Indoor Air Quality Measurements

Abstract

Measurements of gaseous and particulate concentrations are used to characterize the indoor environment, but such measurements may reflect temporary conditions that are not representative of longer time periods. Moreover, indoor air quality (IAQ) measurements are autocorrelated, a result of limited mixing and air exchange, cyclic emissions, HVAC operation, and other factors. This article analyzes the autocorrelation and variability of IAQ measurements using time series analysis techniques in conjunction with a simple IAQ model. Autocorrelations may be estimated using the air exchange rate (alpha) and ventilation effectiveness (epsilon) of the building or room under study, or estimated from pollutant measurements. From this, the variability, required sample size, and other sampling parameters are estimated. The method is tested in a case study in which particle number, fungi, bacteria, and carbon dioxide concentrations were continuously measured in an office building over a 1-week period. The estimated air exchange rate (1.4/hr) for area studied was predicted to yield autocorrelation coefficients of approximately 0.5 for measurements collected on 30-min intervals. Autocorrelation coefficients based on airborne measurements (lag 0.5 hr) ranged from 0.5 to 0.7 for 1-25 microm diameter particles, fungi, and CO2, but near zero for particles < or =1 microm diameter and bacteria. As expected, the variability of measurements with the lowest autocorrelation decreased the most at long sampling times. The implications for spaces with low alpha * epsilon products are that measurements may not benefit significantly from longer averaging periods, measurements on any single day may not be representative, and day-to-day variability may be significant. Steps to determine sample sizes, averaging times, and sampling strategies that can improve the representativeness of IAQ measurements are discussed.