Potential bias from non-continuous monitoring of ambient ozone concentrations for characterizing hourly and daily 7- and 12-hour average concentrations

Researchers using chamber experiments to investigate O₃ effects on vegetation do not always monitor O₃ concentrations continuously. Errors introduced because of non-continuous monitoring during experimental treatments may result in the inadequate characterization of hourly average concentrations, with the consequence that summarized exposure indices may provide inaccurate linkages between experimental results and the standard-setting process. We have investigated the uncertainties associated with limited sampling to characterize (1) hourly average and (2) daily 7- and 12-h average O₃ concentrations. As expected, accuracy increased with increased sampling; over- and underestimation of hourly values was directly related to whether the actual concentrations were increasing, decreasing or stable from one hour to the next. When non-continuous sampling was used at sites experiencing small changes from one hour to the next, more than 85 and 60% of the estimated hourly average values were within ±5 and ±2.5 ppb of the actual values, respectively. For a site where average O₃ concentrations changed rapidly from one hour to the next, a greater error was observed. For the daily 7- and 12-h average values, when hourly average concentrations did not change rapidly from one hour to the next, 85% or more of the estimated values agreed within ±2 ppb, independent of sampling frequency. Greater uncertainty was experienced when hourly average concentrations changed at a faster rate. When considering the uncertainties associated with non-continuous sampling, investigators should specify the level of accuracy needed to characterize hourly average concentrations and then select the number of samples per hour required to meet that objective.