## Data Processing

The concentration of many air pollutants follows a log normal rather than a normal distribution. In a log normal distribution, a plot of the logarithm of the measured values more closely approximates the bell-shaped Gaussian distribution curve than does a plot of the numerical data. Suspended particulate concentrations are a prime example of this type of distribution. In this case, the geometric mean is the statistical parameter that best describes the population of data. The arithmetic mean is of limited value because it is dominated by a few occurrences of high values. The geometric mean, combined with the geometric standard deviation, completely describes a frequency distribution for a log normally distributed pollutant.

Environmental engineers should consider the averaging time over which sample results are reported in processing and interpreting air quality data. For sulfur dioxide, various agencies have promulgated air quality standards based on annual arithmetic average, monthly arithmetic average, weekly arithmetic average, 24-hr arithmetic average, 3-hr arithmetic average, and 1-hr arithmetic average concentrations. The output of a continuous analyzer can be averaged over nearly any discrete time interval. To reduce the computation time, an environmental engineer must consider the time interval over which continuous analyzer output is averaged to obtain a discrete input for the calculations. If a 1-hr average concentration is the shortest time interval of value in interpreting the study results, using a 1- or 2-min averaging time for input to the computation program is not economic.

Environmental engineers must exercise caution in using strip chart recorders to acquire air quality data. The experience of many organizations, both governmental and industrial, is that the reduction of data from strip charts is tedious. Many organizations decide that they do not really need all that data once they find a large backlog of unreduced strip charts. Two cautions are suggested by this experience. First, only data that is to be used should be collected. Second, magnetic tape data storage followed by computer processing has advantages.

The visual display of air quality data has considerable appeal to many nontechnical personnel. Long columns of numbers can be deceptive if only one or two important trends are shown. The use of bar charts or graphs is frequently advantageous even though they do not show the complete history of air quality over a time span.

## Post a comment