Statistical Tools and Optoelectronic Measuring Instruments

Abstract

In the frame of European research projects, several air quality measuring campaigns in cross roads, streets, parks as well in a non ecological waste deposit were realized. The analyzed signals, representing CO, NO2, O3, SO2 and HC concentrations, were measured with several optoelectronic instruments. Two of the utilized optoelectronic devices are shortly presented at the beginning of the chapter. Due to their random character, pollutant concentrations signals can be analysed using statistical processing methods. The main statistical functions and parameters taken into account within this chapter are histograms, correlation coefficients, correlation and covariance functions (Ionel et al., 2009). Actually, statistical tools are usually utilized in analysing ecological data (Zuur et al., 2007) but, as far we know, it is not common to imply statistics in a comparative analysis of optoelectronic devices (Ionel et al., 2007). Specific pre-processing procedures must be used for signal conditioning. Thus, „ideal“ lowpass filtering based on fast Fourier transform can be implemented for the rejection of measurement noise and artefacts from the pollutant concentration signals. On the other hand, „ideal“ high-pass filtering allows the extraction of the variable component of the pollutant level signals. In order to avoid redundant measurements, one can use interpolation for increasing the number of samples, especially in the case of slowly varying meteorological parameters. Computer experiments with real pollutant concentration signals lead to some practical recommendations concerning acquisition parameters like data size and sampling frequency. The most important practical rules are as follow: assure the temporal length of the measured signal, assure the necessary resolution on the time axis, and make interactive verifications of the acquisition parameters during the measuring campaign. Guidance on MATLAB software for calculating statistical functions and parameters are provided. As a particular application, the correlative comparison of two carbon monoxide (CO) measuring instruments is presented. The point source device and the open path optical remote sensing instrument do actually not measure the same quantity but a statistical comparison of the two instruments is still possible. The correlative analysis leads to the expected conclusion that the open path instrument is more suitable for monitoring the pollution level in a large area than the classical point source device.