Both sensory as well as chemical analyses were performed
along with a comparison between the two. Table 1 shows time
based odor concentrations of various odorous samples aftermodels. The raw sewage sample was the most intense from the
odor perception point of view. All the component specific GC–MS
based analysis was also completed using the same raw sewage
sample. Beidler’s law (Appendix A) was applied to convert the
intensity reports to equivalent odor concentration terms in
relevant units. Parameter estimation (based on Levenberg–
Marquardt algorithm) results are presented in Table 5 where
the uncertainties and confidence intervals of each parameter were
given for each of the models for various samples collected from the
STP. Models were ranked according to their performance in the
nonlinear least square fit and their respective values of the
minimized chi-squared function, x2 (Table 5). Figs. Fig. A.1a and
Fig. A.2a show the odor intensity reported by the panel sniffers for
various sewage concentrations made out of the two raw sewage
samples, namely STPA1 and STPA2. Residual intensities were
plotted against measured sewage concentrations in Figs. Fig. A.1b
and Fig. A.2b respectively. In case of trickling filter based STP, the
power law performed better than the Weber–Fechner law for less
intense sewage samples (see Appendix A), using the scaling
technique of category estimation as opposed to ratio scaling.
Beidler’s equation represented the intensity–concentration relationship
for sewage better for samples with more intense odor.