4.0 DISCUSSION 



Laboratory studies have demonstrated the capabihty of UV-R to detect various PAH 

 analytes and quantify concentrations in the range of 10-100 ppm and higher. PAH data 

 determined on traditional chemical analyses of grab samples, show the concentrations of 

 total PAHs at State Pier and Sassafras Point (20 and 30 ppm, respectively) have directly 

 proportional fluorescence intensities (365 nm excitation; 400 and 6(X) maximum ADU, 

 respectively; Figure 3). This suggests a possible correlation between the total PAH field 

 concentrations and the derived intensity spectra from the UV-R, although obviously a more 

 rigorous experimental design is needed to establish this apparent correlation. 



Spectral differences were observed between stations which may be related to the 

 relative concentration of specific PAH analytes, particularly pyrene. Again, definitive 

 analysis of PAH composition will require more extensive field and laboratory testing. 

 Here, one of the primary needs is the development of PAH spectral libraries and non- 

 parametric statistical analysis routines (e.g., signal theory) to decode the PAH signature 

 into its component parts. In this effort, a focus on a few PAH analytes which are 

 surrogates for overall PAH transport pathways (e.g., pyrene for pyrogenic compounds, 

 anthracene for petrogenic compounds) may be sufficient for field reconnaissance mapping 

 to permit characterization of Total PAH concentration. 



Results of sediment column characterizations also show that ambient PAH 

 concentrations at selected locations are near the present detection limit of the instrument. 

 For these locations, approximately 3% of the pixels in the ROI are above internal camera 

 noise and illumination variance. This would suggest that the selected ROI should be kept 

 sufficiently large so as to incorporate an adequate sample size to characterize PAH 

 composition. Increased binning and emission reading times can also increase S:N ratios. 

 In the present data set, the full sediment image (excluding water column) had > 100 pixels 

 for three of four excitation bands (Table 3), which should represent a large enough sample 

 for spectral analysis. The more limited data density for individual ROI (generally <20 

 pixels/ROI; Table 2) would reduce the certainty of spectral characterization. 

 Notwithstanding, smaller ROIs used in the present investigation have permitted the 

 elucidation of more spatially isolated spectra (e.g., "hot spots") which would otherwise be 

 lost in the signal obtained from a full sediment image. 



