wavelengths and, depending on concentration and composition, will emit "red shift" light 

 (i.e., fluoresce) across a broad spectrum of wavelengths. Based on prior investigation and 

 literature values, specific wavelengths were selected to measure the fluorescence response 

 and hence characterize the PAH composition of the sample. Before the red shift light is 

 collected for spectral measurement, a blocking filter is used to block-out all excitation 

 light. The selected emission endpoints include 17 total narrow wavelengths including 15 

 wavelengths between 350-500 nm (at increments of 10 nm) and two additional wavelengths 

 at each of 550 nm and 600 nm bandwidths. In this manner, up to 68 intensity images (17 

 emission x 4 excitation bands) for each image replicate can be obtained. 



Also collected on each sampling day are dark field (camera on, no light) and white 

 field images (camera on, white surface illuminated) for the purpose of correcting internal 

 electrical noise and variation in target illumination, respectively. The resulting data are 

 internally stored then relayed to the surface computer after one or several samples are 

 collected. 



2.4 Data Processing and Analysis 



The primary method of correcting raw image (RI) data for variance in electronic 

 noise and sample illumination is described in Equation 1 : 



CI (i,j) = (RI (i,j) - DF (i,j)) / WF (i,j) (1) 



where CI = corrected image matrix, DF = dark field image matrix, and WF = white field 

 image matrix, and i,j is the row and column location of pixels. In the above corrections, 

 the DF matrix is used to remove electrical noise in the camera system, while the WF 

 matrix permits correction of individual pixels for differences in pixel illumination across 

 the image. Except where noted, the above corrections were applied to all data sets 

 included in this report. 



MATLAB programs (Mathworks, Inc.) were written to visualize the data and 

 perform the above corrections. The program permits further examination of the data using 

 area, intensity, and spectral quality filters for specified Regions of Interest (ROI) within the 

 image. ROIs are selected by the user and depend on the camera penetration depth and 

 other features readily visible in RGB images taken at the same location. For example, 

 ROIs are used to mask out the water column part of each image and to limit the ROI to 

 only those parts of the sediment profile that exhibit fluorescence. 



