Research of dissolved organic matter fluorescence band variability in the South Atlantic using artificial neural networks

Sergey A. Burikov, Tatiana A. Dolenko, Victor V. Fadeev,
Physical Department, Moscow State University, Moscow, 119899, Russia
burikov@lid.phys.msu.su, tdolenko@radio-msu.su, fadeev@lid.phys.msu.su

Igor I. Vlasov
Institute of General Physics of Russian Academy of Sciences, Moscow, Russia
vlasov@kapella.gpi.ru

ABSTRACT

The presentation reports on the results of studying features of the fluorescence band of dissolved organic matter (DOM) in water samples taken along the route of the 19^th sailing of the research vessel "Academic Sergey Vavilov" in the period from November 1 to November 20, 2004 (sections in the area from 30° south to 55° south latitude). The interest in the variability of DOM fluorescence features is connected with an attempt to use DOM fluorescence as a natural indicator of water type.

Fluorescence spectra of water samples were measured on a laser spectrometer: excited by the radiation of a nitrogen laser (λ = 337 nm, peak pulse power about 1 kW, pulse repetition rate 100 Hz), and registered by a multichannel spectra analyzer.

After standard processing (smoothing and normalizing of fluorescence to unit by intensity), the spectra were classified with the help of an artificial neural network. It was revealed that the difference in the values of DOM fluorescence intensities normalized to Raman intensity were consistent with change in the features of the bandshape, registered by classification algorithms. An attempt has been made to connect the results of splitting into classes with hydrological and morphological peculiarities of the region of sampling.

Last Update: 2005-03-16