九州大学大学院農学研究院環境農学部門生産環境科学講座水環境学研究室
Laboratory of Water Environment Engineering, Division of Bioproduction Environmental Sciences, Department of Agro-environmental Sciences, Faculty of Agriculture, Kyushu University
九州大学大学院農学研究院環境農学部門生産環境科学講座水環境学研究室
Laboratory of Water Environment Engineering, Division of Bioproduction Environmental Sciences, Department of Agro-environmental Sciences, Faculty of Agriculture, Kyushu University
九州大学大学院農学研究院環境農学部門生産環境科学講座水環境学研究室
Laboratory of Water Environment Engineering, Division of Bioproduction Environmental Sciences, Department of Agro-environmental Sciences, Faculty of Agriculture, Kyushu University
A fluorescence excitation spectrum was employed to measure in vivo and in situ concentrations of class-differentiated Chlorophyll-a (Chl.a) in a eutrophic water body. We estimated total, Chlorophyceae, and Cyanobacterial Chl-a concentrations, using a three-layered feedforward artificial neural network instead of generating an algal standard solution calibration curve. Here, the fluorescence intensities for nine excitation wavelengths were used as the input variables. The intensities were measured in an actual reservoir, using a multi- excitation wavelength fluorometer. The training datasets for network learning were acquired from two eutrophic agricultural reservoirs, and the estimation model was verified by cross-validation. The network estimated Chl.a in the dominant species and resulted in accurate identification of seasonally prevalent phytoplankton and total Chl-a in the reservoir. Moreover, the network model, although was learnt by single eutrophic water body and single spot datasets, was applicable to other areas and spots. This application enabled a) a continuous observation at a fixed point and b) scheduled observation at multiple points for class-differentiated Chl.a, and provided important information for understanding cyanobacteria bloom prevalence.
Mendeley出力
