Efeitos de diferentes componentes opticamente ativos na reflectância da água para a estimativa remota das concentrações de clorofila-a fitoplanctônica

Abstract

The importance of monitoring phytoplankton communities and variables related to their constituents such as chlorophyll-a is undeniable due to its relation to the quality of the water. Even so, further improvements are necessary on the representativeness of the samples, aiming spatial and temporal pattern recognition of the given variables. Data and techniques of remote sensing have been mentioned as a viable alternative to supply this demand. In contrast, the remote estimation of phytoplankton chlorophyll-a concentration may still be seen as a challenge, mainly in bodies of water of case 2 classification, mostly due to the effects of different optically-active components on the sensing data, including the distinct phytoplankton organisms. For this purpose, algorithms were and currently are being developed. Among those, there are empirical models of a unique spectral band obtained in analysis of simple linear regression. However, this technique is only applicable when a single independent variable affects the dependent variable, which does not occur in bodies of water case 2 classification. Within this context, the general objective of the present work was to verify the adequacy of empirical models developed by simple linear regression to estimate phytoplankton chlorophyll-a from reflectance data in a unique spectral band and to demonstrate that phytoplankton organisms can be considered distinct optically-active components in the water. In order to meet this objective, phytoplankton organisms classified as Cyanophyceae, Euglenophyceae and Chlorophyceae were isolated and cultivated; samples of inorganic sediments were processed and geometries of illumination and acquisition of radiometry data were determined and controlled to perform this experiment in laboratory level, which consisted basically of water reflectance measurements in a tank, in distinct treatments related to the optically-active components present on it. The correlation analysis, simple linear regression and mean squared error, showed that the empirical models of a unique spectral band, developed by simple linear regression, may miscount the concentration of chlorophyll-a even when the water contains only phytoplankton, as the phytoplankton organisms cultivated have promoted distinct and reverse reflectance variation in relation to chlorophyll-a concentration variation and they may be considered distinct optically-active components in the water. In the presence of inorganic suspended solids the estimation errors may be even larger, as additive effects of this component on the reflectance of water variations are not subtracted by the model, but rather considered as being caused by chlorophyll-a. Therefore, the empirical models developed by simple linear regression from reflectance data of a single spectral band are inappropriate for estimating chlorophyll-a phytoplankton in bodies of water, as there is no single wavelength in the visible and infrared regions next to the electromagnetic spectrum that is affected only by the independent variable chlorophyll-a, even when there are only phytoplankton in the water.