PLANKTOTHRIX RUBESCENS IN FRESHWATER RESERVOIRS: THE SENTINEL-2 POTENTIALITY FOR MAPPING PHYCOCYANIN CONCENTRATION
- Authors: Maltese, A; Capodici, F; Ciraolo, G; Corbari, C; Granata, A; La Loggia, G
- Publication year: 2012
- Type: Proceedings
- OA Link: http://hdl.handle.net/10447/66426
Abstract
In December 2006 blooms of Planktothrix rubescens were found in the Prizzi reservoir in Sicily. P. rubescens is sadly famous for producing microcystins (MC), which are harmful hepatotoxins. Recently (2006) P. rubescens has been found in the Pozzillo, Nicoletti, Ancipa, Prizzi and Garcia reservoirs. This paper compares the optical properties of the water of an infested reservoir and those of a clear water reservoir. Furthermore it evaluates an empirical method based on bands product to evaluate the phycocyanin cell density from MODIS, Landsat ETM+ and Sentinel-2 images. Spectroradiometric field campaigns were carried out in February/March 2008 to quantify the spectral transparencies of two water bodies through the calculation of the diffuse attenuation coefficient, by measuring underwater downwelling irradiance at different depths as well as water spectral reflectance. The possibility of detecting P. rubescens using its spectral characteristics through diachronic remote sensing techniques was investigated. During periods of low shortwave irradiance such as winter, when light weakly penetrates water columns and the water cools, P. rubescens filaments float up to the surface, forming redcolored blooms. Analyses were carried out on two reservoirs: Lake Prizzi and Lake Trinità , with and without the P. rubescens algal bloom respectively, during the acquisition of field data. Water samples were collected for both phytoplankton counting and biovolume estimation and for the analysis of MC. Meteorological measurements highlighted the low temperatures (< 2° C) for this region, which occurred between two frosts in December 2007 and February 2008, with a contemporaneous reduction in the incident total solar radiation that probably triggered P. rubescens to float up to the water surface in the infested lake. Underwater spectral irradiances and in-air upwelling spectral radiances were acquired in February/March 2007 in order to characterize the extinction optical properties of the lake infested by the red algae and the uninfested one. The results showed that optical properties are significantly altered when P. rubescens infests lake water. Comparison between the reflectance spectrum of water dominated by P. rubescens and the spectrum of the reference lake water showed higher reflectance in the former case. However, even if Landsat ETM+ images have an appropriate spatial resolution for monitoring Sicilian reservoirs, that are characterized by an average extension of few kilometers; these images cannot be used to distinguish the contribute of chlorophyll from that of cyanobacteria and their pigment (the phycocianin - PC) at early stage formation of cyanobacterial blooms, and their concentration cannot be assessed accurately. When cyanobacteria dominate, the PC absorption is identified by a rising of spectral reflectance in red and near infrared spectral ranges. MODIS images have suitable spectral bands at a spatial resolution of 1 km that, however, is not suitable for lakes monitoring. A time series of MODIS images (spatial resolution of 250 m - red and near infrared bands) have been processed and compared to in situ population density and total concentration of MC. The retrieved empirical equations tend to overestimate the actual values for low (or absent) surface filaments. Sentinel-2 images promise to give an operational solution. Images have a temporal resolution of ≈ 5 days (over our region) that could suitably describe the temporal dynamic of the phenomenon. This experimental evidence could be used to set up an early warning system for P. rubescens blooms based on Sentinel-2 images. This would be based on the reflectance spectral variation though a diachronic analysis of multispectral images and should take into account other meteorological variables such as incident solar radiation.