Monitoramento da Clorofila-a na água por sensoriamento remoto utilizando experimentos em Mesocosmos

Abstract

Eutrophication is one of the main causes of compromising the quality of inland and marine waters. The increase in loads and concentrations of nutrients in the terrestrial environment and, consequently, in aquatic ecosystems, is mainly caused by anthropic actions, such as population growth and agricultural specializations, which are also affected by global climate change. The availability of nutrients, mainly nitrogen and phosphorus, is a limiting factor for the development of phytoplankton, and their excessive growth can result in increased water turbidity and production of harmful toxins to the aquatic ecosystem and human health. Its proliferation is directly related to the concentration of chlorophyll-a, due to its photosynthetic activity. Currently, advances in the area of remote sensing allow carrying out studies of remediation and monitoring of water quality, overcoming some limitations of in situ sampling methods, such as the impossibility of investigating spatial and temporal variations in large bodies of water; high laboratory and logistics costs; and inaccessibility of some locations. In this sense, this work aimed to evaluate remote sensing techniques, from unmanned aerial vehicles (UAV), onboard with RGB sensors, in the quantification of chlorophyll-a in aquatic environments. More specifically, we sought to identify the association of vegetation indices from visible optical spectrum bands with the concentration of chlorophyll-a in phytoplankton, applied in a controlled environment. The experiment consisted of 26 mesocosms, where phytoplankton samples were inserted, simulating small shallow lakes, with gradual additions of nitrogen and phosphorus, until a trophic gradient was obtained. Subsequently, in situ concentrations of chlorophyll-a and aerial images with the UAV were obtained. The images were processed to generate orthorectified mosaics, aiming to calculate eight vegetation indices, by which simple linear regressions were adjusted as a function of chlorophyll-a concentrations. The NGBDI, SI, NGRDI, SCI, VWRI, GLI, EXG and VARI indices obtained R2 = 0.88, 0.85, 0.81, 0.81, 0.78, 0.73 and 0.69, respectively. All resulted in satisfactory estimates, with RMSE < 0.1946 and p-value < 0.001. The assumptions of normality and heteroscedasticity of the model were also met, verified by the Shapiro-Wilk and Breush-Pagan tests, respectively. The results allow us to conclude that the monitoring of water quality, for the evaluated parameters, can be carried out by remotely piloted aircraft, onboard with standard RGB cameras, with faster, simpler and lower cost protocols