Automated monitoring of water resources for the detection of macrophytes using computer vision techniques and bio-optical models integrating multispectral images from satellite and drone

Abstract

Usually the monitoring of macrophytes is carried out with visual inspection and visits to the places of interest and by spot collections in situ, a factor considered to be quite costly for the entire process. The Remote Sensing techniques integrating with in situ collections offer benefits, based on the wide and constant technological development. Such approaches allow the observation of phenomena from their spectral response using geoprocessing technologies. The technical-scientific challenge is related to the detection of macrophyte flowering in its initial stage in aquatic environments, that is, when macrophytes begin to occupy the water layer at the beginning of their appearance, in order to quantify them by remote sensing data. The optimization of the monitoring of the appearance of plants enables the quick decision-making for managing the growth of macrophytes, representing an important opportunity to offer an innovative solution for different markets. Thus, the objective of this project is to develop an innovative solution based on the technical and scientific verification and validation already carried out during the PIPE stage 1 project. The differentiation of the solution proposed in this project is the integration of sensors (multi and hyperspectral) and remote sense technologies to enable a methodology for automated monitoring of aquatic systems and to allow the analysis of macrophyte dynamics and environmental parameters. This application combines orbital and drone sensors to be applied in different scenarios of water resources, such as hydroelectric reservoirs, public supply reservoirs, basic sanitation, fish farming, among others. The methodology proposed in this project differs in two stages: (1) continuous monitoring of the occurrence of macrophytes using orbital images based on optically active parameters, bio-optical modeling integrating computer vision and artificial intelligence techniques; (2) survey with drone in aquatic systems that present the need for details pointed out by the previous step and collection of multispectral / hyperspectral data to precisely quantify the presence of macrophytes. The potential of this solution is confirmed by overcoming all technical-scientific challenges related to the new methodology during PIPE phase 1, including validation with the market from the PIPE Entrepreneur Training. The result of this project has great potential for innovation, since there is no equivalent method to offer continuous, integrated and automated monitoring of macrophytes and other water parameters in a scalable way, without relying on visits for visual inspection, collection and possible laboratory analysis. (AU)