Tracking and Mapping of the Douro River Plume with an AUV

Abstract

In this thesis, we present an algorithm for an AUV (Autonomous Underwater Vehicle) to autonomously track and map the front of the Douro river plume (Porto, Portugal). The plume's front is characterized by a steep gradient of the salinity field and is trackable in real-time by measuring salinity with a CTD (Conductivity Temperature Depth) sensor. The river discharge flows into the ocean carrying sediments, nutrients, and even pollutants that will interact and disperse in the ocean water (with higher salinity and density). These elements propagate with the plume, a less dense water layer flowing on top of the ocean water. For this reason, it is important to study and map the dynamics of the plume, which will help predict its influence on the coastal areas. The study of river plumes has been typically done with the help of research vessels or remote sensing products, provided by drone or satellite imagery. These methods are logistically difficult,expensive, lack the required spacial and temporal resolutions and are not synoptic. The AUV performs an adaptive zigzag-like approach to follow and autonomously track the front, while performing a typical yo-yo trajectory in the vertical plane to sample the plume in 3D. This approach is inspired by previous work done by the Laboratório de Sistemas e Tecnologia Subaquática (LSTS) on the Douro river. Background information on plume models and the Douro river plume are presented in preparation for discussing the state of the art on adaptive sampling of different marine features using autonomous vehicles. The formulation of the problem, organized into several sub-problems, is presented next along with a discussion of the proposed approach. The approach for the tracking algorithm is documented as well as the development of a simulation environment to validate the results using a numerical model that accurately simulates the plume. Following the implementation of the algorithm in the LSTS toolchain, this software will support the deployment of the vehicle. Finally, we discuss the results and future prospects of the system.