Under unprecedented pressure from urbanization and climate change, an ever increasing number of streams worldwide fails to meet good ecological status, thus threatening water quality and ecology, and severely impacting our territories. In this vein, RiverWatch aims to develop a new disruptive monitoring infrastructure for river systems focused on the transport of buoyant plastics, woody material, and floating pollution. The infrastructure will build on current knowledge in image-based hydrological monitoring to explore novel advancements in unsupervised computer vision techniques for environmental analyses. RiverWatch will exploit camera systems on fixed stations as well as volunteer smartphones to build a dense network of monitoring stations potentially along any river system in the world. This may help to overcome the current limitations in the management and maintenance of high cost installations and at the same time allow us to expand our monitoring capabilities. Towards establishing a robust infrastructure, RiverWatch will focus on the Sarno River to develop a dense monitoring system based on cutting-edge unsupervised computer vision.
Notably, the Sarno River is the most polluted river in Europe and a challenging and socially disadvantaged environment to establish monitoring networks. A custom-built mobile app as well as advanced image-based algorithms will be developed to process footage captured by citizens and fixed cameras and collected at a remote server. Image-based algorithms will enable analysis of the river flow along with the estimation of surface pollutants discharge and their characterization. Such data will be published in close to real-time on a web-Gis online platform featuring a storymap and a public database. High-frequency data at several locations in the drainage network will facilitate implementation of simple modeling tools to describe and forecast pollutant transport in the Sarno watershed. RiverWatch objectives will be achieved through the harmonized effort of a multidisciplinary research team with a past experience of collaboration in environmental monitoring through computer vision and machine learning as well as citizen science. The transformative potential of the project is expected to enrich the monitoring tools currently available to National Hydrometric Services and Environmental Agencies worldwide by providing an easy-to-use and inexpensive approach to quantitatively assess river pollution. Finally, RiverWatch will stimulate new research on the pollution dynamics in river watersheds and on the interactions of pollutant transport with climate and anthropic agents.
Prof.ssa Chiara Biscarini
Work packages The RU at UniTus will coordinate the project, design and develop the citizen science initiatives, and collaborate on algorithm development as well as closely work with the other RUs throughout the entire duration of the project. UniBo will develop the mobile app, the imageprocessing algorithms, and coordinate the remote server for image processing. UniNa will coordinate implementation of the fixed camera prototypes, execute validation experiments, and collaborate on algorithm development. UniStraPg will collaborate on the design of the citizen science initiatives, and help with the development of the web-Gis platform. The project will be organized in the following work packages (WPs):
1. Citizen Science initiatives (UniTus & UniNa & UniStraPg): This WP entails the design of the initiatives aimed at involving volunteer citizens: selection of the possible pools of volunteers, preparation of explanatory material and presentation of the RiverWatch mobile app. Tasks include identification of key actors as potential citizen scientists, involving schools and associations, preparing material and organizing meetings and presentations.
2. Fixed camera prototypes installations (UniNa & UniTus): This WP entails the design of lowcost camera prototypes to be permanently installed along key sites of the Sarno drainage networks. Tasks include site selection, prototype design and assembly, coordination with local authorities for installation 3. Mobile app and image-based algorithms development (UniBo & UniTus & UniNa): This WP entails development of the mobile app as well as image-processing algorithms and protocols for extracting hydraulic and pollution information from images. Tasks include image selection, filtering, eventual enhancement, code development and validation.
4. Data validation (UniNa & UniTus): This WP entails development of strategies to validate image-based estimations with data available from fixed camera stations and through dedicated field campaigns at key sites in the Sarno drainage networks. Tasks include collecting data with diverse measurement systems, aggregating them at diverse spatial and temporal resolutions, selecting image training and validation sets, designing and developing codes.
5. Management of the remote server (UniBo & UniTus): This WP entails devising a strategy to manage volunteer and fixed camera videos in real time, applying image processing algorithms and uploading results to the web-Gis platform. Tasks include collecting data, ensuring video compatibility, extracting metadata information (from the phone GPS and accelerometer), applying and developing codes.
6. Modeling pollutant transport (UniNa & UniTus): This WP entails developing a simple modeling framework to describe and forecast pollutant transport in the Sarno drainage network. Tasks include testing and validating algorithms and selecting and aggregating data.
7. Development of the web-Gis platform (UniTus & UniStraPg): This WP entails designing and developing the web platform for showcasing and downloading processed (and selected raw) data. Tasks include developing codes, devising a strategy to communicate with the remote server, and regularly updating the website.
8. Dissemination and communication of results (UniTus, UniBo, UniNa, & UniStraPg): this WP entails paper preparation, data publication in open repositories, participation and organization of conferences, sessions, and special issues, and development of educational activities for students from the middle-school to graduate level.
9. Project management (UniTus): this WP entails regular scheduling, clear task distribution and assignment of responsibilities, and transparent treatment of delays and difficulties. Regular reporting and close communication will enable the PI to closely monitor the quality of the project.