AI-assisted river discharge measurement through citizen science and mobile technology

Vigneswaran, Kayathri; Retief, H.; Clifford-Holmes, J.; Garcia Andarcia, Mariangel; Tennakoon, Hansaka

AI-assisted river discharge measurement through citizen science and mobile technology

Autores

Vigneswaran, Kayathri

Retief, H.

Clifford-Holmes, J.

Garcia Andarcia, Mariangel

Tennakoon, Hansaka

Editor

International Water Management Institute
CGIAR Accelerator for Digital Transformation

Descripción

This report presents a novel protocol for measuring river discharge by integrating artificial intelligence with citizen science in the Limpopo River Basin. Developed under the Enabel/Wehubit “Citizen Science for Water Management” project and aligned with the LIMCOM–UNDP/GEF programme, the approach addresses a core constraint in data-scarce basins: limited observational infrastructure restricts the data needed to power digital decision-support systems. A WhatsApp-based platform enables community members to submit gauge-plate photographs and supporting metadata (station selection, location sharing, and validation) without requiring a dedicated mobile application. The backend includes a Vision API and a custom WhatsApp Bot Service deployed on AWS ECS to manage guided submissions and secure, station-linked user permissions consistent with LIMCOM governance. Gauge readings are derived via a two-step AI method that first detects waterline and scale, then converts these to water level, improving robustness under field conditions. Testing achieved strong accuracy (R² = 0.84; mean error 5.43 cm on high-quality images). Validated water levels are translated to discharge using station-specific rating curves accessed through the FlowTracker API. The protocol demonstrates that hybrid AI and embedded quality assurance can augment hydrological monitoring in transboundary basins while maintaining scientific rigour.