Assessing Smartphone GNSS Accuracy for Reporting and Localisation of Water-Leakage Incidents: A Citizen Cyber-Science and WebGIS approach
Abstract
Citizens are often the first to observe visible water leakages, but an actionable report requires a location that utility personnel can interpret and verify. This study assessed the positional accuracy and operational suitability of smartphone GNSS for approximate localisation of citizen-reported urban water-leakage incidents. Two KoboToolbox-supported workflows were evaluated against the W.Tank geodetic control point at Ardhi University, Dar es Salaam: a structured KoboCollect Android App assessment involving 22 smartphone models and an Enketo/Web Forms assessment comprising 100 repeated observations from 20 respondent-device units. Horizontal error was computed relative to the control point and interpreted using a fitness-for-purpose approach. The structured workflow produced a mean error of 5.40 m, RMSE of 5.91 m and maximum error of 10.03 m. The browser-based workflow had a median error of 10.00 m, but severe outliers increased the mean to 161.63 m, RMSE to 299.21 m and maximum to 856 m. The results show that smartphone GNSS can provide a useful initial location for citizen reports when capture is controlled, but automatic coordinates should not be treated as exact ground positions. A Citizen Cyber-Science and WebGIS workflow should therefore combine GNSS stabilisation, accuracy thresholds, photographs, contextual map layers and a movable location pointer before submission. The approach supports preliminary incident localisation and field navigation; it is not intended for cadastral surveying, legal boundary determination or engineering setting-out.
