Collection and extraction of water level information from a digital river camera image dataset

Area of Science:

• Hydrology
• Remote Sensing
• Environmental Monitoring

Background:

• Accurate water level data is crucial for flood management and forecasting.
• Existing monitoring methods may have limitations in spatial coverage or temporal resolution.
• Image-based data extraction offers a potential complementary approach to traditional river gauging.

Purpose of the Study:

• To present a novel dataset of water levels derived from river camera imagery.
• To detail the methodology for data acquisition and extraction with high spatial accuracy.
• To provide a verified dataset for evaluating and developing automated flood monitoring techniques.

Main Methods:

• Utilized images from four Farson Digital Ltd river cameras during a 2012 Tewkesbury flood event.
• Extracted water level data by referencing measured points within the camera's field-of-view.
• Employed Leica GNSS and Total Station instruments for precise spatial referencing (1 cm accuracy).
• Validated the extracted water level data against established river gauge records.

Main Results:

• A new, short-duration water level dataset for a significant UK flood event.
• The dataset captures the complete flood hydrograph: rising limb, peak discharge, and falling limb.
• High spatial accuracy of the extracted water level measurements was achieved.

Conclusions:

• The presented dataset is valuable for verifying automated water level extraction algorithms.
• This data can aid in the development of advanced automatic flood alert systems.
• The dataset offers potential for data assimilation in numerical models to enhance inundation forecasts.