The Role of Riverbed Background Reflectance in Long Term Turbidity Monitoring Using Camera Systems

The use of digital cameras in river monitoring activities can increase our knowledge of water quality status, solving the cost and spatial and temporal data resolution limitations of the existing techniques. The challenge of image-based procedures using camera systems is the proper red, green, and blue (RGB) bands signal interpretation and processing. The actual water upwelling light that reaches the camera lens is the sum of various reflectance components of the suspended particles, the riverbed background and the water itself. One component could prevail over the others, depending on the variability of hydrological (water level, flow velocity, etc.) and environmental (suspended solids concentration, floating pollutants, etc.) characteristics of the river. The effect of water level and turbidity concentration on the riverbed component of the total water upwelling light can be substantial, especially for shallow water. As a result, the riverbed reflectance component, if neglected, can significantly affect the evaluation of the water reflectance, and hence, water turbidity.

In our field campaign, a synthetic turbidity event was recreated by adding a natural clay tracer into the river, and we monitored it using a camera system. Two turbidimeters were installed within the river section to validate the results. Moreover, a submerged panel was fixed directly on the riverbed. This choice was prompted by the shallow water conditions during the experiment, where the riverbed reflectance significantly contributed to the total upwelling light captured by the camera, particularly under low turbidity levels. We defined a clear water condition in which the panel was fully visible, where turbidity level was considered equal to zero. As turbidity increased and the panel visibility decreased, we applied an image-based procedure to assess the actual river turbidity level. In addition, we applied a pixel-by-pixel mean of the camera frames every 2 minutes, for minimizing the signal distortions due to the effect of ripples, sun glare and shadows within the analyzed region of interest of the river surface.  These methodological steps allowed us to properly decompose the image into different reflectance components, and to enhance long-term monitoring practices that are subject to a wide range of environmental and hydrological variability.

This study focuses on implementing camera systems in real-world settings, supporting existing river monitoring techniques with early warning networks, and developing innovative solutions for water resource management.

Keywords: camera system, river monitoring, turbidity, image processing, remote sensing, water quality