Passive optical remote sensing of Congo River bathymetry using Landsat

Abstract:

While there have been notable advances in deriving river characteristics such as width, using satellite remote sensing datasets, deriving river bathymetry remains a significant challenge. Bathymetry is fundamental to hydrodynamic modelling of river systems and being able to estimate this parameter remotely would be of great benefit, especially when attempting to model hard to access areas where the collection of field data is difficult. One such region is the Congo Basin, where due to past political instability and large scale there are few studies that characterise river bathymetry. In this study we test whether it is possible to use passive optical remote sensing to estimate the depth of the Congo River using Landsat 8 imagery in the region around Malebo Pool, located just upstream of the Kinshasa gauging station.

Methods of estimating bathymetry using remotely sensed datasets have been used extensively for coastal regions and now more recently have been demonstrated as feasible for optically shallow rivers. Previous river bathymetry studies have focused on shallow rivers and have generally used aerial imagery with a finer spatial resolution than Landsat. While the Congo River has relatively low suspended sediment concentration values the application of passive bathymetry estimation to a river of this scale has not been attempted before.

Three different analysis methods are tested in this study: 1) a single band algorithm; 2) a log ratio method; and 3) a linear transform method. All three methods require depth data for calibration and in this study area bathymetry measurements are available for three cross-sections resulting in approximately 300 in-situ measurements of depth, which are used in the calibration and validation.

The performance of each method is assessed, allowing the feasibility of passive depth measurement in the Congo River to be determined. Considering the scarcity of in-situ bathymetry measurements on the Congo River, even an approximate estimate of depths from these methods will be of considerable value in its hydraulic characterisation.