The Congo River provides potential for socio-economic growth at the regional scale, but with limited information on the river dynamics it is difficult for basin countries to benefit from this potential, and to invest in the development of water resources. In recent years, the number of hazards related to navigation and flooding has sharply increased, resulting in high loss of human lives as well as economic losses. Associated problems of river management in the Congo also include inefficiency in hydropower production, an increase in rate of river sedimentation and land use changes. Accurate information is needed to support adequate management strategies such as prediction of navigation water levels and sediment movement, and assessment of environmental impacts and engineering implications of water resources infrastructure. Modelling approaches and space observations have been used to understand the Congo River dynamics, but their effective application has proved difficult due to a lack of ground-based observational data for validation. Recent developments in data capture with acoustic Doppler technologies have considerably improved measurements of river dynamics. As well measuring river discharge, they also allow the analysis of the multiple hydrodynamic features occurring in fluvial systems. This paper presents the results of field measurement campaigns carried out in the middle reach of the Congo River and the Kasai tributary using state of the art measurement technology (ADCP, Sonar, GNSS) for investigation of large rivers. The measurements relate to river flow at multiple transects, river bathymetry, static and continuous water surface elevation, and targeted sediment sampling along the river. The paper provides a descriptive summary of the measurement results, a discussion on the application and performance of the equipment used in the Congo River, and lessons for future use of this equipment for measurements of large rivers in a data scarce environment such as the Congo Basin.

We propose a machine learning-based approach to estimate the flood defense standard (FDS) for ungauged sites. We adopted random forest regression (RFR) to characterize the relationship between the observed FDS and ten explanatory factors contained in publicly available datasets. We compared RFR with multiple linear regression (MLR) and demonstrated the proposed approach in the conterminous United States (CONUS) and England, respectively. The results showed the following: (1) RFR performed better than MLR, with a Nash–Sutcliffe efficiency (NSE) of 0.82 in the CONUS and 0.73 in England. A negative NSE when using MLR indicated that the relationship between the FDS and each explanatory factor did not obey an explicit linear function. (2) River flood factors had higher importance than physical and socio-economic factors in the FDS estimation. The proposed approach achieved the highest performance using all factors for prediction and could not provide satisfactory predictions (NSE < 0.6) using physical or socio-economic factors individually. (3) We estimated the FDS for all ungauged sites in the CONUS and England. Approximately 80% and 29% of sites were identified as high or highest standard (> 100-year return period) in the CONUS and England, respectively. (4) We incorporated the estimated FDS in large-scale flood modeling and compared the model results with official flood hazard maps in three case studies. We identified obvious overestimations in protected areas when flood defenses were not taken into account; and flood defenses were successfully represented using the proposed approach.