A LISFLOOD-FP hydraulic model of the middle reach of the Congo

Abstract

In this paper we attempt to produce a first hydrodynamic model of the middle reach of the Congo river system in order to understand what controls this river’s unique bimodal flood pulse. The model covers the area between Kisangani and Kinshasa on the main stem and includes the major tributaries and the Cuvette Centrale wetland, one of the world’s largest and most understudied lowland regions. A mixture of in-situ discharges and modelled discharge from a basin-wide catchment hydrology model were used to force a four-kilometre resolution hydrodynamic simulation developed using the LISFLOOD-FP model. River channels are represented as sub-grid scale features and their width is therefore decoupled from that of the over-lying floodplain grid. Unknown channel friction and bathymetry parameters were calibrated using ERS-2 and Envisat satellite altimetry measurements of channel water level. The calibrated model simulated channel water surface elevations across the domain with a bias and root mean square error of 0.185 and 0.842 m respectively. The value for root mean squared error is close to that obtained for comparisons of ERS-2 and Envisat satellite altimetry to in-situ water elevation data in similar basins (0.79 m and 0.47 m respectively). The model results imply that the bimodal annual pattern of Congo river discharge is predominantly a hydrological rather than hydraulically-controlled feature, with the channel-floodplain interactions and river constrictions having only a modest impact on the flood wave propagation. Nevertheless, and counter to current understanding, we find that interactions between channels and floodplains do however occur extensively, with over 2100 km of the 13,000 km of channel network in the model identified as zones where water is actively exchanged between channels and floodplains. Whilst the water volume that is exchanged with the floodplain is substantially less than for other large rivers, our results imply that channel-floodplain interactions are a significant feature of Congo flood wave propagation. Overall the model provides insights into the hydraulics of this understudied system that can next be tested both in the field and through more detailed modelling studies.