Uncertainty Analysis of Hydrodynamic Modeling of Flooding in the Lower Niger River Under Sea Level Rise Conditions

Abstract

Uncertainty in modeling results can be introduced via different sources including: the input data, the modeling assumptions, simulations based on hypothetical scenarios, etc. In this paper the uncertainty in modeling results of 1D and 1D/2D hydrodynamic Sobek models of flooding in the Niger River are analyzed. The models were set up with discharge data as upstream boundary conditions and tidal water level data as downstream boundary conditions. The models were run for the years 1998, 2005, 2006, and 2007. Data available for 1998, 2006, and 2007 were for flooding, while 2005 data represents normal flow data. The model setup included 48 cross sections located between Lokoja and the two ends of the rivers Forcados and Nun. The boundary conditions were varied downstream at the mouths of rivers Forcados and Nun using sea level rise (SLR) values adopted from the Rahmstorf predicted values; the simulations were projected for the years 2030 and 2050. Five modeling scenarios were set up to simulate the interaction of river flooding with downstream rise in sea levels. The scenarios were: sea level rise with normal year flow from upstream, sea level rise with a flooding year flow from upstream, sea level rise with flash floods from upstream, sea level rise with subsidence and flooding year flow from upstream, and sea level rise with subsidence and flash floods from upstream. The use of predicted SLR values introduces uncertainties in the model outputs. Another source of uncertainty was the value for land subsidence (25 mm/yr) adopted from estimates by local experts (the exact value is not yet known and might vary within the area). Uncertainty analysis of the modeling results were carried out using probability-based sampling methods in order to determine the uncertainties in modeling results for effects of downstream SLR on flooding extent, flooding time, and change in water depth in the Niger delta.