An integrated framework for hydrology-hydrodynamic modelling of the Barotse floodplains, upper Zambezi river basin.

Loading...
Thumbnail Image
Date
2023
Authors
Chomba, Innocent Chomba
Journal Title
Journal ISSN
Volume Title
Publisher
The University of Zambia
Abstract
In recent years the demand for improved assessments tools to help in understanding the wetland hydrological and hydrodynamic processes for the Barotse floodplains is ever increasing especially with the advent of climate change/variability. Despite several studies in the Upper Zambezi Basin, there was an absence of an integrated online (internal) framework for coupling hydrology and hydrodynamic model to be used to improve modelling of the Barotse floodplains. The objectives of this study were: (i) To investigate the mechanisms of groundwater-surface water interaction in the Barotse floodplains, (ii) To develop a rainfall runoff hydrologic model for Upper Zambezi River basin focusing on modelling Barotse floodplain, (iii) To develop an integrated hydrologic-hydrodynamic online (internal) coupling modelling framework for Barotse floodplains and (iv) To test and evaluate the performance of an online one-way Barotse Coupled Hydrological-Hydrodynamic Model developed based on the framework in objective iii. The study used field data sets, models, remote sensing data, and secondary data to address the objectives at hand. Based on the study objectives, the study has shown that the interaction between groundwater and surface water does occur in the Barotse Floodplain as evidenced by hydro-chemical and isotopic results. Such kind of groundwatersurface water interaction in floodplain, among other factors play a role in the hydrological and hydrodynamic modelling of the Barotse floodplains system. This postulation is partly supported by the results of simulated flows generated by wflow hydrological model, for the upstream gauge stations the results were closely matched the observed flow as indicated by the evaluation statistics; Chavuma, nse =0.738; kge = 0.738; pbias = 2.561 and rsr = 0.511; Watopa, nse=0.684; kge = 0.816; pbias = 10.577 and rsr = 0.557; Lukulu, nse=0.736; kge = 0.795; pbias = 10.437 and rse = 0.509. However, despite that wflow model was able to simulate the upstream hydrology very well, wflow model statistical objective function results of the downstream Barotse Floodplain gauge station (at Senanga) were not as good as the upstream results as indicated by evaluation statistics: nse = 0.132; kge = 0.509; pbias = 37.740 and rse = 0.923. The observed inaccuracy in goodness of fit between observed and simulated at Senanga Gauge Station maybe attributed to model forcing data as well as to the fact the representation of both floodplain channels hydrodynamics and hydrological processes are necessary to correctly capture floodplain dynamics for groundwater dependent systems. This aspect also suggested that standalone hydrological models are not very suitable in modelling the flows of the of groundwater-surface water dependant tropical floodplains. To this effect, one-way online (internal) hydrologic-hydrodynamic coupling framework for Barotse Floodplain has been developed in this study. The coupled model output relatively performed better than a standalone hydrological results. The significant improvements in coupled model results were observed at the downstream gauge station at Senanga with noticeable improvement in nse which improved from 0.132 to 0.535, kge improved from 0.509 to 0.699 and pbias from 37.724 to 21.495 indicating reduced model over estimation. The water levels results output from the online Barotse hydrologic-hydrodynamic model coupling also relatively matched the observed water levels as demonstrated by the statistical goodness of fit objective functions results: Mongu, kge 0.792, rse 0.762, nse 0.55; Senanga; kge 0.6707, rsr 0.827, nse 0.503 and Lukulu, kge 0.630, RSR 0.644, nse 0.301. In addition to water levels comparisons, the simulated results of coupled mode inundation area were compared with MODIS MOD09A1 imagery. The statistical objective functions were nse 0.637, kge 0.731, rsr 0.502 and pbias 25.234. In general terms, the statistical results of the simulated and observed data sets (flow, water level and inundated area) of the on-line hydrological-hydrodynamic coupled model falls within what is deemed to be a good performing model indicating that the integrated framework for online coupling of hydrology-hydrodynamic models for the Barotse floodplain was a successful in coupling the two models used in this study.
Description
Thesis of Doctor of Philosophy in Integrated Water Resources Management.
Keywords
Citation
Collections