Hydrogeological and three-dimensional numerical groundwater flow modelling of the Lake Sibayi catchment, Northern KwaZulu-Natal, South Africa.
Weitz, Jan Christian.
MetadataShow full item record
Lake Sibayi, a topographically closed fresh water lake in northern KwaZulu-Natal, South Africa, and the coastal aquifers surrounding the lake, are important water resources for the local community and the surrounding ecosystem. A significant decline in lake levels has been experienced over the last decade, dropping from approximately 20 m above mean sea level (amsl) in early 2000 to below 16 m amsl at present. It is believed that this decrease could be attributed to an increase in water abstraction from the lake and surrounding groundwater, the rapidly increasing commercial plantations within the catchment and recent droughts. The effective management of this hydrological system needs a thorough understanding of the interaction of the lake with the surrounding aquifer. In recent years, hydrogeological and numerical groundwater flow modelling have become standard tools with which these interactions are studied. This thesis describes the process of conceptual model design through to the development and calibration of steady-state and transient numerical groundwater flow model for the lake Sibayi system. Through a series of field campaigns, on site measurements of depth to groundwater with surface and groundwater sampling, for hydrochemical and environmental isotope analysis, were undertaken. Hydrochemical parameters and environmental isotopes for the various water sources within the Lake Sibayi hydrological system were determined following standard procedures to study the relationship between these resources. A slight distinction between shallow and deep aquifers appears to be present, where the shallow groundwaters are dominated mainly by a Na-Cl hydrochemical facies, while the deeper boreholes are dominated mainly by a Na-Ca-HCO3-Cl hydrochemical facies. Shallow groundwater samples have relatively low EC values averaging 278 mS/m, while the deeper wells had average EC concentrations of 409 mS/m. Groundwater samples collected along the dune cordon, show a similar hydrochemical and environmental isotope composition as that of the lake. Multivariate statistical analyses including principal component factor analysis and hierarchical cluster analysis (HCA) were undertaken on the hydrochemical data. The HCA grouped the water samples into two clusters, which represented surface and groundwaters. Each of these two clusters were in turn divided into two sub-clusters, representing the shallow and deep aquifers, and stream and lake samples, respectively. As part of the conceptual modelling, the long-term water balance of the lake has been quantified by defining the various inflow and outflow components of the lake. All hydrological information including precipitation, evaporation, surface water runoff, abstraction, as well as geological, hydraulic, hydrogeochemical and environmental isotope information were used to conceptualise the hydrological system of the Lake Sibayi catchment. Local geologic, groundwater head distribution, lake level, hydrochemical and environmental isotope data were used to constrain the link between groundwater and the lake. In the western section of the catchment, groundwater flows to the lake where groundwater head is above lake stage, whereas along the eastern section, the presence of mixing between lake and groundwater hydrochemical and isotopic compositions indicate that the lake recharges the aquifer. Stable isotope signals further revealed the movement of lake water through and below the coastal dune cordon before eventually discharging into the Indian Ocean. Groundwater recharge to the catchment was estimated using the chloride mass balance (CMB) method and the results compared with estimates based on published maps. The CMB recharge estimate resulted in 126 mm/a (12 % MAP) against 95 mm/a (10% MAP), estimated using published maps. The total evaporation and evapotranspiration from the lake and its catchment were estimated at 1 495 mm/a and 1 090 mm/a, respectively. Estimated surface water runoff from the catchment to the lake is about 1% of MAP. Calculated lake water outflow to the sea through the dune cordon opposite the lake, along a 12 km seepage face, is 2.3 x 107 m3/a. The total amount of water abstracted from both surface and groundwater resources within the catchment is about 4.5 x 106 m3/a. The water balance of Lake Sibayi shows that lake levels fluctuate in response to varying amounts of groundwater and surface water inflow into the lake, seepage loss through the coastal dune, abstraction, and evaporation from the lake. Based on the conceptual hydrogeological model, a steady-state and transient numerical groundwater flow model, were developed for the Lake Sibayi system using two independent approaches, namely, the High-K method and Lake Package. Groundwater Modelling Systems (GMS), which runs on the modular finite difference code, MODFLOW 2005 with its several packages were used to characterise the three dimensional flow conditions around the lake. Two layer models were used to simulate the lake stage and aquifer conditions over a forty three year period from January 1970 to September 2014. The simulation period was broken down into 536 monthly stress periods with calibrated parameter values for each of the boundary conditions over the simulation period. The calibrated steady-state model simulation results for the two methods were comparable. While, transient model calibration results show that the Lake Package was more suitable in simulating lake level fluctuations with low calibration errors. The calibrated transient groundwater flow models were further used to evaluate the hydrological response of the lake and the groundwater system to various stress scenarios, including changes in evaporation, precipitation and land use. Once again, the High-K method was very sensitive to changes in model input, simulating rapid changes to the system, while Lake Package simulations results were in line with known changes in the system. Therefore, the High-K technique is most suitable for simple applications, while complex lake-aquifer interactions are better simulated using the Lake Package.