Hydrosalinity modelling of the Berg River using ACRUSalinity
Thesis (MScEng (Civil Engineering))--Stellenbosch University, 2008.
In recent years, concern about the water quality in the Berg River received a fair degree of attention, particularly with the imminent construction of the Berg Water Project (BWP). Particular concerns have been expressed about the water quality with respect to total dissolved salts (TDS) at Misverstand Dam. In previous studies (Fourie and Görgens, 1977) it was identified that the saline water was mostly generated in the lower portion of the Berg River Catchment (Matjies, Moorreesburg and Sandspruit Rivers) and that the abstraction of acceptable quality water higher up in the Berg River could possibly result in salinity problems at Misverstand Dam. Contrary to expectation, these studies also showed that for the most saline catchments, a winter peak in TDS concentrations also existed. To help address these concerns, a Water Research Commission (WRC) project was initiated in 2003 in which the newly-developed salinity module of the daily Agricultural Catchment Research Unit (ACRU) agrohydrological model, known as ACRUSalinity, would be configured for the Berg River Catchment. This model had previously been configured and calibrated for the Mkhomazi Catchment (Teweldebrhan, 2003) which exhibited relatively low streamflow TDS concentrations (100 mg/l) and it was deemed necessary to ascertain whether comparable TDS values could be simulated in the Berg River Catchment, where TDS concentration could rise to well above 1 000 mg/l in certain tributaries. In this project, ACRUSalinity was configured for the Berg River Catchment on a distributed basis, aiming to capture the spatial distribution of rainfall and geophysical characteristics which inherently exist in a catchment as expansive as the Berg. Initial application of the "Beta version" of ACRUSalinity to the Berg River Catchment revealed that it failed to produce simulated TDS values which were representative of the observed data. It became evident that the model required both additional salinity-related functions and modifications of existing functions. After the implementation of these algorithm changes the correspondence of simulated and observed TDS concentrations improved markedly. Verification of the ACRUSalinity simulated flows and calibration of the salinity-related parameters was based on the values of predefined objective functions. Reasonably representative flows could be obtained provided that the catchment discretisation and driver rainfall selection process were adequate. Salinity related parameters were determined purely on an iterative basis, although a priori estimation of these parameters was possible. Preliminary interdependency tests of these parameters revealed that the final calibrated set of salinity-related parameters was probably not unique and that some a priori decision making would be required when selecting the most realistic set of parameters. Quantification of the potential effect of the Berg River Dam on the TDS concentrations at Misverstand Dam was achieved as follows: the ACRUSalinity model was verified for flow and calibrated for TDS at available and reliable flow gauging stations. This was then followed by a long-term simulation run which yielded daily TDS time series for comparison, on an exceedance basis, with the observed record. Since the concern about the possible deterioration of water quality at Misverstand Dam was only a winter concern (May to September), comparisons were only drawn over this period. The flow-routing option in ACRUSalinity was not activated and a 1:1 daily comparison of flows and TDS concentrations, based on values of the objective function, was thus not possible. Results from this study showed that even with a daily model, the exceedance percentages of the TDS concentrations after the construction of the Berg River Dam were comparable with the exceedance percentages obtained from the original monthly modelling study (DWAF, 1993). In this study, however, it was possible to capture the increasing TDS concentration which was evident over winter months in the observed data record for the Matjies River and Sandspruit River catchments. The testing of the model’s effectiveness in the evaluation of engineering options was accomplished as follows: several options for ameliorating the possible deterioration of water quality at Misverstand Dam were defined, based on its practicality and cost of implementation. For example, the Withoogte water treatment works abstracts water from Misverstand Dam for supply to the West Coast region when water quality is acceptable (i.e. a TDS lower than 450 mg/l). It was proposed that to minimise the effect of periods when no abstraction from Misverstand could occur due to unacceptable water quality, a second reservoir at the treatment works should be lined and used to provide bridging storage for water from Misverstand Dam when the water quality was acceptable. The calibrated ACRUSalinity model was then modified to reflect the physical attributes of this engineering scenario of interest to produce sets of flow and TDS time series which could be further analysed to determine assurance of supply, in terms of predetermined TDS concentration thresholds in Misverstand Dam. Using this particular engineering option, the analysis revealed that a 300 mg/l TDS upper-limit at Misverstand was too stringent and that 450 mg/l was probably more realistic.