Modelling the impacts of changes in agricultural management practices on water resources with declining hydrometeorological data in the Uthukela Catchment.

Abstract

In order to meet the country’s growing demand for food, and to transform the economy of rural communities, the South African Government aims to develop the agricultural sector in the uThukela Catchment, KwaZulu-Natal Province. Intensification of agriculture will depend on the availability of water resources, with subsequent impacts on the quality and quantity of water resources. Therefore, the aim of this study was to investigate the impacts of proposed agricultural developments on the water flows in the upper uThukela Catchment using the multi-purpose, multi-soil-layered, daily time step ACRU model. The first phase of the study was to confirm the model’s ability to simulate flows in three, relatively small, gauged subcatchments of the uThukela catchment (Quaternary Catchments V11K, V14C and V31F), using current land cover and climate information extending to present day. However, the documented decline in the number of, and quality of data from, hydrometeorological stations, particularly since the year 2000, was concerning. Therefore, the impact of this decline on model performance was investigated in the selected subcatchments by comparing simulated flows to available observed flows in a confirmation study. Configuration of the model to present day conditions was restricted by the unavailability of rainfall stations. In cases where stations were available, there were no nearby stations to patch or compare to, when the record had missing or suspicious values. Given this, the model was set to run from 1960 to the latest record date available for catchments V14C and V31F. For V14C, the model performance decreased when the model was run from 1960 to 2012, compared to 1960-1999. Although a slightly better performance was obtained at V31F, the simulation time period was reduced to 1960-1999 for both catchments due to uncertainties with post 2000 rainfall and streamflow data. However, V14C continued to prove problematic and further investigation using of the Indicators of Hydrological Alteration software revealed a marked change in the flow characteristics between 1980 and 1981. No documentation of developments or substantial changes in the catchment could be sourced. Therefore, Quaternary Catchment (QC) V14C was excluded from further analysis. The ACRU model adequately simulated the flows for V11K and V31F, with the simulated flows being more representative of the observed flows in V31F. With the ability of the ACRU model to simulate the flows in the upper uThukela catchment under various land uses confirmed, the model could be used to investigate the impacts of agricultural land management scenarios on water flows. The agricultural land management scenarios were developed from the national and local government’s plan to expand agriculture to transform the socioeconomic status of the uThukela catchment. To develop scenarios for larger scale modelling, numerous scenarios were tested at QCs V31F and V11K. However, V11K was not responsive to changes in land use; therefore, results from the catchment were not used. For large scale modelling, the Upper uThukela (V1) Secondary Catchment was selected. The scenarios considered were: (i) increasing the fraction of irrigated commercial agriculture into currently dryland commercial fields, (ii) increasing subsistence agriculture through reduction of commercial agriculture (i.e. land reform), (iii) conversion of dryland commercial agriculture into crops with biofuel potential (iv) increased burning, (v) intensified land degradation and (vi) rehabilitation of degraded areas. These were developed from current land cover and compared to a simulation assuming natural conditions. The runoff components of interest were baseflow, quickflow and streamflow, as well as the low, median and high streamflows. Irrigation resulted in the highest flow reductions, with permanent cropping and planting two crops per year resulting in the largest decrease in streamflow at V31F and V1, when compared to natural conditions. These scenarios also had the greates impact on low flows. Plantation of biofuels increased flows, with soya beans having a higher impact on baseflows. Intensified burning and degradation increased quickflow and streamflow, while increasing subsistence agriculture and rehabilitation of degraded areas had little impact on flows. These results were generated from poor climate and land cover input information. Therefore, these results cannot be used at a definite decision-making tool, rather as an indication of the possible impacts of land use change on flows at the uThukela Catchment and similar regions. Efforts should be made to improve and maintain hydrometeorological monitoring stations. In addition, there should be more initiatives to collect land cover and water use data at various catchments in order to improve the quality of input data. Lastly, the current version of the ACRU model requires high computational power for large catchment simulations, lowering the model performance. Investigation into better versions or possible development of the current version should be conducted to enable modellers to finish large projects in allocated time.