Improving rainfall erosivity estimates for the design of soil conservation structures in South Africa.

Abstract

Soil erosion is a major problem, both in South Africa and globally. Soil erosion reduces the productivity of land and has major environmental, as well as economic, impacts. South Africa, in particular, experiences considerable challenges in combatting soil erosion owing to a combination of factors. Examples of these factors include low vegetal cover as a result of arid climatic conditions, as well as intense thunderstorm activity. As more data and computing power become available, it is important that approaches to the design of soil conservation structures and design tools be updated, in order to reduce soil erosion. In this study, literature has been reviewed in order to obtain an overview of mechanical soil conservation measures in South Africa, soil loss estimation models currently used and design approaches to the determination of contour bank intervals. Literature showed that site-specific evaluation, using soil loss prediction tools, is the preferable approach to determining contour bank intervals, rather than the use of empirical equations. It was also found that the Revised Universal Soil Loss Equation (RUSLE) held the most potential as a model, in terms of creating a design tool for the design of soil conservation systems in South Africa. This was due to manageable input requirements as well as reliability – a result of widespread and extensive application of the model. This study applied the erosivity density approach in order to calculate rainfall erosivity (i.e. the ability of rainfall to cause erosion) across South Africa. Owing to the paucity of suitable short duration rainfall data, a second approach was attempted in which rainfall erosivity calculated from short duration data was related to daily rainfall data characteristics. It was found that the both approaches resulted in erosivity density patterns similar to what had been determined in previous studies. The erosivity density method produced results with a very fine level of detail, while the daily data method resulted in a more general overview of erosivity patterns and did not pick up localised variations as effectively. The erosivity density method showed lower rainfall erosivity values than the daily data method, in general. It also produced a much lower maximum annual rainfall erosivity than the daily data method (5 866 MJ.mm.ha-1.h-1 vs. 16 399 MJ.mm.ha-1.h-1). The verification of the interpolation of the erosivity density values gave poor results (an overall error of 75 %), indicating that the spatial density of the data was too low. This was improved in the daily data approach through the use of a greater number of daily rainfall stations, v achieving an overall interpolation error of 43%. However, when verifying the results against observed erosivity at test stations, the erosivity density method performed better, achieving an error of 55 %, compared with 91 % for the daily data method. Both methods showed potential, but require a larger network of short duration stations, in order to improve accuracy. A tool was developed to assist in contour interval determination. This took the form of a Microsoft Excel spreadsheet. The tool utilised the updated rainfall erosivity values determined in the study and focussed on determining recommended contour intervals in sugarcane plantations. The tool took into account the timing of erosive rainfall relative to crop development and tillage operations. Various scenarios were modelled and the results of the spreadsheet were compared to current methods used in the sugarcane industry. The spreadsheet was found to be highly sensitive to slope and the results suggested that soil erosion in sugarcane plantations has previously been underestimated, particularly on steep slopes. The study highlighted the need for ongoing research in the field of soil conservation.