*2.1. Study Area and Data*

The Olifants River Basin is one of the nineteen water management areas in South Africa. It is a principal sub-catchment of the Limpopo River. It originates in the north of South Africa in the province of Mpumalanga and flows northeast through the northern province before joining in Mozambique and emptying into the Indian Ocean (Figure 1). An estimated 3.2 million people live within the catchment area with two-thirds of this population living in the rural community [26]. The Olifants River Basin is recognized as one of the most important basins in South Africa as it contributes largely to the country's economic hub, with an annual contribution of six percent to the Gross Domestic Product arising from agricultural, mining, and industrial activities [27]. The catchment consists of both large and medium-scale agricultural farms that consume a lot of water for irrigation (540 Mm<sup>3</sup> per year) with approximately 130,000 ha irrigated (i.e., 11% of the total cultivated area in the catchment), primarily in the commercial farming sector. The water used for irrigation is obtained from both dams and groundwater in the catchment [28]. Precipitation in the catchment occurs during the summer months from October to April, with average annual rainfall ranging between 500 mm to 800 mm in most parts of the catchment and surpasses 1000 mm along the escarpment which separates the Highveld from the Lowveld. Evaporation varies across the catchment with high levels occurring in the north and west, and lower levels of evaporation recorded in the southeast. Elevations range from 300 m to over 2300 m above sea level, which explains the relatively cool winter and annual widerange of temperature variations of −4 to 35 ◦C [29]. Runoff from the catchment reflects the temporal and spatial distribution of the rainfall with the greatest volumes occurring in the south and along the escarpment. The average annual runoff from the catchment is 37.5 mm (i.e., 6% of the average annual rainfall), which equates to 2040 million cubic meters (Mm3). However, there is considerable inter-annual variation and consecutive years where the flow is below the mean annual discharge [30].

**Figure 1.** Map of Olifants River Basin showing towns, rivers, and irrigated areas.

This study uses climate simulated data (daily rainfall, minimum, and maximum temperature, average relative humidity, 2 m wind speed, and solar radiation) from the Coordinated Regional Climate Downscaling Experiment (CORDEX) database. The data were obtained from http://cordexesg.dmi.dk/esgf-web-fe/. The output from CORDEX RCMs are quality controlled and can be used according to the terms of use (http://wcrpcordex.ipsl.jussieu.fr/). It should be noted that all Coordinated Regional Downscaling Experiment-Regional Climate Models (CORDEX RCMs) are set to 0.44◦ by 0.44◦spatial resolutions, which corresponds to 50 km by 50 km. The area-weighted average method [31] was used to calculate the average climate variables from the CORDEX RCMs over the entire Olifants River catchment (latitude 24◦ and 26◦ and longitude 29◦ and 32◦). Daily climate variables listed above were obtained from a single RCM driven by six Global Climate Models (GCMs) namely, Commonwealth Scientific and Industrial (CSIRO), National Centre for Meteorological Research (CNRM), Canadian Centre for Modelling and Analysis (CCMA), Institut Pierre Simon Laplace (IPSL), Model for Interdisciplinary Research on Climate (MIROC), and Max Planck institute for Meteorological Earth System Model (MPI-ESM) for a period of 30 years (1976–2005). Two projected climate change scenarios, representative concentration pathways (RCP 4.5 and 8.5) were used. The former represents an intermediate stabilization of radioactive forcing by 2100, without surpassing 4.5 W/m<sup>2</sup> (~650 ppm CO2), which constitutes a high mitigation scenario [32]. Whilst the RCP 8.5 scenario assumes that the radioactive forcing pathway reaches above 8.5 W/m<sup>2</sup> (~1370 ppm CO2) by 2100 [33].

Using CORDEX-RCM climate change data, three time periods 2010–2039, 2040–2069, 2070–2099 were considered in this study to project future climate for both greenhouse gas concentration scenarios. These time periods were then compared to the baseline period, 1976–2005. CORDEX climate variables were biased corrected using a linear scaling bias correction method. Observed climate variables obtained from the South African weather

service were used to bias correct the current and projected CORDEX climate data [34]. It was necessary to bias correct the simulated climate data in order to compensate for any over or underestimation of the downscaled variables. The linear scaling bias correction is based on the average difference between daily observed time series data. These differences were then applied to the simulated climate data to obtain bias-corrected variables [34]. The biased-corrected climate variables were then integrated into a decision support system (Water Evaluation and Planning) model to evaluate current and future crop yield and adaptation scenarios using the WEAP-MABIA method.

The WEAP-MABIA model used in this study has a soil profile library functionality that provides typical values for water content at saturation, field capacity, wilting point, and the available water capacity for 12 textural classes. It uses a pedotransfer function to estimate the average soil water capacity. In this study, we assumed scenarios of three textural classes of soil to evaluate its impacts on crop yield under current climate conditions. The three textural classes of soil were (S1-sandy loam, S2-loamy sand, and S3-Sandy clay loam) presented in Table 1.

**Table 1.** Classification soil type used in this study (S1 = Sandy loam, S2 = Loamy sand, and S3 = Sandy clay loam).


Crop parameters were also obtained from the crop library functionality within the WEAP-MABIA. The "Crop Library" provides the required parameters for over 100 crops, some with multiple entries for different climates or regions of the world. The end-user can add, edit, remove, copy, export, import, or search the "Crop Library" for a particular crop. This study selected four crops namely: maize, soya beans, dry beans, and sunflower from the crop library using the crop scheduling wizard. The crop parameter used in this study is presented in Table 2.

**Table 2.** Database of crop parameters used in this study.


The lengths of growth stages (Lini, Ldev, Lmid, Llate) were computed according to the FAO-56 method as a function of vegetation cover (fc). The initial stage (Lini) runs from the sowing date to when the fc reaches a value of 0.1, the development stage (Ldev) runs from a fc of 0.1 to full vegetation cover (fc of 0.9). The mid-season stage (Lmid) runs from the end of the development stage until canopy cover (fc) drops back to the same value it had at the end of the development stage and the beginning of the mid-season period (fc = 0.9). The late-season stage (Llate) runs from the end of the mid-season stage until the end of the growing season.

The basal crop coefficient (Kcb) is defined as the ratio of the crop evapotranspiration ETc over the reference evapotranspiration ETref when the soil surface is dry but transpiration is occurring at a potential rate. Therefore, Kcb represents primarily the transpiration component of ETc. The Kcb coefficient serves as a lumped parameter for the physical and physiological differences between crops. Variation in Kcb between the growth stages is mainly dependent on how the crop canopy develops. The values given in the "crop library" represent a standard climate having mean daily minimum relative humidity (RHmin) equal to 45% and mean daily wind speed measured at 2 m (u2) equal to2ms<sup>−</sup>1.

The depletion factor (p) is the fraction of the total available water (TAW) that can be depleted from the root zone before moisture stress occurs. Different values can be defined to express the variation of the crop sensitivity to water shortage over the different crop stages.

The yield response factor (Ky) is a factor that describes the reduction in relative yield according to the reduction in the crop evapotranspiration (ETc) caused by soil water shortage. Ky values are crop-specific and may vary over the growing season. The values for Ky are given for the individual growth periods as well as for the complete growing season.

The rooting depth for annual crops has three growth stages. The rooting depth is held constant at the minimum depth (Zr min) throughout the initial crop growth stage. The root zone increases linearly to a maximum depth (Zr max) throughout the vegetative growth and development stages

The maximum root depth is attained at the beginning of the mid-season stage (peak growth) and is maintained throughout the mid and late season stage [35].
