A multi-criteria approach of assessing groundwater potential, recharge and flow in the transboundary Tuli Karoo Basin

The current increasing demand for water to satisfy cultural, societal and economic needs in the world can be fulfilled by either groundwater or surface water sources. However, comparing the two, groundwater is the wider distributed and relatively safer (Guppy et al. 2018). Groundwater, as a key component of the global water cycle, is an important resource of fresh water supply for agricultural, industrial, and domestic development in many parts of the world (Frappart & Ramillien 2018), particularly in arid and semi-arid areas (Arshad et al. 2020). Uses of groundwater in irrigation accounts for 43% of the total consumptive irrigation water use worldwide (Frappart & Ramillien 2018; Masocha et al. 2020). Groundwater also provides 50% of the world's drinking water and 40% of the world's industrial water (UNESCO 2015; Rajmohan et al. 2021). It is the primary source of drinking water for more than 1.5 billion people around the world (Frappart & Ramillien 2018). Moreover, groundwater is also used in desalination for domestic and industrial purposes (Panagopoulos 2021, 2020).

In Africa, groundwater is the major source of drinking water and its use for irrigation is forecast to increase substantially to combat food insecurity, particularly in northern and southern Africa (Calow et al. 2010). In southern Africa about 62% of the population depend on groundwater some of which are transboundary catchments. On the mainland of Africa, 72 transboundary aquifers (TBAs) have been identified, underlying 40% of the continent off which 33% of the population lives on TBAs (Nijsten et al. 2018). Of the 72 TBAs in Africa, 14 are in the Southern African Development Community region (SADC) (Davies et al. 2012). An assessment was done by Davies et al. (2012) to determine the degree to which these 14 transboundary aquifers could pose a threat to international relationships and would benefit from shared management through international cooperation. The Tuli Karoo Basin was one of the two aquifers that were identified to benefit from collaborative inter-sate management.

The exploration of groundwater is very much necessary for better development of groundwater resources and improvement of techniques for its investigation (Basavarajappa et al. 2015). Nagarajan & Singh (2009), states that deprived knowledge about groundwater, because of its hidden nature and its occurrence in complex subsurface formations, has been and is still a big obstacle to the efficient management of this important resource. To achieve sustainability on groundwater resources, there is need for a serious protection and monitoring. There are various ways of determining groundwater potential which includes geological and geophysical methods, However, these methods are considered to be time consuming and they require heavy and expensive equipment. Geographic Information System (GIS) and remote sensing as well as hydrogeological modelling has been used as powerful tools for monitoring and predicting the behaviour of groundwater potential in many parts of the world. Studies have been done to delineate groundwater potential zones by a number of researchers in different parts of the world (Magesh et al. 2012; Duan et al. 2016; Pinto et al. 2017; Shamuyarira 2017; Andualem & Demeke 2019; Ajay Kumar et al. 2020; Arshad et al. 2020). In Zimbabwe, a research study was done by Chikodzi & Mutowo (2014) to model the groundwater potential using Geographical Information System. However, the study was done on a larger scale and hence it provides results with a course resolution. Moreover, the study was done at a national level hence not covering the other parts of the transboundary Tuli Karoo Basin.

Groundwater recharge is an important component in the completion of the water balance but it is very difficult to quantify (Day & Simpkins 2018; Lekula & Lubczynski 2019). Understanding the spatial and temporal variation of groundwater recharge helps water managers achieve a better understanding of water availability and aquifer stress (Mushtaha et al. 2019). When abstraction levels exceed the rate of groundwater recharge, groundwater mining results (Custodio et al. 2016). This happened in southeastern peninsular Spain where groundwater was said to have been depleted to such an extent that the aquifer reserves needed more than 50 years to recover (Custodio et al. 2016). Groundwater recharge and groundwater flow models have become increasingly popular nowadays in the management, monitoring, assessing and forecasting of groundwater resources worldwide (Lekula et al. 2018; Smith & Berg 2020). There are some models that can be used to model groundwater recharge, for instance, MODFLOW, MIKESHE, pyEARTH and soil–water–balance (SWB) model (Smith & Berg 2020). SWB model estimates recharge based on the modified Thornthwaite–Mather soil–water accounting method (Thornthwaite & Mather 1957). However, some of these models cannot be used in areas that have challenges of inaccessibility of historical data like the Tuli Karoo Basin. This study therefore aimed at using hydrogeological modelling as well as GIS and remote sensing in assessing the groundwater resource in the transboundary Tuli Karoo Basin making use of readily available data.

The Tuli Karoo Basin lies in a semi-arid region with high water stress and is characterised by low surface runoff and high moisture deficits (Davies et al. 2012). There is little or no reliable and current quantitative information on groundwater resources on the Zimbabwean side of the Tuli Karoo Basin. One of the major constraints in attempting to understand the Tuli Karoo Basin is that there are often limitations on availability of groundwater data. Quantitative, spatially explicit information on groundwater in the Tuli Karoo Basin is required to characterize this resource in ways that can usefully inform strategies to adapt to growing water demand associated not only with population growth but also climate variability and change. The few past studies which have been done in the system do not provide information on the occurrence, the quantity of groundwater recharge as well as groundwater flow direction (Chikodzi 2013; Gomo & Vermeulen 2017; Ebrahim et al. 2019). In this regard, key quantitative information outlining the dimensions of the groundwater resources to date remains unresolved. This project seeks to address this significant knowledge gap by developing quantitative maps of groundwater potential zones as well as modelling groundwater recharge and groundwater flow direction in the Tuli Karoo Basin.

The objectives of the study were to (i) to delineate groundwater potential zones in the Tuli Karoo Basin using GIS and remote sensing, (ii) to determine the distribution of groundwater recharge in the Tuli Karoo Basin as well as to determine the direction of groundwater flow in the Tuli Karoo Basin. This study is relevant to water development policy, water-related development aid, community planning and technical decision making for hydrogeologists, catchment managers, water planners and non-specialists alike. Produced maps for groundwater potential zones can be used by institutes like Zimbabwe National Water Authority (ZINWA), Environmental Management Agency (EMA) and SADC Groundwater Management Institute (SADC GMI) for development, management as well as decision making. Moreover, this research is also important as climate change is reducing availability of surface water due to increased evaporative demand and as such groundwater will become more significant in dry areas.

More than 120,000 people stay in the Tuli Karoo Basin and almost half of the population falls within Zimbabwe. According to Ebrahim et al. (2019), Botswana is the wealthiest on a per capita basis of the three countries, nationally. Both Botswana and South Africa are middle-income countries with GDP per capita above US$6,000 per year. The Tuli Karoo Basin covers a number of economic activities in the fields of mining, agriculture and ecotourism (Masundire et al. 2016). The area contains significant deposits of coal and other minerals. Commercial irrigation exists in Botswana and South Africa. Communities are also involved in small-scale agriculture, both rain fed and irrigation, especially in Zimbabwe. Related, livestock is important, especially in parts of Botswana and Zimbabwe in the Tuli Karoo Basin (Sinthumule 2020).

More than 40% of water in the Tuli Karoo Basin is stored in large dams which are increasing significantly. The majority of this water is extracted and used (Ebrahim et al. 2019). In the Tuli Karoo Basin, irrigation is the main water user (>80%), followed by mining (6.5%), environmental flows (5.3%), water supply service comprising domestic use (2.5%), cattle (0.9%), and industrial use (0.9%), The Tuli Karoo Basin is part of the larger Karoo Super group and is made up of four types of aquifers. It consists of the sandstone aquifer, overlain by Karoo basalts and underlain by low permeability mudstones and fine-grained formations (Gomo & Vermeulen 2017). The aquifer may be confined and semi-confined in some parts of the basin.

To map the groundwater potential zones in the Tuli Karoo Basin, GIS and remote sensing techniques were used as recommended by various researchers (Arshad et al. 2020; Dar et al. 2020). The study involves the integration of nine thematic layers of rainfall, geology, lineament density, drainage density, slope, topographic wetness index, soil texture, soil type and landuse/landcover were prepared mainly in a GIS environment (Figure 3). The sources of data used for this study are shown in Table 1. Selection of the thematic layers was based on literature (Andualem & Demeke 2019; Dar et al. 2020). Some of these studies were also found in arid areas just like the Tuli Karoo Basin (Owolabi et al. 2020).

Soils are an important factor in determining groundwater potential or occurrence due to their major role in determining infiltration. Soil texture is a key aspect in assessing physical properties of soil and it has a direct link to the structure, porosity and adhesion (Dar et al. 2020). Soil data (1 km × 1 km resolution) were downloaded from International Soil Reference and Information Centre (ISRIC) world soil information. This was processed in the GIS environment to prepare a soil texture map. The average weight for soil texture thematic layer was 3.27%. The highest rank was given to soils that have a higher probability of groundwater potential, that is lithosols, because they are defined as shallow soils which lack a defined horizon and consists of imperfectly weathered rock fragments. This then implies that they have high permeability and high infiltration capacity. The least rank was given to cambic arenosols because they have a lower infiltration rate than lithosols. The average weightage given to soil lithology layer was 3.42%.

Drainage density is the total length of all streams or rivers in a drainage basin divided by the total area of the basin. It expresses the closeness of spacing of stream channels (Arshad et al. 2020). Drainage density plays an essential role in runoff distribution and infiltration because of its inverse relationship with permeability (Andualem & Demeke 2019). The less the permeability of a rock, the lesser the percolation of rainfall. This therefore shows that drainage density is crucial for the occurrence of groundwater potential since its computations show important hydrogeologic factors such as infiltration and permeability (Owolabi et al. 2020). To obtain a drainage density map of the Tuli Karoo Basin, a STRM digital elevation model (30 × 30 resolution) was used. DEM hydro-processing was done to extract streams for the study area which were then used for the creation of the drainage density map.

The occurrence and development of groundwater is also determined by landuse/landcover (LULC) (Day & Simpkins 2018). It controls the rate of infiltration and surface runoff, hence playing a significant role in the development of groundwater potential zone map. To add more, the classification of LULC provides crucial environmental insight into areas of groundwater accumulation based on human and interaction with natural settings (Misi et al. 2018). Dense forest area and agriculture have an excellent capability of recharge and groundwater storage, while exposed surfaces as well as bare and built-up areas are the least suitable for infiltration (Arshad et al. 2020). For this study, a prototype LULC was downloaded from the European Space Agency (ESA) and was processed in a GIS environment. The pre-processed prototype LULC map downloaded from ESA has nine classes which are shown on the LULC map in Figure 3.

Rainfall distribution is one of the most considered groundwater condition factors (Arshad et al. 2020). It is considered the major source of recharge hence showing its importance in groundwater potential zone mapping (Rahmati & Samani 2015). Satellite-based annual rainfall data for 1981 to 2020 were downloaded from the CHIRPS database. It incorporated 0.05° resolution. Because of the limitation of rain gauge stations, the study only used two rain gauge stations with rainfall data at daily time step which were cumulated to annual rainfall to validate the satellite-based rainfall estimates. The two stations used were Mzingwane and Bubi. Geology has a significant influence on both groundwater and subsurface fluxes and they should be taken into consideration when mapping groundwater potential zones (Rahmati & Samani 2015). Areas with high permeable subsoil are favourable for infiltration, hence promoting recharge, while areas with impermeable rocks hinder precipitation. Because of this, therefore, geology has become one of the factors to be considered for groundwater potential. In this study, a geology map was downloaded from the Geology of Africa website (https://certmapper.cr.usgs.gov/data/apps/world-maps/, access date:13/12/2020) and it was clipped to the study area in a GIS environment.

Lineaments are geomorphic features that express the zones of weakness or structural displacement of the surface of the Earth (Shebl & Csámer 2021). Lineaments can be identified from satellite imagery by their linear alignments (Arshad et al. 2020). It is a very important hydrogeological factor in determining groundwater potential since they provide pathways for the movement of groundwater potential. Their representation of faulting and fracturing zones increases the rate of secondary porosity and permeability (Dar et al. 2020). This therefore shows that lineament density influences the potential of groundwater and areas with high lineament density are good for groundwater potential. A Landsat image (30 × 30 m) resolution was used to automatically extract lineaments using an algorithm for multi-stage line detection of Canny edge and contour detection. Contour detection enables filtering of edge and curves (Owolabi et al. 2020). Line detection enables a four-stage transformation which includes speculation of maximum error, minimum of length of curves, maximum angle between polylines segments and the minimum distance between two polylines (Owolabi et al. 2020). A lineament density map was then created in a GIS environment.

Slope is also one of the principal groundwater potential controlling factor. This is because it controls the downward movement of water into the subsurface (Andualem & Demeke 2019). Slope is directly proportional to runoff amount and inversely proportional to the infiltration of surface water to groundwater storage. Steep slopes result in a quick runoff, and greater soil erosion rates with reduced infiltration (Arshad et al. 2020). For this study, an SRTM digital elevation model (30 × 30 m resolution) was downloaded from USGS Earth Explorer (http://glovis.usgs.gov/) and was processed to produce an elevation map of the Tuli Karoo Basin. According to Andualem & Demeke (2019), gently sloping areas favour the infiltration and recharge of groundwater, while the steep slopes facilitate surface runoff and therefore relatively less infiltrations. This class was given a weightage of 6.7%.

The stronger the influence of one factor the greater the relative importance, which results in larger weight (Dar et al. 2020). A Saaty scale 1 to 9 (Table 2) was used to resolve the relative significance of each thematic layer, where 1 represents equal importance and 9 represents extreme importance.

For validation of the groundwater potential zone maps, historical borehole yield data were used. The data were collected from the SADC Groundwater Management Institute (SADC GMI) Groundwater Information Portal (SADC-GIP) (https://sadc-gip.org/). The data were downloaded, and overlaid with the groundwater potential map. Values for the groundwater potential map were extracted using borehole yield data in order toto perform a statistical analysis using Kruskal–Wallis test. The Kruskal–Wallis test was used because the borehole yield data were not equally distributed among groundwater potential classes, therefore a non-parametric test which tests the hypothesis of rankings that are the same in different groups was needed. Moreover, for each groundwater potential zone class, a percentage of boreholes falling under different borehole yield ranges was determined. Generally, it is expected that most of the boreholes with low borehole yield will fall under areas classified as low potential zones, while areas with high borehole yields will be found in areas classified as high potential zones (Forkuor et al. 2013).

The model is physically based, deterministic and quasi three dimensional (Smith & Berg 2020). Recharge is calculated separately for each grid cell in the model domain (Westenbroek et al. 2010). For model inputs, climate data and landscape characteristics were used to determine sources and sinks of water within each grid cell. Recharge is calculated as the difference between the change in soil moisture and these sources and sinks.

The SWB model is a simple and reliable method for estimating groundwater recharge. It uses readily available data on soil, topography, and land cover, based on simple mass balance calculated at a daily time step. GIS was used in this project to assemble and generate the requisite input grids of the model. The model requires the user to provide tabular climatological and gridded land surface data in order to calculate a water budget and a recharge estimate for each grid cell (Table 3).

The input parameters needed to run this model are modelling extent, transmissivity, rivers shape file as well as a digital elevation model. The data sources for the groundwater flow map are shown in Table 1.

Figure 3 shows the inputs that were used for the groundwater potential map. The results showed that Tuli Karoo Basin is dominated by clay loam followed by sandy clay loam soil textures which cover 33.4% and 30.2% of the study area respectively. About 5.7% of the area is covered by sandy clay soil texture. Sandy loam soils, which covers about 14.7% of the study area, were given a higher weightage and rank because they have high infiltration and permeability rate. It was then followed by sandy clay loom which covers 33.4% of the area. Results for soil lithology showed that the Tuli Karoo Basin is dominated by lithosols, which covers about 49.9% (6,068.92 km²) of the basin. These are followed by Calcic Luvisols, which cover about 28% (3,410.01 km²) of the basin.

The results also showed that there is high drainage density 0.7 to 1.6 km/km². Tuli Karoo Basin is dominated by shrub cover areas which cover about 66.5% of the total area. This is followed by grassland which covers 25.5%. The results for landuse/landcover were similar to a study which was done by Makonyo & Msabi (2021) in a semi-arid midlands Manyara fractured aquifer which was also dominated by shrub cover areas. The minimum rainfall for the study area was found to be 202 mm with some parts of the basin having 477 mm. Areas like Toporo have low rainfall while some areas have high rainfall (314 to 357 mm). The rainfall results coincided with a observation that semi-arid regions in Africa fall within the 200 to 600 annual rainfall belt (Martiny et al. 2006). The Tuli Karoo Basin is also dominated by mesozoic extrusive and intrusive rocks. The results also showed that, generally, the Tuli Karoo Basin is dominated by low lying areas, covering about 27.2% and 38.5% of the basin.

Generally Tuli Karoo Basin is an area of high and moderate groundwater potential. This can be justified by the fact that the area is underlain by sandstones and Karoo basalt rocks which primarily have high permeability as well as high groundwater storage capacity due to the parent rock which is deeply weathered. To add more, this ensures that most of the rainfall is soaked into the ground and stored in underground aquifers. The higher percentage of groundwater potential in the Tuli Karoo Basin is also explained by the fact that the area has low altitude which promotes the occurrence of groundwater potential (Owen 1989). This notion is supported by Brown et al. (2003), who stated that the Limpopo Basin is largely underlain by older basement that is deeply weathered in many areas and that areas of of greenstones, sandstone, Karoo sequence and high grade sediments can provide moderate to high borehole yields that can be as high as 2.5 l/sec. In these areas, groundwater is readily available and reliable, in most cases, and is suitable for development of primary water at any point by means of boreholes or dug wells.

The results for this study on the Zimbabwean part, are similar to findings by Chikodzi & Mutowo (2014). The study was carried out in order to model the spatial variability of groundwater potential in Zimbabwe using GIS and remote sensing. Their results showed that there is high and moderate groundwater potential in the western and some parts of the southwestern parts where the Tuli Karoo Basin is located. However in a study done by Nijsten et al. (2018), the Tuli Karoo Basin was classified in areas of moderate and low groundwater potential with yields ranging from 0.5 to 5 l/s. This coincides with the results obtained from validation which were done using borehole yield data from SADC-GIP. Boreholes with yield that ranges from 0.1 were found in the low potential class. At the same time there are some areas classified as high which have borehole yield ranging from 40 to 250 l/s. The difference between the results from Nijsten et al. (2018) and the ones for this study might be attributed to the difference in spatial resolution. The latter used 30 m × 30 m resolution while the former used 5 km × 5 km resolution. Classification of groundwater potential zones is important for developers as it helps in identifying areas to prioritise for groundwater development, for example selecting areas of high water insecurity during droughts.

The results for the Kruskal–Wallis test showed that the median values of the borehole yields vary significantly between the different groundwater potential zone classes (P-value <0.0001 Kruskal–Wallis statistic 124.7). This means that the statistical test was highly significant. Although the groundwater conditioning factors used in this study might not be thorough, and the addition of other factors may improve the validation results, the groundwater potential zone map looks to be a reasonable reflection of the situation in Tuli Karoo Basin.

Land use land cover is very important when calculating recharge for the SWB model as it is the basis for the rainfall–runoff relationship (Day & Simpkins 2018). The modelled area contained nine types of land uses and landcover. The dominant class in the modelled area is shrub cover areas which covers 60.7% of the study area followed by grassland which covers 16.8%. Tree cover areas and cropland cover 12.2% and 9.5% of the area respectively. The least class in the modelled area is vegetation aquatic or regularly flooded which covers 0.005% of the modelled area. Tree cover and shrub cover areas have a better potential of groundwater recharge (Westenbroek et al. 2010). The flow direction map showed that 18% of the modelled area flows to the south, 17.1% flows to the north while 15.7% and 15.1% flows to the west and to east. 8.9%, 8.7% and 7.97% flows to the southeast, northeast and northwest respectively (Figure 7(d)).

Initial estimates of groundwater recharge usually include the rainfall analysis technique, where recharge is estimated as a percentage of precipitation. Because of this, precipitation data were compared to recharge. The total precipitation for the 10-year period from 2010 to 2020 was 3,702 mm and the total recharge for the same period was 30.3 mm giving a percent of 0.8 of recharge from precipitation. On an annual basis, recharge as a percent of precipitation ranged from 4.5 in 2011 to 0 in 2017. A linear regression performed to determine the relationship between recharge and gross precipitation showed that there is no strong relationship between gross precipitation and recharge in the Tuli Karoo Basin with R² = 0.0028, P = 0.8752). These results are different from the results obtained in a study in North Central Lowa by Day & Simpkins (2018) where precipitation had a strong correlation with recharge. The difference between these results might be explained by the difference in climate between the two study areas. North Central Lowa is located in humid areas while Tuli Karoo Basin is located in semi-arid regions.

Groundwater recharge is also determined by LULC. Tuli Karoo Basin is dominated by shrub cover areas which covers 66.5% of the study area. This is followed by grassland which covers 25.5% of the study area. The LULC with largest mean recharge was tree cover areas. Although it covers only 2.7% of the study area it had a mean recharge of 5.9 mm for the 10-year period (Table 4). The mean recharge for shrub cover areas and grasslands was 2.5 mm and 2.2 mm respectively for the 10-year period (Table 4). The lowest mean recharge was found in cells classified as bare, built up and open water, which had an average of 0 mm for the 10-year period.

Table 5 and Figure 9 show the relationship between surface water flow and groundwater flow per pixel (10 km × 10 km resolution). A performed correlation test between the two showed that there is no relationship between surface water flow direction and groundwater flow direction in the Tuli Karoo Basin (R² = 0.001857). In a study done by Barackman & Brusseau (2002), usually groundwater flow patterns on a regional scale follow surface flow patterns. However, some studies have shown that groundwater flow directions may be controlled by pumping wells in basins that have extensive pumping of groundwater (Barackman & Brusseau 2002). Barackman & Brusseau (2002) also stated that groundwater flow directions may also fluctuate in lowland areas near surface water as they respond to changes in surface water levels. The latter might be the reason why the groundwater flow direction in the Tuli Karoo Basin differs from that of surface water. However, it should be noted that, due to the inaccessibility of field-based data, this map was not validated.

The following conclusions were drawn from the obtained results:

• 1.

  Tuli Karoo Basin has high groundwater potential indicated by the dominance of high groundwater potential areas which covers 57.8% of the study area with very low groundwater potential covering 0.1%. The validation of the groundwater potential map using borehole data and borehole depth showed that the map produced looks to be a reasonable reflection of the situation in the Tuli Karoo Basin.

• 2.

  Tuli Karoo Basin has low groundwater recharge with a maximum of 13.2 mm per 10-year average. The results from the SWB model showed that there is no strong relationship between gross precipitation and recharge in the Tuli Karoo Basin with R² = 0.0028, P = 0.8752. This implies that groundwater recharge in the Tuli Karoo Basin is not precipitation driven.

• 3.

  The study showed that groundwater in the Tuli Karoo Basin flows towards northeast and in some parts of the basin, groundwater flows from northwest towards southeast where there is the Limpopo River.

It is therefore recommended that:

• 1.

  Reliable hydrogeological data such as borehole drilling logs, stratigraphic data, streamflow data and historical borehole water levels should be provided to promote a better understanding as well as sustainable management and monitoring of groundwater as a resource.

• 2.

  Further studies should be done on estimation of groundwater recharge in the Tuli Karoo Basin using other groundwater recharge techniques such as base flow separation that uses field-based data.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.