#02 2024

Future outcomes of choices made today

By Mkholo Maseko, Junior Groundwater Hydrologist, Grasslands Node

Water security is a fundamental prerequisite for fostering economic development and alleviating poverty in South Africa. However, cumulative impacts on water-providing ecosystems are reducing their potential to supply water, thus hindering prospects for poverty alleviation. This challenge is particularly acute in underdeveloped rural social-ecological systems, where livelihoods are intimately linked to water-providing ecosystems, rendering them especially vulnerable. 

The impacts on water resources stem from two primary drivers: firstly, climate change and its associated extreme climate events, which rural communities have limited immediate control over; and secondly, anthropogenic impacts, including land use change, over which rural communities potentially wield some influence.

The land use choices made presently, as well as those in the past, will either exacerbate vulnerability to climate change-related threats to water security, or may assist in mitigating such impacts. Thus, today’s decisions may significantly shape the adaptive capacity and future trajectory of economic and water security within a region.

A stark illustration of this critical juncture is evidenced by the declining water levels in South Africa’s largest natural freshwater lake, Lake Sibaya.   

Modelling future scenarios 

As part of a three-year interdisciplinary Water Research Commission project (Project No. C2020/2021-0043), a MODFLOW groundwater model previously developed for Lake Sibaya by Kelbe (2020) was updated. This update involved integrating SAEON’s monitoring data and coupling it with the ACRU Agrohydrological surface water model. This loosely coupled modelling approach aimed to assess the impact of historical and future land-use and land cover (LULC) changes in the region, under different future climate storylines.

A noteworthy advancement of this project was the incorporation of dynamic historical LULC data from verified LULC maps and spatiotemporal climate variability, utilising gridded (and corrected) climate products. Future LULC scenarios were collaboratively developed with community members and each of these were simulated in the hydrological models. These scenarios ranged from replacing commercial timber plantations with various alternatives to assessing a “status quo” (no land-use change) scenario and a “no forestry” scenario.

Figure 1. Comparison of past to present and future lake level trends for the “no forestry” LULC scenario relative to the “status quo” LULC, under CMIP6 bias correct wet and ERA5 + 10% wet future climate storylines.

Figure 2. Comparison of 50% increase in bush encroachment (thicket) future scenario against status quo using for C6Wet, C6Dry and ERA5Wet future climate storylines.

The hydrological analysis revealed that if plantation forestry had not been introduced, the drought threshold (no more than five consecutive years with water levels below 16.5 m above mean sea level [AMSL]) would not have been exceeded, and lake levels would have been at least 1.5 m higher by 2020 (Figure 1). In this scenario, water levels would have remained above 16.6 m AMSL until 2020. However, the declining trend in lake levels under this scenario highlighted the influence of climate.

The study utilised three future climate storylines to represent potential future climate trajectories: “worst-case” (dry, with a low frequency of extreme rainfall events – Dry_LER), slightly wetter but still with a low frequency of extreme rainfall events (Wet_LER) and “best-case” (with regular rainfall and frequent extreme rainfall events – Wet_FER).

Looking into the future, under the Dry_LER climate storyline, with a “no forestry” LULC scenario, the 2020 water level crisis would have only occurred in 2040, and by 2050 under the Wet_LER future climate storyline. However, recovery to 19.0 m AMSL by 2050 was simulated under a Wet_FER climate storyline. The declining trend into the future under both LER future climates for the “no forestry” scenario highlights the significant impact of future climate with fewer extreme rainfall events, while the results from the Wet_FER demonstrate the need for multiple extreme events and good rainfall to enable the system to recover.

All other scenarios examined under a FEL climate storyline showed a trend of declining lake levels. However, replacing 50% of commercial plantation forestry with “dryland crop” or “dryland marula” had a more favourable impact on lake levels compared to the status quo, particularly towards the end of the simulation period, despite overall declines. This suggested a probable lag in the system’s recovery.

Conversely, under the Wet_FER climate storyline, all scenarios indicated a recovery, reflecting the impact of increased precipitation characterised by more extreme rainfall events. Retaining the current land use (SQ) alongside bush encroachment emerges as the least favourable option across all climate scenarios.

Efforts were made to integrate the economic system dynamics model with the hydrological model to assess the socioeconomic consequences of the simulated future scenarios. The main integration point was evaluating groundwater availability compared to crop/tree irrigation demand.

The hypothetical scenarios modelled hydrologically were included in the economic system dynamic model (what is the economic impact of converting 50% of commercial forestry plantations to dryland crops, marula or macadamia orchards?). Furthermore, the systems model was designed to include non-land-based options such as increasing tourist occupancy. The impact of bush encroachment on cattle production was also included in the systems model.

The hydrological analysis revealed that if plantation forestry had not been introduced, the drought threshold would not have been exceeded, and lake levels would have been at least 1.5 m higher by 2020 (Photo: Shutterstock/Alexander Narraina)

Among the crops, cassava showed higher cumulative benefits per hectare compared to maize, groundnuts and vegetables, proving more resilient under drier future climate conditions (Photo: PlantVillage)

Economic indicators 

Three economic indicators were used in the systems dynamic model: actual cumulative value (for household activities), employment potential, and net present value (NPV) of net cashflow. Estimated increases in sectoral cashflow NPV ranged from 9% to 33% higher comparing the status quo for the Dry_LER and Wet_HER climate storylines respectively, depending on the LULC scenario.

However, under certain scenarios such as mixed dryland crops, worse economic outcomes were observed, particularly under LER climate futures. Among the crops, cassava showed higher cumulative benefits per hectare compared to maize, groundnuts and vegetables, proving more resilient under drier future climate conditions.

In addition, switching to cassava is likely to yield more favourable outcomes for households compared to other scenarios. Under this scenario, employment is projected to increase significantly, ranging from 18.47% to 205.12% compared to the “status quo” scenario. Conversely, in wetter future climate storylines, several LULC scenarios were identified to result in net gains relative to the status quo. However, a 50% loss of grasslands to bush encroachment is anticipated to decrease livestock productivity, reducing the cattle population by 13% in the area.


The outcomes of different hypothetical land-use scenarios were summarised in a simple matrix for each climate future storyline. Table 1 provides an example of the Dry_LER future climate storyline. These summaries compare the consequences of LULC decisions in terms of water security, employment, economic well-being and natural resource (ecosystem) integrity.

Using the matrix in Table 1, it is evident that, under the Dry_LER future, dryland macadamia, cassava and marula appear to be water-wise and economically viable choices that provide better benefits than the status quo. The status quo, coupled with bush encroachment, resulted in the worst outcomes for all challenges relative to the other choices tested.

While the dryland and irrigated mixed crop LULC had less impact on the water table than the current LULC, they would not be economically viable. The reason for irrigated scenarios not performing as well as one may think, is due to the feedback with the hydrological model which indicated water available for irrigation, which was limited under this future climate storyline.

The scale utilised in the summary matrix demonstrates a proof of concept for a simple storyboard approach for engagement. However, it is currently too crude to discern nuanced differences between choices, particularly regarding the extent of economic and employment returns. A broader scale range is recommended, but this adjustment should be based on more accurately parameterised economic input data.

Table 1. Summary matrix of impacts of different LULC scenarios under the warmer drier (C6Dry) climate future storyline in relation to four key challenges. Green indicates a positive impact relative to the SQ, orange indicates a negative impact relative to the SQ, clear is no change. Red text indicates that there is no allocable water for that scenario based on the water level in relation to the resource quality objectives (RQOs).


The impact of nine land-use choices on water security under three climate future storylines was assessed. Climate, economic and hydrological parameters were integrated using a systems dynamic model to evaluate the consequences of LULC choices. This assessment was supplemented by a summary framework that incorporated additional hydrological consequences. These consequences were evaluated in relation to the impact on the four key challenges identified by the community – water security, employment, economic status and natural resource (ecosystem) integrity under different climate future storylines.

Caricatured climate future storylines and “immediate” LULC switch scenarios were employed to test if integration was feasible. The achieved integration demonstrated a sound proof of concept, which can now be refined, focusing on parameterisation. While the results from individual components offer useful insights, the net effect outcomes should be utilised for demonstration purposes only until various components are more adequately parameterised.