Evaluating the impacts of urbanisation on water resources using a hydropedological approach : a case study in the Crocodile River Catchment, Mpumalanga, South Africa

Abstract

In many parts of the world, urban growth is associated with socioeconomic development and improved lifestyles and is often linked to increased job opportunities. However, recent studies have linked urbanisation to changes in soil properties, including soil structure and soil horizons. Through soil survey and hydrological modelling, this study deepens the understanding of catchment-scale hydrological responses to flow-path modification resulting from land-use activities, such as urbanisation, using X22J, a quaternary catchment of the Crocodile River, as a case study. This study addresses key knowledge and empirical gaps in understanding the impacts of urbanisation on soil morphology and catchment hydrological processes. To assess urban use in 1990, 2007, and 2024, this study used the random forest algorithm in Google Earth Engine (GEE). An 80/20 data split was adopted, with 80% for training and 20% for accuracy assessment and validation. To predict urban land use for 2040, the Multi-Layer Perceptron-Artificial Neural Network-Cellular Automata (MLP-ANN-CA) algorithm within the Modules for Land Use Change Evaluation (MOLUSCE) plug-in in QGIS was utilised, and for validation, a simulated 2024 map was compared with the actual 2024 map. The overall accuracy values for the historical maps were 0.88, 0.91, and 0.87 for 1990, 2007, and 2024, respectively, whereas the predicted 2040 map obtained a kappa coefficient of 0.52 and an overall accuracy of 0.75. This study projected a 4.01% increase in urbanisation by 2040, consistent with the trend of increasing built-up areas since 1990. Consequently, agricultural land and water bodies are projected to decrease by 11% and 0.17%, respectively, by 2040. The transition probability matrix revealed a 20% loss of agricultural land to urbanisation between 2024 and 2040. Land-type data played a vital role in the development and establishment of hydropedological types. Land Type is a country-scale map that depicts a homogeneous combination of terrain type, climate, and, consequently, soil distribution patterns at a 1:250,000 scale. Each Land Type comprises soil forms and their percentage coverages across different Terrain Morphological Units (TMUs). The main hydropedological types were established by disaggregating the Land Type data in QGIS and by mapping them using a set of rules derived from topographic covariates (e.g., slope, curvature, and TMUs). The overall accuracy of the soil map was 0.60, with a kappa coefficient of 0.50. The dominant soil forms were Hutton, Clovelly, Glenrosa, Mispah, Technosols, Champagne, and Fernwood, corresponding to recharge (deep or shallow), responsive (shallow, Hortonian or saturated), and Interflow (soil/bedrock or A/B-horizon) hydropedological types. Subsequently, the study identified and mapped seven hillslope classes based on dominant hydropedological types. To quantify surface and subsurface flows, the hydropedological map was used as input to the Soil and Water Assessment Tool (SWAT) for the years 1990, 2007, 2024, and 2040. This study focused on Lat_Q (Interflow), PERC (recharge), SUR_Q (responsive) and evapotranspiration to simulate the hydropedological flows. Model performance was assessed using observed streamflow data and was acceptable, with NSE (Nash-Sutcliffe Efficiency) and KGE (Kling-Gupta Efficiency) exceeding 0.5 for both calibration and validation. The findings showed an increase in surface runoff with reduced percolation and baseflow in highly urbanised sub-basins from 2024 to 2040. This study further assessed the implications of soil map resolution for model performance by using three soil maps with different resolutions: FAO (53.8 x 107 m), Land Type (460m), and Hydropedology (61 m). The models performed within acceptable thresholds, with the model configured with the hydropedology map showing improved performance compared with models configured with the FAO and Land Type maps. Overall, this study provides valuable baseline insights for managing catchments undergoing urbanisation, with findings applicable to similar contexts beyond the X22J catchment. By integrating soil, land-use, and hydrological data, it provides practical tools to inform sustainable development policies that safeguard soil and water resources. The results underscore the importance of incorporating soil management within water resource and urban planning frameworks. Looking ahead, this research advocates continued innovation, strengthened institutional capacity, and policy reform to enhance resilience to the challenges posed by rapid urban growth.