Makungo, R.Madula, Andy2025-09-152025-09-152025-09-05Madula, A. 2025. Time series modelling of groundwater levels in a selected semi-arid catchment within Vhembe District Municipality, South Africa. . .https://univendspace.univen.ac.za/handle/11602/2940MESHWRDepartment of Earth SciencesThis study is aimed at modelling groundwater levels in a semi-arid catchment within Vhembe District Municipality, South Africa. Auto Regressive Integrated Moving Average (ARIMA) model and Seasonal Auto Regressive Integrated Moving Average with eXogenous variables (SARIMAX) model were used to model the interaction between groundwater levels, temperature, wind speed, evaporation, and rainfall. The unpredictable occurrence of rainfall and other weather conditions in semi-arid and arid regions has caused a restriction in simpler computations of groundwater levels. Historical hydrological data sets of groundwater level were used to model groundwater level and forecast with ARIMA model. Climatic variables like temperature, wind speed, evaporation and precipitation were employed as exogeneous variables of the SARIMAX simulations. The analysis of groundwater levels included the use of Sen’s slope estimator which showed significant fluctuations, with sharp declines, followed by stability and a subsequent increase over time. The ARIMA model’s forecasting of groundwater levels indicated a stable groundwater levels trend post-2016. The training data revealed historical fluctuations, while the test data showed a sharp increase before stabilizing. The SARIMAX model demonstrated a reasonable predictive accuracy for groundwater levels, incorporating significant predictors and seasonal patterns. However, diagnostic tests suggested further model enhancements could improve residual handling. Data stations were selected based on availability of long-term data and considering stations with minimal or no gaps. The data range of the study was 12 years from 2007 to 2018. Both ARIMA and SARIMAX models performed well in predicting groundwater levels. The inclusion of exogenous variables in the SARIMAX model offered a more nuanced understanding of data trends, making it a reliable tool for forecasting. The study findings showed that the groundwater levels in the Luvuvhu River catchment have a stable increase over time and also highlighted the issue of missing data in climatic variables like precipitation which prevented the SARIMAX accuracy in forecasting groundwater levels. This study's insights are valuable for developing effective groundwater management strategies, especially when compared with other studies that highlight the importance of incorporating climate variability into such models for enhanced accuracy. The SARIMAX model's application in predicting groundwater levels is a significant step in environmental modelling, offering insights into subterranean water dynamics crucial for sustainable management.1 online resource (xi, 120 leaves): color illustrations, color mapsenUniversity of VendaUCTDDissertationMadula A. Time series modelling of groundwater levels in a selected semi-arid catchment within Vhembe District Municipality, South Africa. []. , 2025 [cited yyyy month dd]. Available from:Madula, A. (2025). <i>Time series modelling of groundwater levels in a selected semi-arid catchment within Vhembe District Municipality, South Africa</i>. (). . Retrieved fromMadula, Andy. <i>"Time series modelling of groundwater levels in a selected semi-arid catchment within Vhembe District Municipality, South Africa."</i> ., , 2025.TY - Dissertation AU - Madula, Andy AB - This study is aimed at modelling groundwater levels in a semi-arid catchment within Vhembe District Municipality, South Africa. Auto Regressive Integrated Moving Average (ARIMA) model and Seasonal Auto Regressive Integrated Moving Average with eXogenous variables (SARIMAX) model were used to model the interaction between groundwater levels, temperature, wind speed, evaporation, and rainfall. The unpredictable occurrence of rainfall and other weather conditions in semi-arid and arid regions has caused a restriction in simpler computations of groundwater levels. Historical hydrological data sets of groundwater level were used to model groundwater level and forecast with ARIMA model. Climatic variables like temperature, wind speed, evaporation and precipitation were employed as exogeneous variables of the SARIMAX simulations. The analysis of groundwater levels included the use of Sen’s slope estimator which showed significant fluctuations, with sharp declines, followed by stability and a subsequent increase over time. The ARIMA model’s forecasting of groundwater levels indicated a stable groundwater levels trend post-2016. The training data revealed historical fluctuations, while the test data showed a sharp increase before stabilizing. The SARIMAX model demonstrated a reasonable predictive accuracy for groundwater levels, incorporating significant predictors and seasonal patterns. However, diagnostic tests suggested further model enhancements could improve residual handling. Data stations were selected based on availability of long-term data and considering stations with minimal or no gaps. The data range of the study was 12 years from 2007 to 2018. Both ARIMA and SARIMAX models performed well in predicting groundwater levels. The inclusion of exogenous variables in the SARIMAX model offered a more nuanced understanding of data trends, making it a reliable tool for forecasting. The study findings showed that the groundwater levels in the Luvuvhu River catchment have a stable increase over time and also highlighted the issue of missing data in climatic variables like precipitation which prevented the SARIMAX accuracy in forecasting groundwater levels. This study's insights are valuable for developing effective groundwater management strategies, especially when compared with other studies that highlight the importance of incorporating climate variability into such models for enhanced accuracy. The SARIMAX model's application in predicting groundwater levels is a significant step in environmental modelling, offering insights into subterranean water dynamics crucial for sustainable management. DA - 2025-09-05 DB - ResearchSpace DP - Univen LK - https://univendspace.univen.ac.za PY - 2025 T1 - Time series modelling of groundwater levels in a selected semi-arid catchment within Vhembe District Municipality, South Africa TI - Time series modelling of groundwater levels in a selected semi-arid catchment within Vhembe District Municipality, South Africa UR - ER -