Anticipating drought: enhancing prediction models and assessing environmental impact in Eswatini’s Maguga Basin

Drought is considered to be one of the most destructive natural disasters in the world. In Eswatini drought events have been found to be intensifying. They are a slow occurring natural process that has impacts on various environmental and socio-economic aspects. It is this slowness in their occurrence that makes droughts so dangerous. They are usually identified at a very late stage. At this point, the damage caused by the drought would have been severe and cost large amounts of money. It therefore becomes important to quantify the magnitude and forecast future values of drought events before their onset so as to better prepare for it. The Standardized Precipitation Evapotranspiration Index (SPEI) was used as a drought quantification index for this study. This study also finds the best suitable model for predicting droughts by comparing the performance of the Auto-regressive Integrated Moving Average (ARIMA) model with that of Genetic Algorithm (GA) optimized Long Short-Term Memory (LSTM) model. The GA optimized LSTM model was found to outperform the ARIMA model in predicting the drought. The environmental conditions also have to be noted during a drought event as they assist in further quantifying the severity of the drought. Hence, in this study, the water levels at Maguga Dam were also predicted to get a clear picture as to how they would be affected in the event that a drought occurs.