Forecasting of Hydrological drought using Surface Water Supply Index and Streamflow Drought Index at Malewa River Catchment, Naivasha, Kenya

Abstract

Hydrological drought is reduction in rainfall leading to water’ decline under a characterized threshold level over an extended period causing reduction and drying up of waterbodies. Its onset and progress unnoticeably, and slowly, yet its impacts are cumulative and devastating leading to water shortage, conflict, ecosystem degradation, losses of agricultural production, famine and food crisis. Malewa River flow fluctuation leads to decrease in the flow regime and scarcity of water in dry season portending starting of hydrological drought in the Malewa River catchment. The objective was to predict and forecast long-term hydrological drought using Surface Water Supply Index (SWSI) and Streamflow Drought Index (SDI) in conjunction with Artificial Neural Network (ANN) in Malewa River catchment from 2019 to 2050. The SWSI and SDI index values were used to predict hydrological drought in the catchment where feedback NARX network model based on back propagation Levenberg-Marquardt algorithm was developed using the Neural Network Toolbox for MATLAB®. The validation metrics used were R² and RMSE to determine the best-forecasted models. It was projected that the catchment would mainly have near average of hydrological drought using SWSI index with values of -0.9 to 1.0. In SDI, the lead time of 12, 24, and 48 months indicated to be more prone to hydrological drought when compared with 3, 6, and 9 months. In spatial analysis, Eastern part of the catchment was projected to be humid with Southern part expected to be prone to hydrological drought. Forecasting of hydrological drought makes accurate and timely information about drought risks and therefore lays a foundation for long-term water resource management plans and decision-making. Drought forecasting allows National Drought Management Authority (NDMA) to develop drought preparedness plans.