ABSTRACT
Throughout history, most human settlements, however small or prosperous, have had to contend with drought. Droughts can broadly be defined as a negative departure from the normal precipitation over a period of time in a given area. Different definitions of drought exist. They have mostly been linked to changes in the precipitation regime, but other climatic factors, such as high temperatures and high winds can increase the severity of an event. Droughts often become highly visible when they are associated with famine yet, for the most part droughts can occur without resulting in famine, and famines have frequently taken place in the absence of drought conditions.
Given the potential impacts of droughts, it is essential that they are forecasted accurately with sufficient lead time to help mitigate some of the consequences. Drought forecasts can be done using either physical/conceptual models or data-driven models. The latter have become increasingly popular in hydrologic forecasting due to minimum information requirements, rapid development times, and have been found to be accurate in various hydrological applications. This chapter considers the role of various indices, along with artificial neural networks (ANN) and other modelling techniques in the prediction and mitigation of drought.
