Development A Web-Based System for Real Time Prediction of Drought in Northern Nigeria Using Markov Chain Technique

Abstract

Drought is caused by a continuous decrease in rainfall over a long period of time, generally a season or longer, however, other climatic variables (such as high temperatures and low relative moistness) are strongly linked to it in many regions of the world and can exacerbate the severity of drought occurrence. Droughts also spread out more slowly and last far longer than other natural catastrophes (though the precise duration is unknown), making it hard to anticipate when they will start and stop. Because of its erratic rainfall pattern and short rainy season, the effects of drought in Northern Nigeria (North-east and North-west) are extremely severe. Approximately 300,000 animals perished in the North-Eastern portion of Nigeria during the drought of 1972-1973, accounting for roughly 13% of the total cattle population. However, research has shown that if individuals are aware of the approaching drought ahead of time, these impacts can be reduced. This research aims to develop a system that will help with early warning by monitoring and forecasting droughts. In this Project, Droughts are forecasted using a Markov chain algorithm based on data from sensors placed in strategic places, with warnings provided via a web interface. The Markov chain algorithm creates a transition matrix of stochastic processes using prior data, and then forecasts the next state using a dot product of the current state and the transition matrix. For the Markov chain prediction approach, the Mean Absolute Error (MAE) and Root Mean Absolute Error (RMSE) were about 21.48 percent and 21.01 which indicates that the model isn’t totally accurate.