Forecasting Bangladeshi monsoon rainfall using neural network and genetic algorithm approaches

Abstract

The important hydrological event rainfall is the quantity of water falling in drops from vapor condensed in the atmosphere. When water droplets in clouds become too heavy to stay in the air, they fall out towards the ground. This paper proposes an appropriate model to predict monsoon rainfall for different stations of Bangladesh. Making reliable prediction about rainfall is very important in many areas of human activities, indicated as follows: Rainfall mainly controls our water supplies, which are the basis for crops production. It is especially important for rainfed (or dry) land agriculture. For example, crops plant use a huge amount of water and rains, to a major extent, provide this water through the soil for the development and growth of plants. Since Bangladesh is an agro-based country, accurate prediction of rainfall will be useful for proper plannings of cultivation. That’s why those who are involved in agriculture, they will be interested to know whether the next days (or months) will be rainy/non-rainy. Although water is vital to life, yet water can be extremely destructive. Thus, rainfall forecasting can warn of happening flood or drought so that peoples can save their lives and properties. Rainfall forecasting is also important for engineering applications, mainly for the design of hydroelectric power projects, because this system requires prior information about average rainfall, maximum/minimum rainfall, maximum intensity, duration etc. for a year/each month. In urban areas, rainfall also has a strong influence on traffic control, operations of sewer systems and others. Thus, we believe that precise rainfall prediction is important for practitioners who are interested to make wise policies related to this event.