Weather forecasting using soft computing models: A comparative study

Abstract

One of the main fields of weather forecasting is rainfall prediction, which is significant for water resource management, food production plan and different activity plans in nature. The appearance of stretched dry period or intensive rain at the critical stages of the crop growth and development may lead to serious reduce crop yield. Certainly, the accurate forecasting in rainfall could present useful information for water resource administration, flood control and disaster relief. This study proposed several soft computing models for long term rainfall prediction based on monthly meteorological dataset for 13 years, the models are IBK, K-Star, M5P, adaptive neuro-fuzzy inference system (ANFIS), Meta vote, bagging, staking and ensemble by using different machine learning schemes such as hybrid intelligent system, data mining, meta learning and ensemble algorithms.. The results show the accuracies of both ANFIS and the ensemble model are satisfied and ANFIS showed relatively more accurate results.