Variation and Prediction of Rainy Season in Thailand Using Ensemble Neural Model

摘要

The variation and prediction of rainy season play a role in many aspects, especially agriculture and water management resources. In order to examine the variation and prediction of rainfall, an ensemble neural model (ENM) is proposed. The model aims to explore the relationships between rainfall and other weather conditions and also to improve the accuracy in prediction skill. In the experiment, the monthly rainfall data by the Thai Meteorological Department (TMD) from 2013 to 2016 from five meteorological stations are used. They have been interpolated as observed data in the training set including the data from Coupled Model Intercomparison Project Phase 5 (CMIP5) as input data. The analysis provides the temperature, humidity, pressure and geopotential height that affect rainfall in Thailand. The ENM can improve the accuracy in prediction skill compared with the traditional artificial neural network.