Spatio-Temporal Model of Rainfall Data Using Kalman Filter and Expectation-Maximization Algorithm

Abstract

Location and time dimension data modeling, also known as spatial-temporal data, generally has high complexity. This study analyzes a spatial-temporal model of rainfall data and climate variables, namely temperature, and humidity. The complexity of the relationship between variables and parameters in the spatial-temporal model is simplified by a hierarchical approach. The parameter estimation of the ratio-temporal model uses the Kalman Filter approaches and the Expectation-Maximization (EM) method combined with the bootstrap method to calculate the standard error estimation. Implementation of the spatial-temporal model on rainfall data in South Sulawesi Province with temperature and humidity shows that there is a relationship between rainfall and temperature and humidity.