Exploratory Analysis of Rainfall Occurrence in South Sulawesi Region Using Spatial Point Process

Nurtiti Sunusi,Giarno Giarno,Fachrul Nawawi,Erna Tri Herdiani

doi:10.3844/jmssp.2015.113.118

Abstract

This paper study the probability of rainfall occurrence in round year in different segment in South Sulawesi region. In this research, rainfall occurrence in round year described by one line which has divided into 12 months. Each one of those months is assumed that the probability of a rainfall follow a homogeneous Poisson distribution. To modeling the rainfall occurrence in round year, a spatial point process is used. The parameter of the model is estimated by Seemingly Unrelated Regression (SUR) method and Ordinary Least Square (OLS) method with assume that two stations have a correlation in residual model. Results of case study on monthly rainfall data indicate that when the residual correlation (autocorrelation) on all models is weakly and not significant. Thus, it has not good enough to use the SUR method for increase efficiency compared with the OLS method. Moreover, results of the parameter estimation of the model for two selected stations (Paotere and Mandai) showed that the SUR method is more representative than the OLS method.

Highlights

Forecasting is a science to predict events in the future which can be done by using past data into a mathematical model to predict the future of data
Location was divided into two stations that each station modeled into a regression equation in which every equation has a parameter that can be found with the usual Ordinary Least Square (OLS) method
Due to the correlation between the errors that occurs resulting parameters are obtained in theory does not possess Best Linear Unbiased Estimator (BLUE)

Summary

Introduction

Forecasting is a science to predict events in the future which can be done by using past data into a mathematical model to predict the future of data. In forecasting, data that has dependencies on time is used. It was taken in a certain time within the same time interval. The influence of the location (space) is taken into account, or in this case knows as the space-time data. Rainfall phenomena are occurs in random and has dependencies on time. Spatial Point Process is a stochastic model that was built on the site of a phenomenon {Si} on the set X. There is extensive literature on the use of Poisson cluster processes in the stochastic modeling of rainfall, stemming largely from (Rodriguez-Iturbe et al, 1987; Onof et al, 2000; Cameron et al, 2000). Rainfall modeling can generally be classified into four categories (Onof et al, 2000): (1) Meteorological models involving complex sets of differential equations representing the physical processes controlling precipitation and other weather variables; (2) stochastic multiscale models describing the spatial evolution of the rainfall process independently of scale; (3) statistical models which can allow for the modeling of trends; and (4) stochastic process

Results

Discussion

Conclusion