Generalized additive models fitting with autocorrelation for sea surface temperature anomaly data

8th Annual International Conference 2018 on Science and Engineering

工业技术（总论）

Ananda, S.^1 ; Miftahuddin^1

Department of Statistics, Faculty of Mathematics and Sciences, Syiah Kuala University, Banda Aceh

23111, Indonesia^1

关键词: Auto-correlation value;    Autocorrelation structures;    Climatic conditions;    Generalized additive model;    Indian ocean dipoles;    Predictor variables;    Sea surface temperature (SST);    Sea surface temperature anomalies;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/523/1/012002/pdf
DOI  :  10.1088/1757-899X/523/1/012002

学科分类：工业工程学

来源: IOP

【 摘 要 】

Climatic conditions in Sumatra Island are affected by Sea Surface Temperature Anomaly (SSTA) in the Indian Ocean or more commonly referred to as Indian Ocean Dipole (IOD). Extreme climate events also closely related to SSTA. Several climate features that affect SSTA such as air temperature, precipitation rain, relative humidity, wind speed, and solar radiation. SSTA is an increase or decrease in the mean of Sea Surface Temperature so required the analysis to assess extreme climatic events to the risk due to the occurrence of anomalies. Generalized Additive Models (GAM) with autocorrelation can be used to model this phenomenon. GAM method accommodates the nonlinear influence between response variables and predictor variables. The data used in this research is the time series data, daily data from 2006-2017 and there are gaps in it, where there is an autocorrelation value. The purpose of this research is to get a representative model and to know the factors that influence to SSTA. The results show that GAM's best model with autocorrelation is GAM model by including month and year variables with the monthly autocorrelation structure. Factors that affecting SSTA are the air temperature, month and year as time covariates.