Modelling of sea surface temperature by using generalized additive mixed models in risk detection

Rising temperatures from year by year caused an extreme weather which could result in all kinds of natural disasters. One of methods that can analyze the factors which affect sea surface temperature (SST) is Generalized Additive Mixed Model (GAMM). The GAMM is able to analyze more complex, especially those related to random effect and data which have not normal distribution. The purposes of this study is to construct a model of SST dataset by interacting with several variables, then the factors that affect SST be discovered. The type of data that used in this research is secondary data obtained through the National Oceanic and Atmospheric Administration website. The data has been taken from 2 locations, namely points 4 degrees N90 degrees E and 1.5 degrees N90 degrees E. The data used is daily unitin periods from September 17, 2006 to June 14, 2017. The variables used are SST, wind speed, air temperature, dynamic height, heat content, rainfall, relative humidity, shortwave radiation, sea surface density, and sea surface salinity, and a random effect namely location. Based on the results obtained the best model is the third model with Akaike Information Criterion, Bayesian Information Criterion and log-likelihood values respectively are -986,021, -894,911, and 515,752. Then the factors that significantly affect SST are wind speed (X-1), air temperature (X-2), dynamic height (X-3), heat content (X-4), sea surface density (X-8), and sea surface salinity (X-9). Furthermore, there are several interacting predictor variables that affect SST including interactions between air temperature and dynamic height, dynamic height and sea surface density, wind speed and rainfall, and air temperature with sea surface density.