Estimation of Spatio-Temporal Additive Model Using Mixed Model Approach with Application to Ozone Data in Surabaya City

Abstract

Spatio-temporal models arise when spatial data are collected over continuous time, so that one must take account of spatial correlations as well as temporal correlations. This research is focused in modelling spatia-temporal data using additive models. Spatio-temporal additive models are combination of time series additive models and spatial additive models. Functional form of predictors and response are modeled u sing P-spline. Since P-spline has connection with linear m.ixed models, the estimation of spatia-temporal additive model can be approached by linear mixed models. The models are applied to air pollution Ozone from AQMS in Surabaya. In modelling Ozone we add meteorological factors as covariates. This resulting model is used for spatial interpolation in unmonitored location. The study show that addition meteorological factors in spatio-temporal models improved the accuracy of models, although spatial interpolation are similar with models without meteorological factors.