Statistical model for land surface temperature change over mainland southeast Asia

Abstract

© Geoinformatics International. This study presents an alternative statistical methodology for estimating changes in land surface temperatures over mainland Southeast Asia (SEA). The method comprises of seasonal adjusting and autocorrelation filtering of MODIS LST time series obtained from 2000 to 2019 at systematic 45 sample locations. Furthermore, the filtered seasonal-adjusted LST time series were estimated to quantify the decadal change of LST using linear regression model. The long-term dynamic of temperature change was revealed by curve fitting using a spline model with different knots. The overall LST changes in sub-regional and regional scale were estimated using multivariate regression model which adjusted for spatial correlation and aggregated information of LST change from all individual sample locations irrespective of their strength of statistical evidence (p-value). The final result showed that the surface temperature change in the SEA region increases by 0.126 °C/decade. 95% confident interval for increasing ranges between 0.04 to 0.21 °C/decade, which shows evidence of substantial warming surface in this region.