Statistical Modelling for Forecasting Land Surface Temperature Change in Thailand

Abstract

It is anticipated that Thailand's climate will keep changing. Due to its increased sensitivity to natural hazards for example severe rainfall, floods, forest fires, deforestation, and vegetation fluctuation. It is extremely vulnerable to the effects of climate modification. However, the variations in the Land Surface Temperature (LST) are a major cause of climate change. The study aims are to explore the daytime LST yearly seasonal patterns and trends, and to forecast LST variation in sub-regions and regions in Thailand. The daytime LST time series data from 2000 to 2023 was downloaded from the Moderate Resolution Imaging Spectroradiometer (MODIS) website. To model the yearly seasonal patterns of LST in the daytime was applied the natural cubic spline method with eight knots. The linear regression model was used to demonstrate the LST trends. Moreover, to forecast LST trends over 23 years was applied a cubic spline with 2, 3, and 4 knots. Finally, to adjust spatial correlation and to estimate the increase in daytime LST was used the multivariate regression model. The results show that, there was an increasing LST trends in Thailand. The daytime LST by sub-regions were increased in the southern and central of Thailand. Furthermore, the southern and central regions demonstrate the increasing trends of the daytime LST. The eastern region shows the stable while the northern region was likely increase of LST trends. The mean increase of LST per decade was 0.115 °C. Therefore, by applying a cubic spline with three knots to forecast the daytime LST trends demonstrates the significant rise trends compared with other spline knots.  Nevertheless, LST of the daytime in Thailand is steadily increasing. The increasing reasons is needed to be explored in future studies.