Modelling MODIS Land Surface Temperature Change in Antarctica from 2000 to 2019 Using Cubic Spline Model

Abstract Land surface temperature (LST) data derived from the satellite is increasingly required to supplement the limited weather stations for assessing temperature trends in Antarctica. This study analyses the LST based on data from the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard NASA’s satellite length from 2000 to 2019 at a systematic 108 sub-regions. Antarctica was divided into 12 regions, each consisting of 9 sub-regions. A cubic spline model adjusted for seasonal patterns and the autoregressive process adjusted for time series correlation. Change in LST in sub-regions was estimated by fitting the simple linear model, while cycle and acceleration were estimated using cubic spline models. Multivariate regression adjusted for spatial correlation and was used to estimate the LST increase in regions. The seasonal patterns for all 108 sub-regions were found to be quite similar. Out of 108 sub-regions, only 30 had statistically significant decreasing trends. The 12 regions showed that most temperature trends decreased, although only 5 regions were statistically significant. The results for the entire Antarctic continent showed a statistically significant decrease and a 95% confidence interval ranging from -0.668 to -0.068 °C per decade. Keywords: Land surface temperature, MODIS, Cubic spline, Autocorrelations, Spatial correlations