Examining climate-related satellite data that strongly relate to seasonal phenomena requires appropriate methods for detecting the seasonality to accommodate different temporal resolutions, high signal variability and consecutive missing values in the data series. Detection of satellite-based Land Surface Temperature (LST) seasonality is essential and challenging due to missing data and noise in time series data, particularly in tropical regions with heavy cloud cover and rainy seasons. We used a semi-parametric approach, involving the cubic spline function with the annual periodic boundary condition and weighted least square (WLS) regression, to extract annual LST seasonal pattern without attempting to estimate the missing values. The time series from daytime Aqua eight-day MODIS LST located on Phuket Island, southern Thailand, was selected for seasonal extraction modelling across three different land cover types. The spline-based technique with appropriate number and placement of knots produces an acceptable seasonal pattern of surface temperature time series that reflects the actual local season and weather. Finally, the approach was applied to the morning and afternoon MODIS LST datasets (MOD11A2 and MYD11A2) to demonstrate its application on seasonally-adjusted long-term LST time series. The surface temperature trend in both space and time was examined to reveal the overall 10-year period trend of LST in the study area. The result of decadal trend analysis shows that various Land Use and Land Cover (LULC) types have increasing, but variable surface temperature trends.
This is an open access article distributed under the Creative Commons Attribution License
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited