The multi-timescale temporal patterns and dynamics of land surface temperature using Ensemble Empirical Mode...

Area of Science:

• Earth and Environmental Sciences
• Remote Sensing
• Urban Climatology

Background:

• Understanding temporal Land Surface Temperature (LST) patterns is vital for urban thermal environment studies.
• Existing tools lack integrated capabilities for extracting multi-timescale LST patterns from satellite time series data.

Purpose of the Study:

• To present a novel workflow for extracting multi-timescale temporal patterns and dynamics from nonlinear, non-stationary time series LST (TSLST) data.
• To apply and validate this workflow using MODIS TSLST data for Wuhan, China (2003-2017).

Main Methods:

• Generated a continuous, monthly TSLST dataset using non-parametric Multi-Task Gaussian Process Modeling (MTGP) from 8-day MODIS products.
• Segmented the study area into time series clusters using k-means for urban planning relevance.
• Decomposed TSLST data using Ensemble Empirical Mode Decomposition (EEMD) to reconstruct annual, interannual, and trend components.

Main Results:

• Annual components showed consistent rhythmic patterns across clusters, driven by Earth's revolution.
• Interannual components exhibited similar shapes but varying amplitudes, potentially linked to ENSO.
• Overall trends varied significantly, categorized into three types, likely influenced by localized urbanization and environmental management.

Conclusions:

• The developed workflow effectively extracts multi-timescale TSLST patterns, enhancing understanding of urban thermal dynamics.
• The findings highlight the impact of meteorological cycles and urbanization on LST variations.
• The workflow is adaptable for analyzing other cities and facilitating comparative urban environmental studies.