Overcoming cloud-induced gaps in surface urban Heat Islands intensity monitoring using annual temperature cycle modeling across Taiwanese cities

Thermal infrared remote sensing offers a promising technique to measure Surface Urban Heat Island Intensity (SUHII) induced by urbanization. However, a significant challenge lies in obtaining spatially continuous Land Surface Temperature (LST) data due to cloud contamination. This study aims to reconstruct continuous SUHII time series and examine its temporal dynamics in response to urbanization across five major Taiwanese cities (Taipei, Taoyuan, Taichung, Tainan, and Kaohsiung) from 2003 to 2020. We applied the Annual Temperature Cycle (ATC) model using MODIS LST data, along with surface features such as impervious surface area (ISA), Enhanced Vegetation Index (EVI), and Normalized Difference Latent Heat Index (NDLI). Urban-rural zones were defined using built-up intensity gradients, and the Mann-Kendall method was used to assess interannual trends. Correlations between ATC parameters (mean, amplitude, phase shift) and surface features (ISA, EVI, NDLI) were analyzed with climatically averaged SUHII. Results showed strong correlations between ATC-derived parameters and observed clear-sky SUHII, with a significant increasing trend across all urban zones and cities. Mean SUHII was strongly correlated with ISA ( r = 0.990–0.999) across urban-rural gradients and years, while it was negatively correlated with EVI and NDLI. Daytime SUHII was higher than nighttime and peaked in summer. Furthermore, Taipei and Taichung have higher SUHII dynamics relative to the other cities. These findings offer insights into the dynamics of urbanization-driven SUHII, establishing a basis for predictive modeling and heat mitigation strategies in subtropical urban areas where cloud cover is a significant issue. • We assessed SUHII dynamics for five Taiwanese cities using the ATC model. • More significant increases in SUHI were observed during the day than at night. • We found the changing rate of S M increased from the exurban to the urban core. • S M is significantly and positively correlated with ISA across years and UDI gradient.