Impact assessment of land cover, urban morphology, and roofing materials on surface urban heat island estimation using multi-temporal landsat imagery in Tegal City, Indonesia
Files
Farania_Azrina_05682500_2026.pdf
Open access - Adobe PDF
- 7.95 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- This study evaluated the influence of land cover composition, Local Climate Zone (LCZ), and roofing material properties on the spatial distribution of land surface temperature and surface urban heat island (SUHI) pattern in a tropical coastal city. The research was conducted in Tegal City, Indonesia, using multi-temporal Landsat 8/9 thermal imagery across six heatwave dates (September 2023 to October 2024). The PSF convolution approach was used to homogenize land cover and LCZ composition at the pixel level by aligning them with the thermal resolution of the Landsat thermal sensors prior to regression modeling. Urban surfaces were derived from Tegal City Government’s land cover data (0.5m resolution) updated with Sentinel-2 spectral indices. Morphological analysis was performed by classifying land cover into LCZs based on form (e.g., compact, open, or vegetation) using rules, and by classifying urban surfaces based on their thermal characteristics using a Normalized Brightness Index approach. Land cover and LCZ-based models were constructed using linear regression and linear mixed-effect modeling, and the thermal effects of each surface class were determined and compared. The land cover model achieved higher predictive performance (R² = 0.84, RMSE = 1.3459 °C) than the LCZ model (R² = 0.79, RMSE = 1.5309 °C), suggesting that fine-resolution land cover composition captures thermal variability more precisely than urban morphological classification alone. In land cover models, strong warming effects were observed from railway networks (+10.63 °C) and built-up area (+8.75 °C). Broadleaved trees (-7.11 °C) and fish ponds with water (-6.05 °C) yielded significant cooling. For LCZ models, stronger warming was recorded from urban LCZ 2-Compact medium-rise (+11.32 °C) and LCZ E-Bare rock or paved (+4.76 °C), with LCZ A-Dense trees (-8.10 °C) and LCZ G-Water (-5.58 °C) as the strongest coolers. Among urban morphological parameters, Pervious Surface Fraction (PvSF) showed the strongest correlation with LST anomaly (r = -0.647). Roofing material analysis revealed that tile/clay roofing was associated with higher LST anomalies, while metal roofing showed a cooling effect. A statistically significant but weak coastal cooling gradient was identified to have a negative correlation. Based on findings, practical recommendations can be provided for urban mitigation: to invest in cool roofing and minimum vegetation, limit building densities, and protect coastal areas. The methodology employed can also be used in other tropical coastal cities for heat-resilient urban planning.