The urban thermal climate of Singapore
Chang, Chew Hung.
Date of Issue1998
The urban heat island phenomenon was examined at three spatial scales in this study. At the island-wide scale, the existence of the nocturnal heat island phenomenon at 2200 hrs for various "seasons" of the year in Singapore was confirmed. The study was undertaken at 2200 hrs to allow for comparison with an earlier study in 1981. The intensity of the phenomenon varies both spatially and temporally. The nocturnal phenomenon has changed since it was last studied in 1981. Secondary heat islands are now present over some sub-urban areas which were previously rural. Previous studies of Singapore's nocturnal heat island have recognised and described the phenomenon substantially but have failed to establish statistical relationship between factors and the phenomenon. This study established some statistical correlation between some factors (such as urban canyon geometry and size of estate) and the nocturnal heat island phenomenon through empirical data examined at local scales. Urban canyon geometry, in terms of height-to-width ratios, and the physical size of the housing estates have positive correlations but the areal extent of waterbodies has negative correlations with heat island intensities. The role of canyon geometry in the processes involved in the genesis of the phenomenon was further examined at a micro-scale, where three canyons were studied for spatial and temporal variation in ambient temperatures, insolation and wind speed. Results show that canyons with higher height-to-width ratios tend to have weaker winds at the canyon floor and large spatial difference in ambient air temperatures between canyon floor and canyon top. Using the statistical results, a model to predict 2200 hrs heat island intensities based on urban canyon geometry, size of estate and areal extent of waterbodies was proposed. This model gives an overall p-value of 0.0002 with urban canyon geometry being the most statistically significant of the three parameters included in the model.
DRNTU::Social sciences::Geography::Environmental sciences