Abstract
This paper proposes a method to estimate the urban surface heat fluxes using Landsat TM data in conjunction with routine data from meteorological observatory and direct field measurement data. The sensible and latent heat fluxes are evaluated independently using bulk transfer formulae. The ground cover classification, surface albedo, surface temperature and the normalized difference vegetation index (NDVI) are deduced from two Landsat TM images on Aug. 6, 1990 and Dec. 6, 1988. A field measurement is carried out to obtain the relationship between the moisture availability and relative humidity for no-vegetation ground using band-pass covariance technique to measure the turbulent sensible and latent heat fluxes over concrete surface.
The distribution of each pixel and mean value for each kind of ground cover show that the net radiation in urban area has a relatively conservative behavior with little spatial variability. Conversely, the sensible and latent heat fluxes have clear spatial variability. The high sensible heat flux corresponds to high building (urban center) and residential area, while the relatively low sensible heat flux corresponds to forest area in parks and open water. The latent heat flux is very low in high building and wooden residential area (mean values less than 37Wm-2 in summer and 1Wm-2 in winter). But in residential area with some vegetation and forest area, latent heat flux increases to 77 and 194Wm-2, and the mean Bowen ratio decreases to 4.6 and 1.6 in summer, respectively. Since the amount and activity of vegetation is higher in summer, there is a greater spatial variability of heat fluxes in summer than in winter.
