UBC Theses and Dissertations
The effect of preferential view direction on measured urban surface temperature Adderley, Christopher David
This study aimed to determine the complete surface temperature T0,C of an urban system and quantify the bias of sensors with preferential views (satellite, hemispherical) on sensed urban surface temperature. A thermal camera emplaced on a tower in a street canyon in Vancouver, Canada recorded panoramas of surface temperature. Techniques from micrometeorology, and computer vision/graphics were combined to project this data onto a 3D model of the urban structure surrounding the tower, which was derived from photogrammetry and airborne LiDaR data, then correct sensed brightness temperatures for atmospheric and emissivity effects. Facets of the 3D model not seen from the tower were gap filled statistically with data from other areas. Roofs were the warmest in the daytime and coolest at night, while the ground followed an opposite trend. Wall surface temperatures fell in-between with variation dependent on orientation and material. T0,C followed the trends in Voogt & Oke (1997), lower than roof and ground surface temperatures but higher than wall surface temperatures during the daytime, with an opposite trend at night. A narrow field of view sensor and a hemispherical radiometer were simulated at varying locations in and above the canyon. Recovered simulated brightness temperatures showed that T0,C was exceeded during the daytime by simulated at-nadir brightness temperatures TB,nad by up to 2.2 K and undervalued by up to 1.6 K at night due to emissivity and anisotropic effects (-0.7 K and +2.9 K respectively). This difference is due to anisotropic effects which were up to 3.5 K for this urban structure, following street canyon orientation. Simulations of radiometers with hemispherical views showed that the horizontal placement of a radiometer measuring L↑ in and above a street canyon causes the recovered flux to vary by up to 120 W m-2. Horizontal variation was reduced when the radiometer is placed at sufficient height above the ground surface (5 times building height), where simulated L↑ measurements converge to a single value. Surface temperatures determined from L↑ are in approximate agreement with T0,C but showed an RMSE of 1.8 K which is larger than the RMSE between T0,C and TB,nad (1.4 K).
Item Citations and Data
Attribution-NonCommercial-NoDerivatives 4.0 International