Study of urban geometric effects on urban surface temperature retrieval and energy flux modelling : an application in Hong Kong

Abstract

Remote sensing data are widely used in urban heat island research. Urban geometry poses a great challenge for Urban Surface Temperature (UST) retrieval, however. UST is an important parameter for Surface Urban Heat Island (SUHI) research and also a crucial parameter to estimate radiative and convective fluxes in urban areas. However, most current methods to retrieve UST do not consider geometric effects, probably due to a lack of accurate regional data on urban geometry. In this study, a method for UST retrieval by considering geometric effects was developed and the impacts of USTs derived with and without considering geometric effects on estimated urban energy fluxes were analyzed. Firstly, an Improved Urban Effective Emissivity Model based on the Sky View Factor (IUEM-SVF) was developed. The IUEM-SVF was then evaluated by numerical experiments with a numerical model of urban micro-climate, i.e. the Temperatures of Urban Facets in 3-D model, (TUF-3D). Results from the comparison with TUF-3D show that the IUEM-SVF performs well under conditions of uniform and isothermal urban targets, i.e. the correlation coefficients of effective emissivity are 0.99 and the root-mean-square deviations (RMSDs) are less than 0.005 when the component facets are isothermal, and material emissivity is larger than 0.80. Thermal and material heterogeneity affects the accuracy of the IUEM-SVF. When effective emissivities in the ASTER AST_05 Band 13 products on March 13th 2013 and August 4th 2013 were used to evaluate the effective emissivity derived from IUEM-SVF, high correlations between ASTER emissivities and emissivities modelled by the IUEM-SVF were observed (r² = 0.97 for 13th March 2013 and 0.83 for 4th August 2013). Secondly, an urban exitance model based on the single channel method (UEM-SCM) was developed with IUEM-SVF. Results show that the temperature bias caused by geometric effects can reach 2 K in summer and 3 K in winter over built-up areas. The Split-Window (SW) algorithm was also used to retrieve UST and the results show that IUEM-SVF can be used to distinguish geometric and atmospheric effects when applying the split-window algorithm. When the Temperature and Emissivity Separation (TES) algorithm is used for UST and emissivity retrieval, the relationship between minimum emissivity and maximum-minimum difference can be reconstructed by considering urban geometric effects. However, multiple spectral methods (e.g. SW and TES) are sensitive to the thermal and material heterogeneity for UST retrieval over urban areas. Lastly, geometric effects on the urban albedo and net radiation were analyzed based on the TUF-3D and the findings show that geometric effects may cause 34.76% bias in estimated net radiation. The different USTs (with respectively without considering geometric effects) cause 9.8% to 23.5% differences in estimated sensible heat fluxes. In view of the complex characteristics of the urban landscape and urban geometric effects, methods developed in this study for UST retrieval and urban energy flux estimation can be applied over urban areas to enable better estimation and monitoring of urban heat island effects.