Modelling Spatial and Temporal Road Thermal Climatology in Rural and Urban Areas Using a GIS

A 1-D road surface energy balance model was modified to account for the geographical variables of latitude, optical depth, sky-view factor, slope and slope orientation for the West Midlands (UK). The physical variables of albedo, emissivity and surface roughness are also included. Using a satellite land cover classification, aided by a field analysis of urban canyon characteristics, it was possible to estimate the spatial variation of surface variables across the West Midlands. Spatial analysis of the topography was achieved using a Geographical Information System (GIS) database which calculated values at 1 km(2) resolution for the geographical variables. This enabled a spatial and temporal analysis of road surface temperatures (retrospectively) across the West Midlands. Sensitivity analysis shows that the geographical variables which have the most significant influence on the model are slope angle and sky-view factor. Validation of the model (West Midlands grid model, WMG) against actual road surface temperature for 15 road weather sensors distributed around the West Midlands, for February 2000, gave R-2 values as high as 0.84; however regression indicated that for 79 nights in the period December 1999 to February 2000 the model overestimated the minimum road surface temperature with a bias of 0.65degreesC (RMSE 2.07degreesC), as opposed to the Met Office model (MOM), which underestimated with a bias of -2.03degreesC (RMSE 3.09degreesC). Time slices of the model output, covering an area of 2400 km(2), show the development of a surface urban heat island in the West Midlands. The intensity of the modelled heat island is sensitive to the values used for the sky-view factor in the rural areas surrounding the urban conurbation. Winter solstice heat island intensity for calm clear nights in the West Midlands is calculated to be 4.7degreesC. The structure of the heat island suggests that current Open Road weather forecast zones are not applicable in simulated clear calm conditions because of the wide range of road surface temperatures caused by the degree of urbanisation.