The rapid urbanization process has greatly changed the spatial pattern of cities, especially artificial facilities. As an important municipal infrastructure, urban road systems occupy more than 1/4 of the built-up area. It supports an open space for material flow, energy flow and information flow, and its developing speed and distribution are far ahead of other urban infrastructure. Urban roads are also a kind of typical impervious surface. Compared with natural green space, the building materials like cement and asphalt make urban road system a different thermal characteristic, and in terms of spatial distribution, those traits may eventually result in the significant impact on the pattern of urban thermal environment. Many studies involved in urban underlying surfaces and their relationship with urban heat island (UHI), while the relationship between the spatial structure of urban roads and UHI has not been reported. More information about the interactive impact between the spatial structure of urban roads and urban environment needed to be quantitatively assessed during the process of rapid urbanization, which could provide some scientific guidance for urban infrastructure planning and management toward a sound and eco-friendly development.
In this paper, through field survey (by using Laser distance meters and GPS to collect valid information from different sections of 238 roads within the study area) and high resolution remote sensing data (Quick Bird data source) recognition based on the object-oriented method, a spatial database of urban roads within 5^th ring road Beijing was established, which included the length, width and intersection nodes of different classes of urban roads. A thermal infrared inversion of Landsat TM was conducted for retrieving LST and NDVI,which are the two common indicators to characterize thermal environment and vegetation information. Then a correlation analysis between the road system characteristics and LST and NDVI was carried out.
The results showed that there was a significant correlation among parameters of road system, LST_mean and NDVI_mean in larger grids (4×4). Meanwhile, LST_min and NDVI_min were related with parameters of road net in moderate grids (8×8, 16×16). In terms of small grid (32×32), besides the mean and maximum value of LST and NDVI, the correlations were observed between certain road indicators and LST_max and NDVI_max;
The total length of roads is the best variable described the change of LST and NDVI among all the parameters. When the number of nodes was weighted for area, the correlation coefficients with NDVI in 8×8 and 32×32 grids were increased. At all levels of analysis grids, all four road indicators presented a significantly negative correlation with NDVI_mean, which implied that the area of urban road green space was too small to play an important roles on the intensity and spatial distribution of NDVI. The factual of linkages among various spatial attributes of urban road networks to the intensity, quantity, and distribution of LST and NDVI may help develop a new approach to quickly identify the overall UHI spatial patterns and further study the relations between urban thermal environments and human activities.