Spatio-temporal Modeling and Analysis of Urban Heat Islands in China

The urban climate has changed much more dramatically than the global climate change over the past decades, with the urban heat island (UHI) as the typical urban climatic feature. As the UHI has deep impact on the local/regional climate and the lives in cities, UHI studies are of great importance for both the climate change research and the urban planning. Although great progresses have been achieved, the UHI studies are still of great challenge. The complex urban systems accompanied by the intense human activities make the patterns and impact factors of UHI varied spatially and temporally. Rapid urbanization has occurred in China over the past decades and it will be accelerated in future considering that China has high speed of development and a large population at present. Therefore, we chose China as the study area and hope to provide a systematic research framework for an in-depth comprehensive understanding about the variations and associated factors of UHI, and finally to enrich our knowledge on the UHI and provide references to mitigate UHI effect in China.;This dissertation developed a systematic research framework following the outline of "retrieving land surface temperature (LST), describing the spatio-temporal patterns, revealing the associated factors, and simulating the response to urban expansion". The multisource datasets were employed, including basic geographical information system (GIS) datasets, remote sensing (RS) images and statistical data. Firstly, a spatio-temporal image fusion model was developed to generate the Landsat-like LST for intra-city surface UHI (SUHI) studies. With the support of the available Landsat data and predicted Landsat-like data, spatial statistical models were used to explore the seasonal variations in the relationships between LST and the normalized differences vegetation index (NDVI) in Beijing city. Secondly, a modified regression tree model linked to R language was employed to assess the associated factors of SUHI across 67 major Chinese cities. Meanwhile, the statistical methods were used to comparatively analyze the spatio-temporal variations in SUHI in China. Finally, the coupled "Weather Research and Forecasting/Urban Canopy Model (WRF/UCM)" numerical modeling system was adopted to simulate the response of UHI to the future urban expansion over the Beijing-Tianjin-Hebei (BTH) metropolitan area. The improvements of methodologies and main findings in this research are summarized below.;1) A spatio-temporal image fusion model based on bilateral filtering was proposed to produce high spatio-temporal Landsat-like LST. Its accuracy was higher than the widely used "Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM)" algorithm because of its advantage accounting for both the warming and cooling effects of surface features in the urban environment. Based on the retrieved LST, the seasonally varied relationships between LST and NDVI in Beijing city were explored. It is suggested that the seasonal variations should be considered when concluding the contribution of vegetation to SUHI.;2) An improved regression tree model was introduced to assess the specific associated factors of SUHI based on the social, economic and natural factors by taking the urban as the complex ecological system. The results indicated that the SUHI varied significantly in character between northern and southern China. Southern China experienced more intense and stable daytime SUHI intensity (SUHII) than northern China, while the patterns were opposite for nighttime SUHII. Vegetation was found to be the most important explanatory factor in northern China implying that increasing vegetation coverage may mitigate daytime SUHII effectively. In southern China, annual electricity consumption and the number of public buses were found to be important. These results have important theoretical significance and may also be of use to decision-makers aiming to mitigate UHI effects.;3) The coupled WRF/UCM modeling system was presented to simulate the response of UHI to future urban expansion taking the BTH metropolitan area as the study area. The results indicated that the coupled models embedded with fine spatial resolution land use/land cover (LULC) data can well simulate the response of UHI to the urban expansion. The urban expansion has significant impact on the temperature, especially on the minimum temperature, which was raised by about 5 K over the newly urbanized area. The temperature in old urban areas was also increased a little (less than 1 K). It indicated the more uncomfortable urban environment in the future, especially during the nighttime when the minimum temperature changes are larger due to the urban expansion.