Spatial Downscaling Of Land Surface Temperature Retrieved From Remotely Sensed Thermal Infrared Imagery

Global climate change is one of the most severe challenges that confronting human beings in the21st century. The Earth’s climate system is experiencing a significant change characterized by global warming. Study of urban heat environment has become hot spots in a number of research areas, such as ecological, environmental and climatic research of cities, etc. Land surface temperature retrieved from remotely sensed thermal infrared bands can be used for urban heat environment study. However, the rather low spatial resolution of the remote sensing thermal imagery has become the bottleneck for its further application in practice. With the aid of spatial downscaling technique, different spectral bands with various spatial resolutions from the same remote sensor can be integrated and fully utilized, thus the spatial resolutions of the thermal bands can be improved. Therefore, study of thermal infrared imagery spatial downscaling techniques has a very important significance to urban heat environment research.In this thesis, a full comparison and assessment of three different algorithms, including TsHARP, EM and HUTS, for spatial downscaling of land surface temperature (LST) was made, not only from the qualitative but also from the quantitative (with multiple indices) perspective. Based on that, by introducing a new parameter, namely, land surface emissivity (ε), and by replacing the original NDVI with fractional vegetation cover (fc), an improved algorithm with better performance, MHUTS (Modified HUTS) was put forward. Three Landsat TM/ETM+imageries over the study area of Fuzhou, SE China, with similar summer season, acquired on June15,1989, May29,2003and June6,2009, respectively, were used in this study. The MHUTS algorithm was chose to downscale the LST imageries retrieved from the Landsat data to the spatial resolution of30m, and the spatial-temporal variations of the LSTs of the study area during the period from1989to2009, were explored and analyzed.The research results of this thesis will provide useful experiences and ideas for other similar research, especially in the way of algorithm improvement and assessment. Furthermore, by using Landsat5and Landsat7data to study the techniques of spatial downscaling of thermal infrared images and its utilization in urban heat environment research, this study will serve as good references for application of other newly launched remote sensing data (for example, Landsat8) in future.