Using a Modified HUTS Algorithm to Downscale Land Surface Temperature Retrieved from Landsat-8 Imagery: A Case Study of Xiamen City, China

Urban heat environment has a close relation with quality of life of urban citizens. Land surface temperature (LST) retrieved from remotely sensed thermal infrared imagery can be used for urban heat environment studies. However, the rather low spatial resolution of currently available remote sensing thermal imagery (e.g., Landsat ETM+, its thermal band is in the spectral range of 10.40-12.51 mu m and at 60 m spatial resolution) has become the bottleneck for its further application in the real word. By using the spatial downscaling techniques, different spectral bands with various spatial resolution from the same remote sensor can be integrated and fully utilized, thus the improvement of the spatial resolution of the thermal imagery can be realized. So far, various algorithms or methods for spatial downscaling of remotely sensed thermal images have been put forward. Among them, the HUTS (High-resolution urban thermal sharpener) algorithm, proposed by Dominguez et al. (2011), has been accepted and used by many researchers. However, its applicability in more study areas with more remotely sensed data, especially recently launched data (like Landsat-8 data), still need to be further examined. In this paper, taking Xiamen City, China as the study area, the original HUTS method was modified and improved by taking the following two measures: (1) introducing a new parameter, namely, land surface emissivity; (2) replacing the original scaling factor NDVI with fractional vegetation cover. A Landsat-8 (path 119, row 43) imagery, acquired on August 4, 2013, was used and the revised Generalized Single-channel Method proposed by Jimenez-Munoz and Sobrino (2014) was applied to retrieve LST in this study. A full assessment and comparison of the original and the modified HUTS algorithms were made, both from the qualitative and the quantitative (by using multiple indices, such as R, RMSE, RMSPE, RASE, and so on) perspectives. The research results showed that, the modified HUTS algorithm, presented in this study, had better performance than the original one, thus was proposed as a useful technique of spatial downscaling of thermal infrared image in future urban heat environment studies.