Principal component based fusion of land surface temperature (LST) and panchromatic (PAN) images

The spatial details of panchromatic (PAN) images are always higher than land surface temperature (LST) images. The main aim of this paper is to develop a fusion technique for PAN and LST images of the LANDSAT8 satellite. The key is to appropriately estimate the spatial details of the PAN images while preserving the LST image’s thermal contents. The existing methods are incapable to fuse the thermal details of LST images while fully considering the PAN image’s structure, resulting in inaccurate LST estimation and spectral distortion. Principal components (PC) of PAN–LST images can efficiently transfer the spatial details of the PAN image in the spectral information of the LST image. In this paper, a novel fusion algorithm has been proposed named as “intensity transformation fusion model” (ITFM), to downscale LST images using the PC1–PC4. The results have shown that the root mean square error of PAN fused LST images were minimum for PC1 (0.63 °C) and maximum for PC4 (1.04 °C), respectively. The proposed ITFM method has enhanced spatial resolution and visual distinctiveness of LST images as well as precisely preserved the LST data. The fusion algorithm would help in studies related to the detection of land cover’s thermal emissions, thermal comfort monitoring, urban heat island effect analysis, and LST downscaling applications.