Downscaling land surface temperature with high-resolution image: a difference mean distribution (DMD) post-processing method

ABSTRACT

Land surface temperature (LST) inversion using remote sensing is a crucial tool to study the land surface thermal environment. To improve the resolution of the inverted LST, a downscaling method is used but it often reduces the accuracy. Here, a difference mean distribution (DMD) method was developed to post-process the downscaled LST. Landsat 8 and Sentinel 2 images were used to test and Field measured LSTs in Tianjin, China were used to verify the method. Results show DMD method reduces MAE and RMSE by 1.36°C and 1.50°C, respectively indicating that it can improve the accuracy of the downscaled LST.

KEYWORDS:

• Land surface temperature (LST)
• downscale
• difference mean distribution (DMD)
• gradient
• accuracy

Disclosure statement

No potential conflict of interest was reported by the author(s).

Author contributions

All of the authors contributed extensively to the work. Jinlong Zang participated in the method construction and wrote the paper. Qiaozhen Guo proposed key ideas and a large number of paper revision work. Xiaoxu Wu analyzed the result and helped complete the modification of the language. Other authors participated in the collection and processing of the data.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China, grant number [No.41971310] and the Scientific Research Project of Tianjin Municipal Education Commission, China, grant number [No. 2020KJ051].