Retrieval of Rugged Mountainous Areas Land Surface Temperature From High-Spatial-Resolution Thermal Infrared Remote Sensing Data

• Published in: IEEE transactions on geoscience and remote sensing Vol. 61; pp. 1 - 16
• Main Authors: He, Zhi-Wei, Tang, Bo-Hui
• Format: Journal Article
• Language: English
• Published: New York IEEE 2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adjacent effect; Algorithms; Atmosphere; Atmospheric modeling; Atmospheric water; Brightness temperature; Canopies; Canopy; Climate change; Data analysis; Emissivity; Infrared imagery; Land surface; Land surface temperature; Landsat; Landsat-9 data; Mathematical models; Mountain regions; Mountainous areas; mountainous surface temperature; Mountains; Parameters; Pixels; Plant cover; Radiance; Radiation; Radiative transfer; Remote sensing; Retrieval; Root-mean-square errors; Rough surfaces; Satellite imagery; Scattering; Spatial distribution; Surface radiation temperature; Surface roughness; Surface temperature; Surface topography; thermal infrared (TIR) remote sensing; Thermal radiation; topographic effect; Topographic effects; Water vapor; Water vapour
• DOI: 10.1109/TGRS.2023.3316624
• ISSN: 0196-2892

Mountainous land surface temperature (MLST) is one of the key surface feature parameters in studying mountainous climate change. However, for the high-spatial-resolution thermal infrared (TIR) remote sensing images, the current land surface temperature (LST) retrieval algorithms were developed without enough accounting for terrain geometry and adjacent effect, which is not suitable for retrieving LST over rugged mountainous surfaces. To overcome this problem, a new method was developed to estimate the small-scale self-heating parameter (SSP) of mountainous pixels to quantify the proportion of internal thermal radiation intercepted, and the mountainous canopy effective land surface emissivity (MLSE) was defined and modeled based on SSP. A novel mountainous canopy multiple scattering TIR radiative transfer (MMS-TIR-RT) model based on SSP and sky-view factor (SVF) was developed to eliminate thermal radiance contribution from inside/adjacent pixels and the atmosphere and to restore the thermal radiation characteristics of pixels themselves. Based on this model, a new framework of mountainous single-channel (MSC) algorithm was developed for MLST retrieval from TIR data of Landsat-9 TIRS-2 sensor. In accordance with simulated data analysis, SSP, SVF, atmospheric water vapor content (WVC), land surface emissivity (LSE) of target pixel, and mean LST and LSE of the proximity pixels are the main influence factors on the magnitude of the topographic effect and adjacent effect (T-A effect). Among them, SSP plays a decisive role in the mountainous canopy effective emissivity when the emissivity of the original material is low. The retrieval LST differences (<inline-formula> <tex-math notation="LaTeX">\delta </tex-math></inline-formula>LST) between the MSC algorithm and conventional single-channel (SC) algorithm (without considering the T-A effect) from Landsat-9 TIR images are related to SSP and SVF. The results showed that the inversion of LST can be overestimated by up to 4.5 K without considering the T-A effect correction in the rugged mountainous areas. Comparing brightness temperature (BT) at the top of atmosphere (TOA) simulated by the discrete anisotropic radiative transfer (DART) model and TOA BT from radiance of TIRS-2 band 10, there are good consistencies between the spatial distributions at the three subregions, with the root-mean-squared error (RMSE) less than 0.62 K. The findings demonstrated the need for T-A effect to be considered in the retrieval of high-precision MLST.