Spectral unmixing for thermal infrared multi-spectral airborne imagery over urban environments: day and night synergy

Carlos Granero-Belinchon (1) , Aurélie Michel (1) , Véronique Achard (1) , Xavier Briottet (1)

1 DOTA, ONERA, Université de Toulouse [Toulouse]

This article aims to show that TRUST thermal unmixing, providing intra-pixel information 1 of material abundances and their temperatures, allows a better description of the urban thermography, 2 which not only supplies land surface temperatures at better resolutions but also allows to link them to 3 specific materials. Moreover, this article presents an improved version of TRUST, called TRUST-DNS 4 (Day and Night Synergy), which takes advantage of daytime and nighttime acquisitions to improve 5 the unmixing performances. Both TRUST and TRUST-DNS are applied on airborne thermal images 6 of Madrid city center obtained during the DESIREX campaign 2008, initially at 4m resolution and 7 undersampled to 8m resolution. This undersampling allows to use the 4m ground classification done 8 during the DESIREX campaign as reference for the abundance retrieval study. Thus, this article aims at 9 showing the applicability of TRUST thermal unmixing on a highly challenging study case: the analysis 10 of a heterogeneous urban environment with airborne multispectral (8 thermal bands) images at 8m 11 resolution. In this study, seven different endmembers (natural and manmade materials) are considered as 12 composing the Madrid city center, and TRUST and TRUST-DNS are used to recover intra-pixel abundance 13 and temperature maps at 8m resolution for each endmember.

Mots clés: Thermal unmixing; TRUST; Urban environment; Airborne remote sensing; LST 15; thermal unmixing; urban environment; airborne remote sensing; LST

Domaines: Sciences de l'ingénieur [physics]; Physique [physics]

Carlos Granero-Belinchon, Aurélie Michel, Véronique Achard, Xavier Briottet. Spectral unmixing for thermal infrared multi-spectral airborne imagery over urban environments: day and night synergy. Remote Sensing, 2020, 12 (11), pp.1871. ⟨10.3390/rs12111871⟩. ⟨hal-02776492v2⟩