A Multimodal Deep Learning Approach for High-Resolution Land Surface Temperature Estimation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

A Multimodal Deep Learning Approach for High-Resolution Land Surface Temperature Estimation

Résumé

Urban Heat Islands (UHI), characterized by elevated temperatures, present important challenges to sustainability. This study introduces a novel multimodal approach for high-resolution Land Surface Temperature (LST) estimation, a critical component in addressing UHI. The methodology initially employs RGB orthophotography for LST estimation and progressively integrates additional relevant variables correlated with LST, including elevation and land cover. Leveraging conditional Generative Adversarial Networks (cGANs), LST maps are generated, enabling informed urban planning. Experimental results highlight the potential of this multimodal approach, emphasizing that the combination of all data variables yields the most favorable outcomes. These findings advance UHI research and support data-driven urban climate management.
Fichier principal
Vignette du fichier
Article_1570959384.pdf (760.45 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04460000 , version 1 (15-02-2024)

Identifiants

  • HAL Id : hal-04460000 , version 1

Citer

Issam Khedher, Jean-Marie Favreau, Serge Miguet, Gilles Gesquière. A Multimodal Deep Learning Approach for High-Resolution Land Surface Temperature Estimation. 7th International Symposium on Signal Processing and Intelligent Recognition Systems (SIRS’23), PES University, Dec 2023, Bangalore, India. ⟨hal-04460000⟩
62 Consultations
31 Téléchargements

Partager

Gmail Facebook X LinkedIn More