Airborne Hyperspectral LWIR Emissivity Data for Urban Surface-Material Characterisation: A Case Study in Dessau, Germany

Urban environments consist of highly heterogeneous surface materials whose physical and radiative properties influence surface energy exchange, heat storage, and urban microclimate conditions. While urban materials have been extensively investigated in the visible, near-infrared, and shortwave-infrared regions, their characterisation in the long-wave infrared (LWIR) domain remains comparatively limited, particularly using airborne hyperspectral imaging. This study assesses the potential of airborne hyperspectral LWIR emissivity data for urban surface-material characterisation and discrimination. Hyperspectral data were acquired over Dessau-Roßlau, Germany, using the Telops Hyper-Cam Airborne Mini. The radiance data were geometrically corrected, mosaicked, atmospherically compensated, and processed using temperature-emissivity separation to derive a 66-band emissivity mosaic. After MNF-based denoising, 11 material endmembers were extracted from a representative urban-industrial subset, including asphalt, concrete, cobblestone, bitumen roofing, metal surfaces, solar panels, storage silos, tree cover, and sparse grass. Selected airborne endmembers showed generally consistent spectral shapes when compared with independent ECOSTRESS and LUMA laboratory reference spectra. Linear and radial basis function Support Vector Machine classifications were then applied and evaluated using spatially independent validation regions. The linear SVM achieved the highest overall accuracy of 84.95%, balanced accuracy of 85.30%, macro-averaged F1 score of 82.34%, and a kappa coefficient of 0.83, while the RBF-SVM reached 81.19%, 83.06%, 77.68%, and 0.79, respectively. Solar panels, concrete, cobblestone, metallic materials, tree cover, and sparse grass showed high class-specific performance, whereas aged asphalt, bitumen roofing, and storage silos remained more difficult to distinguish. Overall, the results demonstrate the strong potential of airborne hyperspectral LWIR emissivity imaging for urban material discrimination under real-world conditions and support the development of dedicated LWIR spectral libraries for urban environments.