Emissivity image generation of Gaofen-5's thermal infrared bands using ASTER data

摘要

To explore the application of Gaofen-5's thermal infrared (TIR) data, algorithm development and validations needs to be conducted before its launch; and image simulation is necessary to provide datasets for such research. In this paper, using ASTER imagery as the data source, a simulation method is proposed for Gaofen-5's four TIR bands' emissivity images. For the first 3 bands, emissivity images are generated through translation models built on spectral correlation between emissivity of Gaofen-5 bands and those of ASTER bands. For the 4^th TIR band of Gaofen-5, we propose a spectral matching and spectral mixing method for emissivity image generation. To test the feasibility of our simulation method, ASTER thermal infrared emissivities are simulated and then compared with their actual values. Average relative errors of these simulated emissivities are 4.04%, 2.98%, 1.74%, -4.02% and -2.79% in 5 ASTER TIR bands, indicating that our proposed method can be used for emissivity image generation.