Emissivity image simulation for thermal infrared bands on Gaofen-5 using airborne hyperspectral data

摘要

Gaofen-5 (GF-5) satellite is designed to provide high resolution imagery for land observation and disaster monitoring and will be launched in 2017. As true GF-5's data is unavailable, image simulation can be a useful tool to provide test images for research on relevant key techniques. In this paper, an emissivity image simulation method is proposed for 4 thermal infrared (TIR) bands on one of GF-5's instruments. In this method, thermal infrared images of airborne hyperspectral sensors are chosen as data source. For bands TIR1-3, their emissivity images are simulated through spectral integration based on hyperspectral emissivity images. For band TIR4, abundance inversion is conducted on emissivity images of the data source with surface endmembers, and emissivity images of band TIR4 are finally generated through spectral mixing. As the data source has a higher spatial and spectral resolution than the sensor on GF-5, this method can reproduce emissivity images with good accuracy.