[Study on the modeling of earth-atmosphere coupling over rugged scenes for hyperspectral remote sensing]

Adjacency effects may introduce errors in the quantitative applications of hyperspectral remote sensing, of which the significant item is the earth-atmosphere coupling radiance. However, the surrounding relief and shadow induce strong changes in hyperspectral images acquired from rugged terrain, which is not accurate to describe the spectral characteristics. Furthermore, the radiative coupling process between the earth and the atmosphere is more complex over the rugged scenes. In order to meet the requirements of real-time processing in data simulation, an equivalent reflectance of background was developed by taking into account the topography and the geometry between surroundings and targets based on the radiative transfer process. The contributions of the coupling to the signal at sensor level were then evaluated. This approach was integrated to the sensor-level radiance simulation model and then validated through simulating a set of actual radiance data. The results show that the visual effect of simulated images is consistent with that of observed images. It was also shown that the spectral similarity is improved over rugged scenes. In addition, the model precision is maintained at the same level over flat scenes.