Sensors (Dec 2020)
A Systematic Sensitivity Study on Surface Pixel Shifts in High Spatial Resolution Satellite Images Resulting from Atmospheric Refraction in the Sensor to Surface Ray Path
Abstract
When viewing Earth’s surfaces from a low Earth orbiting (LEO) satellite platform with an optical sensor, the upward light propagation path from the ground to the satellite is affected by atmospheric refraction. For imaging sensors with a spatial resolution of about one km on the ground, atmospheric refraction is typically neglected during geo-registration of the satellite images. However, for high spatial resolution imaging systems with surface pixel sizes of approximately one meter or finer, the neglect of atmospheric refraction effects can typically introduce errors of a few meters in the spatially registered images. The atmospheric refraction effects need to be properly taken into consideration during the spatial registration of high spatial resolution satellite images. We have found that, with minor modifications, the ray tracing models implemented inside the LOWTRAN series of atmospheric radiative transfer codes developed in the 1970s and 1980s, in particular LOWTRAN7 in late 1980s, can be used for modeling the pixel displacement resulting from atmospheric refraction for satellite observations. The LOWTRAN series models were originally designed for calculating atmospheric transmittances and radiances for radiation going through long paths of the Earth’s atmosphere. In the ray tracing portions of the codes, a spherical model atmosphere from the ground to 100 km is finely divided into about 30 thin atmospheric layers. The refraction angles for ray paths between consecutive layer boundaries are accurately calculated. We make a new use of the refraction angles calculated by the LOWTRAN7 code to study the surface pixel shift resulting from atmospheric refraction for satellite observations. In this letter, we report the modeling results on surface pixel displacements for different satellite altitudes and downward view zenith angles, several atmospheric temperature and pressure profiles, a few surface elevations, and wavelength dependencies from blue (450 nm) to near-IR (865 nm). These results can have reference values for researchers to estimate refraction-induced pixel displacements in their high spatial resolution satellite images. The results may also potentially help in designing spacecraft algorithms for accurate instrument pointing and mission tasking to automatically capture short-lived science events.
Keywords
