Using Rayleigh scattering over the ocean for satellite in-flight polarization calibration

Currently, an increasing number of spaceborne polarization sensors have been launched and are operating in space. However, there are very few satellite in-flight polarimetric calibration methods using natural scenes, compared to the numerous well-established vicarious methods for radiometric calibration. It's well-known that Rayleigh scattering is the primary contributor of the reflected radiative signal at the top of the atmosphere (TOA) over clear dark ocean at the short visible spectrum, whose degree of linear polarization (DOLP) ranges from 0 to 0.94 and can be accurately simulated using radiative transfer models. In this study, the potential of using Rayleigh scattering over the ocean for satellite in-flight polarization calibration is evaluated. Firstly, the achievable accuracy of simulated DOLP at TOA (<i>P</i>_cal) is evaluated, considering the typical uncertainties of input atmospheric and oceanic parameters. The <i>P</i>_cal error (<i>δ</i><i>P</i>cal<i>f</i><i>a</i><i>c</i><i>t</i><i>o</i><i>r</i>) in the visible spectrum (i.e., 490 nm and 670 nm) caused by various factors, including aerosol optical depth (AOD), aerosol model (AM), absorbing gases concentration (column water vapor, CWV; column ozone concentration, O₃), sea surface wind speed (WS), and chlorophyll concentration (Chl), is evaluated via radiative transfer simulation. The dependency of <i>δ</i><i>P</i>cal<i>f</i><i>a</i><i>c</i><i>t</i><i>o</i><i>r</i> on the wavelength and solar-viewing geometry is analyzed in detail. It's shown that the AOD and AM uncertainties are the primary error sources in Rayleigh scattering polarization calibration, followed by the WS, while the influences of Chl and absorbing gases can be ignored. The total DOLP error (<i>δ</i><i>P</i>calTotal) is mainly dominated by the scattering angle (SCA) and increases with the solar zenith angle (SZA) and viewing zenith angle (VZA) slightly. At the same time, <i>δ</i><i>P</i>calTotal increases rapidly with wavelength. Under an optimized solar-viewing geometry condition (i.e., SZA ≤ 40° and VZA ≤ 50° and the sun glint angle (SGA) ≥ 60° and 108° ≤ SCA ≤ 145°), the average <i>δ</i><i>P</i>calTotal at 490 nm and 670 nm are about 0.0088 and 0.0217, respectively. Then the simulated DOLP at the TOA of the Rayleigh scattering dominated region over the ocean was applied to calibrate the in-flight polarimetric measurements of the Directional Polarimetric Camera (DPC) onboard GaoFen-5(02). The standard errors (SE) of DOLP for the well-screened calibration samples are 0.0078 at 490 nm and 0.0213 at 670 nm, which is consistent with the theoretical error budgets. Furthermore, by comparing the calibration results of the six globally distributed ocean regions, all three indicators (i.e., SD, SE, and MAE) show much more stable polarimetric calibration results after the optimized solar-viewing geometry screening of the samples. It indicates that it is feasible and practicable to calibrate the satellites' short visible bands polarimetric measurements using Rayleigh scattering over the ocean under favorable geometric conditions. This study presents a method for satellite in-flight polarization calibration.