In order to detect the tiny change of long-term global climate from satellite observations， the Earth Radiometry Benchmark Infrared Sounder is designed to provide the irrefutable benchmarking record by measuring the Earth-Atmospheric upwelling infrared radiance with an ultra-high accuracy of 0.1 K （

k

=2） over the 600–2700 cm

-1

（3.7–16.7 μm） spectral band. In the calibration model research， it was found that the combined polarization of the scene scanning mirror （SSM） and the instrument’s primary optics could modulate the incident radiation， ultimately resulting in a radiometric calibration deviation， which could be a significant radiometric uncertainty contributor. Thus

， it is necessary to investigate the mechanism and distribution pattern of the polarization-induced deviation， leading to the formulation of a robust correction scheme.The expression for the polarization-induced radiometric deviation was derived from the principles of polarized optics. To investigate the influence of various factors on the polarization-induced deviation， including instrument temperature difference， scene brightness temperature， spectral channel， scan angle

δ

， instrument polarization angle

α

and combined degree of polarization

p

r

p

t

， a simulation analysis of its distribution pattern was conducted based on the currently assumed scan geometry of the benchmark sounder. The results indicate that due to the difficulty in deploying temperature sensors on the scanning mirror surface and the challenge of establishing accurate temperature model， the temperature difference between the SSM and the calibration blackbody （BB） introduces additional deviations in the polarization-induced deviation calculation， with a maximum exceeding 1 K. To address this issue， an engineering solution utilizing a “SSM-BB” thermal equilibrium method to reduce this deviation is proposed. A rational simplification of the polarization deviation equation is then performed， followed by a corresponding radiometric uncertainty analysis. Moreover， when assessed in terms of brightness temperature （BT）， the non-linear conversion relationship between brightness temperature and radiance leads to a key characteristic： as radiance decreases， an identical amount of radiative modulation produces a larger brightness temperature deviation. Consequently， the polarization-induced deviation exhibits a strong spectral and scene temperature dependence. Specifically， the deviation increases with decreasing scene brightness temperature and with increasing wavenumber. In addition，

α

determines the phase offset of the polarization deviation across different scan angle

s. Notice that， when

α

is 0° or 90°， the deviation is symmetrically distributed about the nadir.To validate the effectiveness of the polarization correction model， a variable-temperature blackbody calibration experiment—which provides a reliable reference source with known truth through its precisely calculable spectral radiance and primary design for simulating on-orbit Earth scenes—was conducted based on the analysis of polarization-induced radiometric deviation patterns. The instrument polarization angle

α

differed across the long-wave infrared （LWIR）， mid-wave infrared （MWIR）， short-wave infrared （SWIR） bands due to the polarization basis state rotation introduced by the dichroic beam splitters， measured as 115.4°， 116.2°， and 91.8° respectively. And the variable-temperature blackbody temperatures

T

obs，BB

were set to 210 K， 230 K， 256 K， 287 K， and 310 K to cover the Earth's radiance brightness temperature range. The results indicate that polarization effects introduced a certain deviation trend across the three spectral bands. After applying the correction， the polarization-induced deviation was effectively reduced， with maximum corrections of 0.14 K， 0.4 K， and 3 K in the LWIR， MWIR， and SWIR， respectively.Following the validation of the correction model， a comprehensive on-orbit uncertainty analysis was performed according to the Guide to the Expression of Uncertainty in Measurement （GUM） to assess the impact of polarization-related parameters on the radiometric calibration accuracy. Key uncertainty sources included the BB temperature （

T

BB

）， SSM temperature （

T

SSM

），

α

， and

p

r

p

t

. Their individual contributions were quantified， and the combined standard uncertainty was calculated using the root sum of squares （RSS） method. The analysis covered three representative scene temperatures： 287 K， 256 K， and 210 K. A clear trend was observed as

the scene temperature decreased： the calibration uncertainty contributed by

T

BB

​ diminished， while those from

T

SSM​

，

α

and

p

r

p

t

increased significantly. For the 287 K scene， the uncertainty with polarization correction was dominated by

T

BB

， with a maximum combined uncertainty （

k

=2） of approximately 0.04 K. For the 256 K scene， while

T

BB

remained the primary contributor in the LWIR and MWIR， the influence of

T

SSM

and

α

became noticeable in the SWIR， with the total uncertainty remaining below 0.1 K across most channels. For the critical 210 K low-temperature scene， polarization-related parameters emerged as the dominant uncertainty source， significantly exceeding the contribution from

T

BB

. The uncertainties contributed by

T

SSM

，

α

， and

p

r

p

t

were most significant， yet the combined uncertainty for spectral channels below 1700 cm⁻¹ was lower than the 0.1 K threshold.In conclusion， this study has demonstrated the effectiveness of the polarization deviation correction model， with the thermal equilibrium design playing a critical role in diminishing additional calibration deviation caused by the “SSM-BB” temperature difference. Following the future successful launch of the Earth Radiometry Benchmark Infrared Sounder， it will be further applied to the polarization-induced deviation correction of actual atmospheric observation data.