Industrial chimneys, ship exhaust, and volcanic eruption processes can emit large amounts of harmful SO₂ into the atmosphere, causing serious pollution to the environment. The development of effective SO₂ monitoring tools can provide a strong guarantee for atmospheric environmental management. In recent years, SO₂ ultraviolet (UV) cameras have been rapidly developed and widely applied by virtue of their high spatio-temporal resolution, high detection sensitivity, and two-dimensional detection imaging capability. Due to the limitation of physical principles, the initial amount measured by the SO₂ UV camera is the optical thickness of SO₂ gas, which needs to be retrieved into a concentration image with the help of calibration curves, and the accuracy of calibration curves directly affects the accuracy of SO₂ concentration results. Cell calibration and differential optical absorption spectroscopy (DOAS) calibration are two main methods for obtaining calibration curves. In terms of equipment cost, easy operation, and system stability, the cell method is significantly better than the DOAS method, but its calibration accuracy is seriously affected by the light dilution effect, reflections on the windows of the calibration cell and filter, and aerosol scattering factors. Additionally, with the rising detection distance, the above factors, especially the influence of the light dilution effect, become increasingly more serious. To improve the calibration accuracy of the cell method, we research the calibration error correction method to address the practical problem of inaccurate cell method calibration in remote SO₂ monitoring.

In practice, since the factors affecting the accuracy of the cell method are mainly from the light dilution effect, window reflection, and the scattering of aerosols, it is necessary to correct each of these factors. The specific method is as follows. Firstly, the image correction method (ICM) is proposed for correcting the light dilution effect, and the extinction coefficient is obtained by fitting the intensity information of the measurement points at different distances in the UV camera images. Additionally, the optical thickness image of the cell at the measured distance is calculated by the extinction coefficient, and then the calibration curve with the correction of the light dilution effect is obtained. Then, based on the analysis of window reflection and aerosol scattering effect, the influence of the reflection effect and scattering characteristics on the calibration results are quantified. Finally, the calibration curves with the correction of light dilution effect and scattering characteristics are calculated by combining the above influencing factors.

Based on the Etna volcanic plume image data captured by Professor Jonas Gli? from the Norwegian Air Research Institute using a SO₂ ultraviolet camera, the Etna volcanic plume SO₂ concentration image is retrieved by calibration curves before and after the correction of the light dilution effect. The results are compared with the retrieval results of the DOAS calibration curve, and the results show that the correction of the light dilution effect can reduce the differences between the cell method and the DOAS method from 59.0% to 31.3%, which verifies the effectiveness of ICM in correcting light dilution effect. After correction for reflection and scattering effects, the difference between the cell method and the DOAS method is reduced to 7%. The cell method and DOAS method show good agreement in the time domain after correction, and the fitting curve slope of the primary function of the calibration results is 0.924, with a goodness-of-fit of 0.998.

The results show that the proposed error correction method for cell calibration of the SO₂ UV camera can improve the calibration curve accuracy. The fitting accuracy of the extinction coefficient and the measurement accuracy of the filter reflectance and the quartz window directly affect the accuracy of the calibration curve. The error analysis results show that a 10% shift in the extinction coefficients εA and εB obtained from channels A and B fitting will cause an error of 8.44% and 13.57% for SO₂ column density retrieval respectively, while a 10% shift in background light intensity will result in an error of 4.98% for SO₂ column density retrieval. Additionally, a 10% error in the filter reflectance and the quartz window will result in a 6.26% and 1.95% shift in the SO₂ column density respectively. Increasing the interval distance of sampling points and the number of sampling points can improve the fitting accuracy of the extinction coefficient. The high-resolution UV spectrometer ensures that the filter reflectance and the quartz window are accurately measured to control errors caused by the reflectance uncertainty. The proposed error correction method for calibration curves solves the limitation that the cell method cannot be applied to monitor the plumes at long distances and high carbon black concentrations, which is important for better applications of SO₂ UV cameras in volcanoes, ships, and industrial chimneys.