TG-2/MAI CCD像元响应特性不一致性的在轨分析及校正

CCD像元响应特性的差异是制约MAI成像质量及其数据定量化应用的主要因素之一。为了提高MAI数据质量，本文基于全量程多段分析与校正法，利用2016年9月至2018年3月期间共104403帧观测数据，分别对MAI偏振通道和非偏振通道的像元响应特性的不一致性开展了在轨分析与校正，并利用GOME-2和MODIS数据产品对校正结果进行了验证。首先，假定观测样本足够多，即每个CCD像元观测的样本具有相同的遍历性，则各CCD像元对应的所有样本的平均DN值可以代表CCD各像元的响应特性；其次，利用104 403帧观测数据构建各个通道的参考图像，并利用MAI中心5×5像元给出各参考图像对应的标准DN值；在此基础上分别对MAI偏振通道和非偏振通道开展了像元响应特性的分析，结果表明，MAI各通道均存在CCD像元响应特性不一致的问题，各通道的不一致性大约在4%～10%之间，对偏振通道而言，同一偏振波段的三个偏振通道之间像元响应特性的不一致性有一定的相似性，像元响应特性不一致性的差异基本在1%以内。然后，将MAI近两年的观测数据分为前后两个时间段进行对比分析。结果表明：前后两个时间段偏振通道和非偏振通道的图像均具有很好的一致性，即CCD像元响应特性未随时间发生显著变化，这也进一步验证了前面MAI数据量充足的假定。因此，可以利用全量程多段校正法逐通道逐像元开展CCD像元响应特性不一致性的校正。基于该方法校正后，MAI图像质量得到显著改善，图像四周响应偏低的区域明显改善，基本和周围像元的响应达到了同一水平；图像更加平滑，颗粒感基本消除；部分区域的场景发生了变化，特别是碎云等反射率介于中低反射率之间的目标。基于GOME-2的交叉对比结果表明，MAI 565，670和763 nm波段反射率与GOME-2的参考反射率之间的平均绝对偏差分别由校正前的1.6%，4.2%和2.2%减小至校正后的0.5%，2.6%和0.4%；此外，基于多通道云识别方法开展的云检测表明，校正后的MAI云检测结果与MODIS云检测产品一致性更好。因此，全量程多段校正方法可以有效解决MAI CCD像元响应特性的不一致性，显著提高MAI在轨观测的质量，且该方法也可以应用于其他CCD仪器的在轨校正。

On-Orbit Analysis and Correction of the Inconsistency in the Response Characteristics of TG-2/MAI CCD Pixels

The difference in the response characteristics of charge-coupled device (CCD) pixels is one of the main factors restricting the quality of the Multi-angle Polarization Imager (MAI) and its quantitative application. In order to improve the quality of CCD imaging, this paper used a total of 104,403 frames of observational data from September 2016 to March 2018, based on the full-range multi-section analysis and correction method, to realize the analysis and correction for the inconsistency in the pixel response characteristics of MAI polarized and non-polarized channels. And the results were verified using the Global Ozone Monitoring Experiment 2 (GOME-2) and Moderate-Resolution Imaging Spectroradiometer (MODIS) data. Firstly, assuming there are sufficient observational samples, that is, the objects observed by each pixel have the same ergodicity, and the average Digital Number (DN) value of all samples corresponding to each CCD pixel could represent the response characteristics of each pixel of the CCD. Secondly, constructing reference images for each channel by using 104 403 frames observation data, and the 5×5 pixels in the CCD center are used as standard DN values corresponding to each reference image. Next, the response characteristics of the MAI polarized and non-polarized channels are analyzed respectively. The results show that there is significant inconsistency in the response characteristics in each MAI pixel and each channel. The inconsistency in each channel falls roughly between 4% and 10%. And for polarized channels, the inconsistencies in the pixel response characteristics between the three polarized channels in the same polarization band have some similarities, but there are certain differences, and the difference in pixel response inconsistency is basically within 1%. Then, the observational data for two years are divided into two periods for comparative analysis. The result shows that the CCD pixel response characteristics do not decay over time. This also shows that the amount of reference image data is sufficient, which further rerifies the rationality of the above assumptions. Therefore, the full-range multi-section correction method can be used to correct the inconsistency of each pixel’s response characteristics on a channel-by-channel basis. After correction was performed based on this method, the image quality of MAI is significantly improved. The dependence on the observational zenith angle of the CCD pixel response is significant. The image is smoother, and the graininess is basically eliminated. Also, the scenes of some areas have changed-especially targets with a reflectivity between low and medium reflectivity, such as broken cloud and so on. Compared with GOME-2, the average absolute deviations between the reflectance of the MAI 565, 670, and 763 nm bands and the GOME-2 reference reflectance are reduced from 1.6%, 4.2%, and 2.2% to 0.5%, 2.6%, and 0.4% after correction, respectively. In addition, the cloud detection result based on the multi-band cloud identification method shows that, compared with the MODIS cloud detection product at a similar time, the corrected MAI cloud detection result looks more accurate. Therefore, the full-range multi-section analysis and correction method can realize the monitoring and correction of the inconsistency in the response characteristics of the MAI CCD pixels, which significantly improves the quality of the on-orbit observations of this instrument. And this method can also be applied to the on-orbit calibration of other CCD instruments.