摘要
卫星宽幅推扫成像云图定位精度受光学系统畸变、CCD阵列形变等因素影响,其系统性评估与精确校正是提升数据产品应用价值的关键环节.针对一种星载多相机推扫载荷的宽幅成像定位精度优化需求,本研究创新性地提出了一套完整的解决方案:首先建立了基于单通道图像特征自动识别与海岸线匹配的大样本特征点提取方法,进而构建了融合顺轨平移、跨轨偏移以及扫描线拉伸和旋转效应的四参数非线性校正模型.实测数据分析表明,该载荷原始图像存在显著的跨轨方向位置依赖性定位误差,表现为从星下点向图像边缘呈梯度增大的特征,其中拉伸和旋转因素导致的几何畸变最大可达10像元和20像元.通过30余天的独立观测数据验证表明,所提方法在无需任何辅助信息的情况下,可有效校正图像边缘区域的定位偏差,使各相机顺轨和跨轨方向的平均定位误差从初始的1~3像元优化至亚像元级.特别地,全幅宽平面定位误差经校正后降至0.46像元,相对改进幅度超过75%.本研究的主要贡献在于:解决了推扫式成像系统特有的跨轨方向非线性几何畸变校正难题;提出的方法适用于宽幅云图产品定位结果的常态化监测与精度提升;建立的模型具有计算高效、便于业务化应用的特点.该成果为同类卫星载荷图像的定位精度保障提供了可靠的技术支撑,对提升气象海洋卫星数据的应用效能具有重要意义.
The geolocation accuracy of satellite wide-swath push-broom cloud imagery is affected by factors such as optical system distortion and CCD array deformation.Systematic evaluation and precise correction are crucial for enhancing the application value of data products.To address the geolocation accuracy optimization requirements for a multi-camera push-broom payload,this study innovatively proposes a comprehensive solution:First,we established a large-sample feature point extraction method based on automatic single-channel image feature recognition and coastline matching.Then,we constructed a four-parameter nonlinear correction model incorporating along-track shift,cross-track offset,scan line stretching,and rotation effects.Experimental data analysis reveals that the original images exhibit significant cross-track position-dependent geolocation errors,showing a gradient increase from the nadir to the image edges.The maximum geometric distortions caused by stretching and rotation factors reach 10 pixels and 20 pixels respectively.Validation using over 30 days of independent observation data demonstrates that the proposed method can effectively correct geolocation deviations at image edges without requiring any auxiliary information,reducing the average along-track and cross-track geolocation errors from initial 1~3 pixels to sub-pixel level.Notably,the full-swath planar geolocation error is reduced to 0.46 pixels after correction,representing a relative improvement exceeding 75%.The main contributions of this study include:(1)solving the unique challenge of nonlinear geometric distortion correction in the cross-track direction of push-broom imaging systems;(2)developing a method suitable for routine monitoring and accuracy improvement of wide-swath cloud product geolocation;(3)establishing a computationally efficient model that facilitates operational applications.These achievements provide reliable technical support for ensuring geolocation accuracy of similar satellite payloads and significantly enhance the application effectiveness of meteorological and oceanographic satellite data.
作者
姚志刚
江军
赵增亮
YAO ZhiGang;JIANG Jun;ZHAO ZengLiang(Beijing Institute of Applied Meteorology,Beijing 100029,China;Key Laboratory of Smart Earth,Beijing 100029,China)
出处
《地球物理学报》
北大核心
2025年第9期3353-3366,共14页
Chinese Journal of Geophysics
基金
智慧地球重点实验室自主科研项目(SYS-ZX05-2024-01)资助.
关键词
卫星
云图
宽幅
定位
评估
校正
Satellite
Cloud Imager
Wide swath
Geolocation
Evaluation
Correction
