高分一号宽幅多光谱影像辐射定标偏差及其植被指数影响

Abstract

PDF HTML阅读 XML下载 导出引用 引用提醒 高分一号宽幅多光谱影像辐射定标偏差及其植被指数影响 DOI: 10.5846/stxb202205241462 作者: 作者单位: 作者简介: 通讯作者: 中图分类号: 基金项目: 国家重点研发计划课题项目2019YFB2102004）；福建省自然科学基金项目（2021J011190）；厦门理工学院高层次人才项目（4010520004）；江西省水利科技项目（202224ZDKT11） Radiometric calibration bias of the Gaofen-1 WFV multispectral imagery and its influence on vegetation index Author: Affiliation: Fund Project: 摘要 | 图/表 | 访问统计 | 参考文献 | 相似文献 | 引证文献 | 资源附件 | 文章评论 摘要:自2013年4月在轨运行以来，高分一号（GF-1）宽幅（WFV）多光谱相机已实现持续对地观测，为包括生态学在内的相关研究领域与行业应用提供了丰富的数据源。当前场地定标频次偏低以及定标参数更新、发布滞后的现状，一定程度上影响了GF-1 WFV多光谱数据的定量应用。然而，现有文献仅在数据预处理流程中谈及对GF-1 WFV影像的辐射定标处理，很少讨论定标参数选取不当甚至误用产生的可能影响。基于已公布的辐射定标参数（2014-2021年）和4景GF-1 WFV Level1A级影像产品数据，重点围绕辐射定标偏差及其对多光谱波段星上反射率和常用植被指数的影响等展开模型分析和讨论。结果显示：即使两个相邻年份间，在多数情况下误用辐射定标参数会导致不可忽视的波段星上反射率相对偏差；进而给不同类型植被指数的实际应用带来不同程度的挑战。因辐射定标偏差的影响，常用的两波段归一化型植被指数在监测稀疏植被覆盖区时会存在明显的误差；而对高植被覆盖区的监测时采用两波段简单比值型植被指数将面临更大的挑战。针对存档GF-1 WFV Level1A数据的应用需求，提出利用时间距离加权的线性内插法来修正基于公开定标参数的辐射定标结果，并通过案例分析表明了该处理方法的有效性。最后，希望研究结果能引起普通用户对卫星遥感影像辐射定标的关注和重视。 Abstract:Since its on-orbit operation in April 2013, the wide field of view (WFV) sensors onboard Gaofen-1 (GF-1), where "Gaofen" means high resolution in Chinese, have continuously collected multispectral imagery of the Earth's surface. The GF-1 WFV sensors have provided abundant data sources for many fields including ecology. The fact that insufficient site calibration (i.e. once a year) as well as the updates lag of calibration parameters might limit the quantitative application of the GF-1 WFV imagery to a certain extent. However, current literature has only mentioned radiometric calibration of the GF-1 WFV images as a pre-processing procedure, whereas has rarely discussed the possible impacts caused by improper selection or misuse of calibration parameters. Based on the radiometric calibration parameters published for the GF-1 WFV imagery (from 2014 to 2021) and four scenes of the Level1A data products, the issue on radiometric calibration bias was investigated. Furthermore, impacts of the radiometric calibration bias on the top of atmosphere (TOA) reflectance and on several vegetation indices commonly used were discussed accordingly. It showed that, generally, even the calibration difference between two neighboring years were considered, in most cases, improper selection of the calibration parameters could result in a significantly relative bias in the TOA reflectance. The bias in TOA reflectance further challenged different types of vegetation indices with varying degrees of patterns in practice. In particular, mainly due to the radiometric calibration biases, the obvious errors were more likely in the normalized difference vegetation index based on two bands, for monitoring sparse vegetation coverage area. Actually, the normalized difference vegetation index has been commonly used in assessing vegetation status as well as surface dynamics. At the same time, for monitoring high vegetation coverage area, the simple ratio vegetation index with two bands was possibly confronted with greater challenges. Consequently, to make full use of the GF-1 WFV Level1A products, it is critical to solve the radiometric calibration problems. In this study, a time weighted linear interpolation method was proposed. In the interpolation process for a specific GF-1 WFV imagery, both radiometric parameters obtained closely before and after the imagery acquisition were integrated. A case study suggested the effectiveness of the proposed processing to improve the radiometric calibration of the GF-1 WFV images in Level1A products, as compared against the results obtained merely based on public calibration parameters. Finally, general users should pay much attention to the radiometric calibration of satellite remotely sensed data (e.g., the GF-1 WFV multispectral imagery) for their quantitative applications, as discussed in this investigation. 参考文献 相似文献 引证文献