Upwelling spectral radiance as measured in different depths at radiometer station PS73/062-2 (corrected version)@en

In the previous upwelling radiance (Lu) profile data set (doi:10.1594/PANGAEA.820814), Lu data were wrongly shifted by 3 nm (which responds to the originally resolution of the measurement). The results of the publications related to these data were not affected. Taylor et al. 2013 investigated only Lu data incorrectly shifted by 3 nm, so the analysis was consistent within itself. Taylor et al. 2011 and Bracher et al. 2015 used the remote sensing reflectance (RRS) data calculated from the incorrectly shifted Lu data. The resulting RRS data only showed inconsistencies were the incoming downwelling irradiance changes quickly with wavelengths, which resulted erroneous features <430 nm. We have used both versions (previous and the one presented here) Lu and RRS data and got the same results since the wrong features <430 nm are consistent and are neither used for the hierarchical cluster analysis (only data >430 nm are considered) performed in Taylor et al. 2011 nor for the pigment predictions (since these features are contained in the first empirical orthogonal function) in Bracher et al. 2015.在既往的上行辐亮度（upwelling radiance, Lu）剖面数据集（DOI:10.1594/PANGAEA.820814）中，Lu数据存在3 nm的偏移错误（该偏移对应于该测量的原始分辨率）。基于该数据集的相关研究成果并未受到影响。2013年Taylor等人仅针对存在3 nm偏移错误的Lu数据开展研究，因此其分析内部逻辑自洽。2011年Taylor团队与2015年Bracher团队均使用了由存在偏移错误的Lu数据计算得到的遥感反射比（remote sensing reflectance, RRS）数据集。由此得到的RRS数据仅在入射下行辐照度随波长快速变化的区间出现不一致，进而在波长小于430 nm的波段产生了误差特征。本次研究同时使用了旧版与本次发布的Lu及RRS数据，最终得到了一致的结果：波长<430 nm的误差特征具有一致性，且2011年Taylor团队的层级聚类分析（仅考虑波长>430 nm的数据）与2015年Bracher团队的色素预测（该误差特征包含于第一经验正交函数中）均未用到这部分数据。