光谱仪作为可见-近红外光谱分析核心工具, 对于光谱分析的影响还有待深入研究。 这里, 以黄渤海沉积物为研究对象, 利用安捷伦Cary 5000、 ASD FieldSpec 4、 海洋光学QEPro三种光谱仪, 研究有机碳含量的反射光谱及其光谱分析在不同光谱仪间的差异。 其中, Cary 5000为室内台式光谱仪, FieldSpec 4和QEPro是便携式光谱仪; QEPro只能采集200~1 000 nm的反射率, 而且反射率在三种光谱仪中最高; Cary 5000和FieldSpec 4能够采集完整的可见-近红外反射率(350~2 500 nm), 光谱曲线几乎完全一致, 尤其是近红外部分, 但Cary 5000采集的反射率高于FieldSpec 4。 三种光谱仪的光谱分析能力也存在显著差异, 其中, Cary 5000的光谱分析能力最强, 其光谱模型具有很强的预测能力, 建模集 r²高达0.99, 验证集 r²高达0.86, 均方根误差(RMSE)在建模集和验证集分别为0.04、 0.11, 相对分析偏差(RPD)高达2.6, 显示了对黄渤海沉积物有机碳强大的预测能力。 FieldSpec 4的建模集 r²很高, 达到0.98, 但验证集 r²仅为0.56, RMSE从建模集的0.06降为预测集的0.19, RPD仅为1.4, 显示在黄渤海沉积物有机碳的预测中准确度很低。 QEPro的建模集和验证集 r²都很低, 分别为0.75、 0.59, RMSE保持稳定, 均为0.18, RPD刚超出1.5(RPD=1.6), 显示在黄渤海沉积物有机碳的预测上具有一定的准确度。 可见, 室内光谱仪比便携式光谱仪具有更强的光谱分析能力, 可能是因为室内光谱仪的配置更好。 同为便携式光谱仪, 在黄渤海沉积物有机碳的光谱分析方面, QEPro具有和FieldSpec 4一样的光谱分析能力, 甚至在稳定性上更好。 由于QEPro的价格相对低廉, 因此在室外沉积物/土壤碳氮光谱分析中, 具有更好的性价比。 该研究深入比较了不同光谱仪在相同样品的光谱分析中的差异, 为不同研究的光谱分析及其模型转移提供了有效参考。

Spectrometers are the core tool in spectral analysis, but it is still unclear how spectrometers influence the results of spectral analysis. Here, we studied the spectral analysis of organic carbon in sediments of the Yellow Sea and Bohai Sea using Agilent Cary 5000, ASD FieldSpec 4, and Ocean Optics QEPro and compared differences in the reflectance spectra of organic carbon and their spectral analysis. Cary 5000 is an indoor spectrometer, and FieldSpec 4 and QEPro are portable spectrometers. QEPro could only collect the reflectance between 200 and 1 000 nm, and the reflectance is the highest among the three spectrometers. Cary 5000 and FieldSpec 4 could collect the reflectance of the complete visible and near-infrared waveband (350~2 500 nm), and both spectral curves were almost identical, especially in the near-infrared bands. However, the reflectance collected by Cary 5000 is higher than that by FieldSpec 4. The abilities of spectral analysis of organic carbon concentrations in the Yellow and Bohai Sea were also different across the three spectrometers. Cary 5000 had the strongest ability to perform spectral analysis. The spectral models in Cary 5000 had a strong prediction ability of organic carbon concentrations in sediments. In Cary 5000, the r² of the calibration set was as high as 0.99, and the r² of the validation set was as high as 0.86; the root mean square error (RMSE) of the calibration set and validation set was 0.04 and 0.11, respectively; the relative prediction deviation (RPD) was as high as 2.6, showing a strong ability to predict sediment organic carbon. In FieldSpec 4, r² of the calibration set was as high as 0.98, but r² of the validation set was only 0.56; RMSE decreased from 0.06 to 0.19, and RPD was as low as 1.4, showing a low prediction ability of sediment organic carbon. In QEPro, r² of the calibration set and validation set were both low (0.75 and 0.59, respectively), RMSE were stable and as high as 0.18, and RPD was larger than 1.5 (1.6), showing a convincing prediction ability of sediment organic carbon. Results showed that the portable spectrometers were worse than indoor instruments in spectral analysis due to their lower technological performance. For the portable spectrometers, results in spectral analysis were not different between FieldSpec 4 and QEPro, and even the results of QEPro were more stable. Therefore, we think QEPro can be replace with FieldSpec 4 in rapidly determining sediment organic carbon by spectroscopy because QEPro is cost-effective. In this study, the differences among three types of spectrometers in the spectral analysis of the same sample were compared, which provided an effective reference for the spectral analysis and model transfer of different studies.