Nitrate Measurement in Seawater Based on Environmental Correction Algorithm

The accuracy of the in situ nitrate measurement in seawater is affected by many environmental factors, such as salinity, temperature, and turbidity. Along these lines, in this work, an environmental correction (including salinity, temperature, and turbidity) algorithm was proposed and experimentally verified, demonstrating a good prediction accuracy of the nitrate concentration. The orthogonal projection to latent structures (OPLS) algorithm was used for performing one-time turbidity and temperature correction of the acquired spectral data. After the environmental correction, the support vector machine (SVM) was used to establish the nitrate calculation model. The OPLS-SVM algorithm was systematically examined in spiked samples with four different groups of seawater matrices (the seawater samples from Western Pacific, Aoshan Bay of Qingdao, China, South China Sea, and Yellow Sea with addition of various nitrate concentrations and turbidity). From the extracted results, it was proven that the SVM calculation model with the OPLS correction algorithm could significantly improve the accuracy of the nitrate measurement. More specifically, a prediction performance with root-mean-square error (RMSE) of
$0.22 \mu $
mol/L and
${R}^{{2}}$
of 0.999 was demonstrated when the temperature and turbidity ranges were 5 °C–25 °C and 0–50 nephelometric turbidity unit (NTU), respectively. The introduced method is suitable for measuring the nitrate concentration in turbidity seawater with a high degree of accuracy, which is undoubtedly of great importance for performing in situ nitrate measurement in complex seawater environments.