Keywords: 
• nitrate measurement /
• pressure correction /
• ultraviolet spectroscopic measurement system /
• multiplicative scatter correction

Abstract:   Objective  Optical nitrate sensors have advantages for in-situ exploration and the potential for long-term observation in deep-sea environment. However, measurement of optical nitrate is influenced by substrates in seawater, especially bromide ions (Br^-), within the relevant spectral range, causing a spectral shift in the ultraviolet (UV) intensity spectrum. In recent years, the pressure coefficient of the UV absorption spectrum of bromine under seawater pressure of 2 000 meters has been studied and experimentally confirmed, that the bromide in seawater affects the UV absorption spectrum. Considering that the seabed minerals are primarily distributed undersea at depths ranging from 1000 to 6000 meters, achieving accurate in-situ detection of nitrate concentration becomes crucial for the assessment of impact of seabed mining on the marine eco-system, establishment of an early warning system for the marine mining environment. Currently, products of nitrate sensors are limited to the submerged depths of approximately 2000 meters. No nitrate sensor product is available beyond this depth. This paper reports a calibration method for nitrate measurements within the pressure range of 0-50 MPa (0-5000 meters), aiming to improve the accuracy of nitrate measurements in deep-sea environments.  Methods  A system capable of measuring the UV spectrum of seawater under deep-sea pressure is constructed in this work. The light emitted from a deuterium lamp is transmitted through a fiber-optic beam splitter, dividing it into two paths. One path passes through a fiber-optic attenuator, while the other path goes through a pressure vessel. The two signals are then combined at an optical switch and selectively transmitted through it. Finally, the data processing module performs data acquisition and calculation. To simulate the deep-sea environment, the pressure vessel is connected to a weight manometer. The UV absorption spectra at different pressure are measured by controlling the external pressure. A continuous flow analyser was used to calibrate the nitrate concentration in the seawater samples collected from Aoshan Bay. Different levels of nitrate (0-50 μmol/L) were added to the Aoshan Bay seawater, and these seawater samples with different nitrate concentrations were measured by the measurement system.  Results and Discussions  The measurement results revealed a decrease in the absorbance of seawater samples with an increase in pressure. To investigate the pressure-induced changes in different substrates in seawater, identical pressure tests were conducted for nitrate solution (50 μmol/L), Aoshan Bay seawater, and sodium bromide solution (840 μmol/L). The absorbance results obtained are depicted in Fig.2(a). Notably, the absorbance of seawater and bromide under pressure exhibited a similar trend, whereas the absorbance of nitrate remained largely unaffected by pressure. Subsequently, pressure correction of seawater UV absorption spectra was conducted at pressures ranging from 0 to 50 MPa using two algorithms for spectral pre-processing, including standard normal variate transform (SNV) and multiplicative scatter correction (MSC), and regression prediction with the partial least squares regression (PLS) algorithm. The results are presented in Fig. 2(b). It is evident that the R² is 0.991, MAE is 1.980 μmol/L, MBE is −0.042 μmol/L, and root mean square error (RMSE) is 2.505 without using any pressure correction. The R² is 0.997, MAE is 1.294 μmol/L, MBE is 0.037 μmol/L, and RMSE is 1.620 using the MSC-PLS algorithm. The R² is 0.989, MAE is 2.308 μmol/L, MBE is 0.098 μmol/L, and RMSE is 3.085 μmol/L using the SNV-PLS algorithm. Therefore, with the utilization of the MSC-PLS pressure correction algorithm, the prediction results are superior to those without using any pressure correction. This suggests that the pressure correction algorithm improves measurement accuracy. The MSC-PLS algorithm has the highest R² and the smallest error range, indicating its superior pressure correction and data prediction capabilities.  Conclusions  The primary objective of this study is to enhance the accuracy of optical nitrate measurements in the deep-sea environment by addressing the influence of substrates such as bromide on UV absorption spectra. A system capable of measuring the UV spectrum of seawater under deep-sea pressure is constructed, utilizing a deuterium lamp, fiber-optic components, and a pressure vessel. The experimental results demonstrate variations in UV absorption spectra between 200-240 nm under different pressure conditions at the same nitrate concentration. The SNV and MSC algorithms are employed for pressure correction, and MSC-PLS algorithm exhibits superiority in predicting nitrate concentrations under the pressure range of 0-100 MPa (R² of 0.997). Therefore, the proposed method offers potential applications in mining exploration and environmental monitoring.