Investigation into Intelligent Spectral Nitrate Sensor Temperature and Salt Correction

This study proposes an intelligent temperature and salinity correction scheme to address the primary issue that the accuracy of ultraviolet spectral seawater nitrate sensors is influenced by temperature and salinity disturbances in the real marine environment. The experimental results indicate that the bromide absorbance is significantly influenced by temperature, whereas the salinity has a relatively weak effect and is linearly related to the bromide concentration. To achieve accurate nitrate concentration inversion, this scheme extensively uses wavelet denoising, baseline correction, and piecewise deep learning. Additionally, the salinity credibility factor is introduced to reconstruct the bromide ion temperature and salinity compensation function. The improved algorithm’s mean absolute error (MAE) and root mean square error (RMSE) are 0.008 and 0.010, respectively, lower than those of the conventional polynomial fitting method. This greatly enhances the sensor’s accuracy and stability under complex, and fluctuating temperature and salinity conditions, and provides reliable technical support for real-time monitoring of the marine ecological environment.