Hi Peter, chemometrics ...

Hi Peter, chemometrics methods   such as Partial Least Squares regression applied on UV spectroscopy data   200-400nm is very appropriate to obtain a robust calibration model for NO3 using as refernce data your actual sensor/method.   This model will work accurately when experimental conditions  do not change significantly with time. These conditions include a similar concentration range, stable background of impurities (operational procedures do not change permanently).  Otherwise , frequent recalibration steps would be needed.  Choosing this chemometric approach you will need to create an experimental system based on continuous flow or  fiber optic cables  going to a classical UV desktop instrument and kind of chemometric software (commercial) or R, or Python scripts to calculate. Look at this article for information

  Dahlen et al. Chemosphere 40 (2000) 71-77.

As alternative to this let say laboratory approach,   on the market, also exist UV submergible sensors   measuring NO3  and other parameters  on line, using their global calibration models.

Hope to be useful,

Stefan