Computation of nitrate concentrations in coastal waters using an in situ ultraviolet spectrophotometer: Behavior of different computation methods in a case study a steep salinity gradient in the southern North Sea

Abstract

Absorption spectra of seawater can be used to estimate the concentration of nitrate based on the UV absorption characteristic of nitrate. However the results of that estimation show an increased uncertainty compared to wet chemical methods. This is caused by the close proximity and the magnitude of the bromide peak (as the main component of seawater salt) close to the nitrate signal in the UV. Current data processing methods are optimized to give good results under constant conditions in terms of temperature, salinity, and CDOM concentration. However, in coastal regions all three parameters are highly variable.

In this work three methods to determine nitrate concentration from the seawater UV spectrum are compared: (A) via the subtraction of the seawater spectrum and CDOM absorbance from the total absorbance of the sample and then fitting the nitrate absorption to the remaining absorbance, (B) the subtraction of the seawater spectrum and fitting the spectral signature of nitrate and CDOM as suggested by Sakamoto et al. (2009) and (C) the direct determination via the fitting of the spectral signature of all components to the sample spectrum. The results of all three methods correlate ($R>0.99$) very well with each other as well as to the results of the wet chemical analysis.

An extensive dataset of a transect from the Southern North Sea into the Weser estuary (RV HEINCKE transect 345), which covers a broad salinity range as well as a broad range of nitrate concentrations, is used to exemplary show the potential and the limitations of all three methods under these conditions.