Interpreting the colour of an estuary

This paper explores the possibility of using water colour to quantify the concentration of coloured dissolved organic matter (CDOM), and through it, dissolved organic carbon (DOC) and salinity in a turbid estuary in which suspended sediments also influence water colour. The motivation of the work is that the method could be applied to water colour measurements made remotely from an aircraft (or, in larger estuaries, a satellite) enabling near-synoptic mapping of surface salinity and DOC distributions. The paper describes observations at 29 stations distributed along the salinity gradient of the Conwy estuary in North Wales. At each station, surface water samples were collected and analysed for salinity, concentrations of DOC, chlorophyll and suspended particles and absorption spectra of CDOM, or yellow substance. Profiles were made of both upwelling and downwelling irradiance in four narrow band channels, and these were used to calculate irradiance reflection and attenuation coefficients. Results show that spectrally averaged light absorption in the estuary is caused principally and equally by mineral suspended solids and yellow substance, with water and chlorophyll in third and fourth place. The CDOM is strongly correlated (R 1⁄4 0:99) in a negative sense with salinity, and more weakly correlated with DOC. There is a linear relationship between CDOM and the ratio of reflection coefficients in the red (670 nm) and blue-green (490 nm) parts of the spectrum, which could be applied to remote sensing; the slope and intercept of the relationship are however different to those found in less turbid water bodies. It is shown that the change in slope and intercept are consistent with the presence, in the Conwy estuary, of suspended particles which influence the water colour. A method is described and tested for inverting water colour measurements in a turbid estuary to give estimates of CDOM in the presence of suspended particles. The solution, which has not been adjusted to fit the data, produces profiles of CDOM, DOC and salinity, which behave reasonably but which currently have a limited accuracy. RMS differences between measured and optically derived parameters for the entire data set are 0.27 m 1 (CDOM), 4 PSU (salinity) and 67 mM (DOC) although better accuracy is obtained on individual surveys. The fact that there is little bias between predicted and observed parameters indicates that much of the scatter is caused by random measurement error and that the approach is fundamentally sound. 2003 Elsevier Ltd. All rights reserved.