In recent years, remote sensed measurements of sea surface salinity have been identified as a key requirement in mixing and pollution studies. Existing methods of remote sensing salinity are largely limited to microwave radiometry, which operates on the principle that the microwave emissivity of the sea is strongly salinity dependent. Orbiting microwave radiometry sensors (SMOS) however, lack the spatial resolution to adequately resolve most estuaries, whilst airborne microwave radiometry is subject to radio frequency interference, and is therefore often unreliable in estuarine regions that are near cities. In many estuaries, the concentration of coloured dissolved organic matter (CDOM) shows a strong, linear relationship with salinity, since it mainly enters estuaries through
land runoff. The aim of this study, was to develop a method by which the concentration of CDOM, and hence salinity, could be derived from remote sensed water colour in the optically complex Conwy Estuary. To achieve this aim, measurements of the specific inherent optical properties of the dominant optically active in-water constituents (OICs) were made, and in-situ measurements of water colour collected to enable both forward and inverse modelling of water colour. Airborne remote sensed water colour was obtained in the estuary during spring, summer, and autumn months to test the success of the developed inversion technique.
Specific CDOM absorption was determined spectrophotometrically, whilst specific absorption coefficients (a*) of mineral suspended solids (MSS) and phytoplankton pigments (PIG) were measured for particles concentrated on filter papers, using the Qualitative Filter Pad (QFP) technique. Measurements collected over two years showed a clear seasonal variation in the shape of the a*PIG spectra, with an exponential shaped spectra present in winter months and a twin peaked shaped spectra, typical of chlorophyll a, present in summer. A simple time varying model of a*PIG was thus developed, and enabled the forward modelling of pigment absorption to within an RMS error of 0.07 m·'. Multivariate linear regression was employed to retrieve a*MSS, which along with the time varying model of a *PIG was used to forward model total particle absorption to within an RMS error of 0.37 m·' for an independent dataset. Backscattering in the Conwy Estuary was found to be dominated by MSS, and was relatively wavelength independent.
Forward modelled water colour showed close agreement to the in-situ values, within measurement uncertainties, and revealed that whilst MSS significantly affected all colour ratios, the irradiance reflectance at far-red wavelengths was dominated by water and MSS alone. In addition, modelled far-red/red colour ratios were seen to be dominated by MSS and pigments, whilst far-red/green
colour ratios were dominated by MSS and CDOM. Theoretical water colour inversion methods were tested using an in-situ dataset, and included both solving simultaneous equations of water colour and over-determined matrix inversion. The results of both techniques were inadequate, and analysis using synthetic data showed both techniques were highly susceptible to small measurement uncertainties in the measured colour ratios. An alternative spectral windowing inversion technique was therefore developed, in which the far-red irradiance reflectance was used to obtain the concentration of MSS, the far-red/red colour ratio was used to obtain the concentration of pigments, and the far-red/green colour ratio used to obtain the concentration of CDOM.
Remote sensed estimates of MSS concentration were in excellent agreement with values determined in-situ, and showed the presence of a turbidity maximum in the upper reaches of the estuary. The axial gradients of remote sensed pigment concentration were also in close agreement to those observed in the in-situ values, and both showed a twin peaked phytoplankton maximum was also
present in the Conwy Estuary, slightly seaward of the turbidity maximum. Remote sensed estimates of CDOM concentration were less accurate, with the most successful estimates overestimating high concentrations, whilst underestimating low concentrations. This was largely attributable to inaccuracies in the atmospheric correction of the remote sensed data. Nevertheless, the remote sensed CDOM concentrations were used to produce remote sensed salinities that revealed a very similar axial variation in salinity from the mouth to the head of the estuary as that observed in-situ.