Predicting visual clarity and light penetration from water quality measures in New Zealand estuaries

Estuaries are biogeochemically and physically dynamic environments, resulting in diverse optical water properties that vary in space and time. This natural variability challenges effective monitoring. We develop semi-empirical linear models of the two main water clarity aspects, visual clarity and light penetration, as functions of more routinely monitored and/or modelled water quality measures, to simplify approximations and allow more expansive modelling in space and time. At multiple points along salinity gradients in six northern New Zealand estuaries, we made discrete measurements of horizontal black disk visibility (yBD) and downwelling irradiance attenuation (Kd(PAR)), with surface concentrations of the main light-attenuating constituents (LACs): total suspended solids (TSS), coloured dissolved organic matter (CDOM – ag(340)), and phytoplankton chlorophyll a (Chl.a). Estuaries on the east coast (Tauranga and Tairua Harbours) were relatively clear compared to those on the west coast (Kaipara and Raglan Harbours). Although Kd(PAR) (controlling light penetration and euphotic zone depth) and yBD (inversely proportional to beam attenuation at 550 nm, controlling visual clarity) were strongly corelated over all estuaries (r2 > 0.8), there were small, yet significant, differences between regression lines in different estuaries. This suggests regional differences in light scattering and absorption properties of the LACs. Variability in water clarity was mainly accounted for by TSS alone in west coast Kaipara Harbour sub-estuaries (r2 > 0.7), while in east coast estuaries, TSS was a weak predictor (r2 0.7). West coast Raglan Harbour had intermediary responses. In most cases, the inclusion of CDOM significantly improved estuary models, but Chl.a was a significant term only in a few cases, such as during one survey on the Kaipara River sub-estuary. Co-linearity of LACs was an issue in some cases for semi-empirical models. The application of a literature-derived Lambert-Beer Law (LBL) type model to our data more effectively partition the contributions of the LACs, which confirmed and quantified differences between west coast (TSS dominant), and east coast (TSS + CDOM) estuaries. Although we expect that ‘simple’ semi-empirical models will have wide practical application for estimating visual clarity and light penetration in estuaries in New Zealand and elsewhere, consideration of all LACs in an LBL type model is advised across entire estuaries, and local verification and ‘tuning’ is desirable.