A Computationally Efficient Parametrization of the Underwater Light Field for Biogeochemical Models in Coastal Waters

Jochen Wollschl├Ąger1, Beke Tietjen1 and Oliver Zielinski1,2, (1)University of Oldenburg, Marine Sensor Systems Group, Institute for Chemistry and Biology of the Marine Environment, Oldenburg, Germany, (2)German Research Center for Artificial Intelligence, Marine Perception Group, Oldenburg, Germany
Light is a parameter relevant for all kinds of aquatic life. Its most fundamental role is that of a resource driving photosynthesis and thus primary production, essentially fueling the whole marine food web. A common measure for biologically relevant light availability is the photosynthetically active radiation (PAR), which integrates the spectral distribution of photon flux between 400 to 700 nm into a single value. This simplification reduces the e.g. the computational effort of biogeochemical models when dealing with light. Through the water column, PAR is attenuated by absorption and scattering processes, summarized by the diffuse attenuation coefficient of downwelling radiation (Kd). Kd is a function of the optical properties of the water itself as well as of its optically active constituents (mainly phytoplankton, chromophoric dissolved organic matter, and non-algal particles). Since biogeochemical models usually do not include Kd, for their purposes, light attenuation has to be parameterized by variables commonly included, e.g. chl-a and suspended matter. Efforts in this respect have been done in the past, but rarely for optically complex coastal waters.

In this contribution, the results of in situ hyperspectral underwater light field measurements made in a coastal environment (North Sea) are set in context with inherent optical properties and direct water constituent measurements. The collected data were used to parameterize the attenuation of PAR in an extended bimodal approach in order to obtain a sufficiently accurate, but still computational affordable linear representation of this process for modeling purposes. Modeled and measured data are compared, and strengths and weaknesses of the chosen approach are discussed.