Modeling attenuation of photosynthetically active radiation across the optical gradient in the Laurentian Great Lakes with application to Lake Erie

Abstract

Representation of subsurface photosynthetically active radiation (PAR) in biophysical models of the Laurentian Great Lakes (LGL) is imperative to their utility as tools for research and management. Here we consolidated measured vertical profiles of subsurface PAR with concurrent water quality (WQ) data from four LGL. We estimated the diffuse attenuation coefficient of PAR (K_d(PAR)) by fitting an exponential function to measured PAR over depth, and evaluated 68 regressions predicting K_d(PAR) as a function of water quality variables (K_d-WQ regressions). We compare four of the top cross-lake calibrated regressions against two published regressions trained on western Lake Erie (WLE) data. Then, as a case study, we demonstrate the utility of our cross-lake calibrated K_d-WQ regressions with a simplified biophysical model of Lake Erie consisting of the Finite Volume Community Ocean Model with submodules for simulating suspended sediment and dissolved organic carbon (FVCOM-SS-DOC). Twenty-five K_d-WQ regressions were identified as candidates for use in biophysical models based on their skill determined via cross-validation. WLE-trained K_d-WQ regressions were less able to simulate K_d(PAR) and PAR in more transparent waters compared to cross-lake calibrated K_d-WQ regressions, which translated to considerable differences in primary production estimates for the central and eastern basins when using WQ data simulated by FVCOM-SS-DOC. A cross-lake calibrated K_d-WQ regression was installed into FVCOM-SS-DOC, which then simulated spatial patterns of suspended sediments and K_d(PAR). These calibrated K_d-WQ regressions can be used in a variety of biophysical models across optically-distinct waters of the LGL to support adaptive management of nutrient inputs and fisheries.