Bio-optical modeling of primary production on regional scales: the BermudaBioOptics project

Regional to global scale estimates of primary production must rely on remot ely sensed quantities. Here, we characterize in situ light-primary producti on relationships and assess the predictive capability of several global pri mary production models using a 6-yr time series collected as part of the US JGOFS Bermuda Atlantic Time Series (BATS). The consistency and longevity o f this data set provide an excellent opportunity to evaluate bio-optical mo deling methodologies and their predictive capabilities for estimating rates of water-column-integrated primary production, SPP, for use with satellite ocean-color observations. We find that existing and regionally tuned param eterizations for vertically integrated chlorophyll content and euphotic zon e depth do not explain much of the observed variability at this site. Fortu nately, the use of these parameterizations for light availability and harve sting capacity has little influence upon modeled rates of SPP. Site-specifi c and previously published global models of primary production both perform poorly and account for less than 40% of the variance in SPP. A sensitivity analysis is performed to demonstrate the importance of light-saturated rat es of primary production, P*(sat), compared with other photophysiological p arameters. This is because nearly one-half of SPP occurs under light-satura ted conditions. Unfortunately, we were unable to derive a simple parameteri zation for P*(sat) that significantly improves prediction of SPP. The failu re of global SPP models to encapsulate a major portion of the observed vari ance is due in part to the restricted range of SPP observations for this si te. A similar result is found comparing global chlorophyll-reflectance algo rithms to the present observations. More importantly, we demonstrate that t here exists a time-scale (roughly 200 d) above which the modeled distributi ons of SPP are consistent with the observational data. By low-pass filterin g the observed and modeled SPP time series, the model's predictive skill le vels increase substantially. We believe that the assumptions of steady stat e and balanced growth used in bio-optical models of SPP are inconsistent wi th observational data. Most of the observed variance in SPP is driven by a variety of ecosystem disturbance processes that are simply not accounted fo r in bio-optical models. This puts important bounds on how SPP models shoul d be developed, validated and applied. (C) 2001 Elsevier Science Ltd. All r ights reserved.