MEASURED AND MODELED PRIMARY PRODUCTION IN THE NORTHEAST ATLANTIC (EUMELI JGOFS PROGRAM) - THE IMPACT OF NATURAL VARIATIONS IN PHOTOSYNTHETIC PARAMETERS ON MODEL-PREDICTIVE SKILL

Use of ocean color satellite data in global biogeochemical studies req uires models to predict primary production from the satellite-derived chlorophyll fields. In this paper, measured biooptical and photo-physi ological data are used in place of standard (constant) parameters to a djust a previously published primary production model. In the JGOFS-Fr ance program, systematic studies were carried out at three locations i n the tropical northeast Atlantic, selected to represent typical EUtro phic, MEsotrophic and oLIgotrophic regimes (EUMELI cruises). During cr uise no. 4, these studies included the spectral measurements of the ph otosynthetically available radiation at sea level and within the water column, the determination of the algal absorption spectra and the det ermination of the physiological parameters derivable from P versus E e xperiments (photosynthesis-irradiance responses). The model prediction s are compared with in situ determinations made by the C-14 technique (JGOFS core parameter). At the three sites, the physical structure (mi xed layer and euphotic depths), the algal abundance and community stru cture, as well as their bio-optical and physiological properties, are very different, so that the predictive performance of the model was te sted in trophic conditions that span most of those expected in the glo bal open ocean. The model, when adjusted by entering the actual physio logical parameters (chlorophyll-specific absorption of algae, maximum quantum yield, and light saturated carbon fixation rate), provides sat isfying results compared to those observed in situ. The relative roles of the physiological parameters are analyzed and sensitivity studies are performed. For global applications, and in the absence of specific information when all seasons and provinces of the world ocean are con sidered, it will remain necessary for a while to rely on generic model s and a selected standard set of physiological properties. The sensiti vity studies here presented help in this choice, and a modified set of parameters is proposed and tested. With this set, reconstructed produ ction profiles are close to those determined in the field, and the int egrated values are retrieved with no bias and a reduced scatter (18% a t one SD) for 17 stations (cruises 3 and 4) and daily production rangi ng from 0.3 to 2.3 gC m(-2). Copyright (C) 1996 Elsevier Science Ltd