Bayesian estimation of ecosystem model parameters in the Black Sea: Integrating satellite observations to reduce uncertainty
OCEAN MODELLING, vol.201, 2026 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Volume: 201
- Publication Date: 2026
- Doi Number: 10.1016/j.ocemod.2026.102712
- Journal Name: OCEAN MODELLING
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Artic & Antarctic Regions, Compendex, Geobase, INSPEC
- Keywords: Bayesian ecosystem modeling, Black Sea, Uncetainity quantification, Parameter estimation, Data assimilation
- Dokuz Eylül University Affiliated: Yes
Abstract
The Bayesian Hierarchical Modeling (BHM) approach, developed by Parslow et al. (2013, Ecological Applications, 23:679-698) to estimate biogeochemical parameters in a stochastic nutrient-phytoplankton-zooplankton-detritus (NPZD) model, was implemented in the open waters of the Black Sea. Using the Particle Markov Chain Monte Carlo (PMCMC) method, the model integrates satellite-derived chlorophyll-a data with physical variables such as sea surface temperature and mixed layer depth. Joint inference of ecosystem states and model parameters resulted in a substantial reduction in posterior parameter uncertainty. Posterior distributions indicated strong constraints on biological parameters, with uncertainty reductions ranging from 60% to 97% for phytoplankton growth and zooplankton dynamics. We show that predictive uncertainty in a marine biogeochemical system organizes into flow-dependent structures under changing physical regimes, with uncertainty quantification metrics and clustering revealing the limits of satellite-based monitoring. Parameters that regulate vertical processes-like detritus sinking and light attenuation-remained difficult to identify, underscoring the need for complimentary in situ observations.