Uncertainty in reservoirs ecological modelling: propagation from physical to phytoplankton dynamic predictions

Abstract

Ecological models are commonly recognized as useful scientific tools to describe, analyse and predict ecosystem dynamics. Although, to be used in management, model uncertainties, together with their magnitude and sources, need to be assessed. In this work we applied a 1D coupled physical-ecological model to a deep Mediterranean reservoir to determine if the uncertainty in physical predictions would affect the ecological ones, and then quantify this uncertainty. Sources of uncertainty include light penetration in the water column, inflow mixing and geometry, and boundary conditions at free surface (e.g. wind speed and air temperature variability). Uncertainty evaluation was conducted through Montecarlo simulations. Results show that uncertainties in the physical model propagate to the ecological results. Depending on the magnitude of error accepted in physical prediction the prediction on phytoplankton abundance and on the timings of the successional changes in phytoplankton composition will change. Thus, the evaluation of uncertainty is essential to properly express model results and to guide management decisions and monitoring programs.