Propagation of uncertainty in ecological models of reservoirs: From physical to population dynamic predictions
Date
2012
Authors
Rigosi, A.
Rueda, F.
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Journal article
Citation
Ecological Modelling, 2012; 247(2012):199-209
Statement of Responsibility
Anna Rigosi and Francisco J. Rueda
Conference Name
Abstract
Ecological models are widely accepted in the scientific community as tools to describe, interpret and predict ecosystem functioning. However, to be used in environmental management, model uncertainties, their magnitude and sources need to be carefully assessed. A one-dimensional coupled physical–ecological model is applied to a deep Mediterranean reservoir (Lake Béznar) to determine whether or not the uncertainty existing in physical predictions affects ecological predictions, and, then to quantify this uncertainty. The sources of uncertainty include light penetration in the water column, inflow mixing and geometry, and boundary conditions at free surface. Uncertainty in the model results was evaluated following the procedures outlined in Beven (2001), based on Montecarlo simulations. At least during summer time, the largest sources of uncertainty in the physical predictions are associated to the input variables used to construct the surface (heat and momentum) boundary conditions. Uncertainties in the physical model propagate to the ecological results. Average chlorophyll-a concentration predicted by the ecological module in the water column, their standard deviations, and the timings of the successional changes in the algal community all vary depending on the magnitude of the error accepted in the physical predictions. Our results illustrate that the analysis and quantification of model uncertainty are fundamental to properly express model results and, consequently, to optimize monitoring programmes and guide management decisions.
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
© 2012 Elsevier B.V. All rights reserved.