Aquaculture CRSP
Work Plan 7, DAST Study 5
John P. Bolte and Shree S. Nath
Department of Bioresource Engineering
Oregon State University
Corvallis, Oregon, USA
Abstract
Manual parameterization of the fish growth model in the decision support
system POND© has proven to be time-consuming and complicated
because of interactions among variables considered in the model and the large
parameter space to be searched. This has limited the use of the software for
examining production potential for different pond culture species. Traditional
parameter estimation techniques typically require partial derivative
evaluations; however, because this is a difficult task for simulation models
that consider several variables, the use of an iterative, non-linear, adaptive
search method (genetic algorithm or GA) for automatic parameterization of the
fish growth model was explored. As with other parameter estimation techniques,
the objective function to be optimized must be specified. The objective
function chosen for GA testing is the minimization of the absolute error
between observed and predicted fish growth. Adequate convergence to acceptable
parameter values was obtained for the three species (channel catfish, tambaqui
and pacu) chosen to evaluate GA's as an effective parameter estimation
technique. This methodology has been incorporated directly into
POND© to enable users to rapidly customize the software for
alternate culture species and locations.
Interim Work Plan, DAST Study 2
Daniel Jamu and Raul H. Piedrahita
Department of Biological and
Agricultural Engineering
University of California
Davis, California, USA
Abstract
Enhanced nutrient cycling through waste reuse and sustainable production
are the major objectives for integrating aquaculture and agriculture. This
report describes the initial steps in the development of a simulation model to
analyze the effects of integrating aquaculture with agriculture on nutrient
cycling, whole system productivity, and of the impacts of various management
actions on the potential for enhancing pond sediment quality. The model
consists of three modules: fishpond, crop, and terrestrial soil nitrogen.
Inputs of nitrogen into the pond include feed/fertilizer and influent water and
outputs from the pond include uptake by fish, effluent water, and removal of
pond sediments. The three modules are linked through the use of sediment from
ponds as crop fertilizer and/or the use of wastes from crops as feed/fertilizer
to aquaculture ponds. Preliminary results demonstrate that feed quality and
digestibility of feed need to be considered to improve overall estimation of
organic matter and nitrogen production in the fish pond, and of fish growth.
PD/A CRSP
