Monitoring and parameter estimation in ecological models including fish balances for lake restoration planning with dynamic optimization techniques

VANINA G. ESTRADA; JIMENA DI MAGGIO; FABIAN GROSMAN; PABLO SANZANO; VIVIANA COLASURDO; JOSE MARIA GUERRERO; JUAN BAGLIVI; MARTA CRISAFULLI; GUILLERMO JELISNKY; LAURA FRITZ; MONICA HOFFMEYER; MARIA SOLEDAD DIAZ

Salt Lake City

Congreso; American Institute of Chemical Engineers Annual Meeting; 2015

AIChE

Algal blooms constitute one of the most important environmental problems in many lakes and reservoirs. During the last decades, several restoration strategies have been implemented to address eutrophication (Paerl and Otten, 2013; Jeppesen et al., 2012). In this sense, artificial wetlands have been used to decrease external nutrient loading from nonpoint sources and perform bottom up control on phytoplankton growth. However, biogeochemical processes that take place within water bodies delay restoration (phosphorus and nitrogen recycles, as well as nutrient release from sediments). To accelerate restoration, bottom up control strategies are complemented with several inlake strategies, such as chemical treatment, hypolimnetic oxygenation and biomanipulation (Sondergaard et al., 2007).The application of any restoration strategy requires deep knowledge of both the ecological state and the dynamics of the water body, as well as the development of reliable ecological models to plan and predict restoration effects (Estrada et al., 2011, Di Maggio et al., 2015).In this work, we address monitoring, parameter estimation and restoration planning for Paso de las Piedras reservoir (38° 22´ S and 61° 12´ W), a warm polymictic lake which is the drinking water source for two cities (population above 450,000) and the most important petrochemical complex in Argentina. Field data have been collected throughout 2014-2015 and include nutrient, phytoplankton, dissolved oxygen, zooplankton and fish biomass concentrations. Phytoplankton and zooplankton data have been collected weekly and twice a month, respectively, while fish data have been collected every three months during the first year and bimonthly in the second year (Di Maggio et al, 2014).  Collected data have been used to calibrate a mechanistic ecological model (Estrada et al., 2011) that includes dynamic mass balances for three phytoplankton groups (cyanobacteria, diatomea, chlorophyta); two zooplankton groups (cladocera, copepoda) and two size classes of local zooplanktivorous fish, as well as dissolved oxygen and main nutrients. Algebraic equations stand for forcing functions profiles, such as temperature, solar radiation, river inflows and concentrations, etc.As a third step, we address the optimal planning of alternative restoration strategies through advanced dynamic optimization techniques. We have implemented some of these strategies as different optimal control problems, by formulating the ecological water model within a dynamic optimization environment. The dynamic optimization problem is formulated within a control vector parameterization framework (PSEnterprise, 2013).Field data show recurrent algal blooms and abundance of zooplanktivorous fish, which are in poor conditions, as well as piscivorous fish. The model is being re-calibrated with these data, providing a reliable tool for restoration planning. Dynamic optimization results include optimal profiles for control variables to plan alternative restoration actions, as well as a quantitative estimation of restoration effects on the water body