1st International Congress on Environmental Modelling and Software - Lugano, Switzerland - June 2002
Keywords
eutrophication, fuzzy systems, pattern recognition, decision support systems
Start Date
1-7-2002 12:00 AM
Abstract
The Orbetello lagoon is a shallow brackish waterbody subject to intense and diverse eutrophication (phytoplancton, macroalgae and macrophytes). Periodically a large amount of algae must be artificially removed, their collection and disposal representing a considerable management cost. This paper describes the design of a bloom predictor based on the daily fluctuations of simple water quality parameters such ad dissolved oxygen, oxidation-reduction potential, pH and temperature. The task of the fuzzy predictor is to recognise the possibility that a bloom of the macroalgae population is about to occur based on the changing daily pattern of these variables. The fuzzy predictor is based on a number of fuzzy rules derived from experimental observations and expert knowledge. A whole year of hourly data was analysed and used to form the initial knowledge-base. The tests show that the inferential engine has good predictive capabilities, which could be improved when more data becoming available.
Fuzzy Prediction of the Algal Blooms in the Orbetello Lagoon
The Orbetello lagoon is a shallow brackish waterbody subject to intense and diverse eutrophication (phytoplancton, macroalgae and macrophytes). Periodically a large amount of algae must be artificially removed, their collection and disposal representing a considerable management cost. This paper describes the design of a bloom predictor based on the daily fluctuations of simple water quality parameters such ad dissolved oxygen, oxidation-reduction potential, pH and temperature. The task of the fuzzy predictor is to recognise the possibility that a bloom of the macroalgae population is about to occur based on the changing daily pattern of these variables. The fuzzy predictor is based on a number of fuzzy rules derived from experimental observations and expert knowledge. A whole year of hourly data was analysed and used to form the initial knowledge-base. The tests show that the inferential engine has good predictive capabilities, which could be improved when more data becoming available.
