Abstract: Anzali wetland, which is located in north part of Iran, is one of
the most important ecosystems in the world considering economical and
environmental features, and is in the list of Ramsar Convention. Receiving
wastewater with minimum treatment caused increasing water pollution in the
wetland. In this regard, having enough and accurate data and precise
interpretation are necessary which may help water quality management. There are
many methods for interpretation of data. In this work, prediction of COD
parameter in Siakeshim station in south part of the wetland was studied using
Artificial Neural Network (ANN) considering effects of five parameters TP, TN,
(NO_3-N), (NO_2-N) and (NH_4-N) as the most important features of nutritional
materials and creation of utrification and also predictors of pollution. The
ten years average monthly data of five mentioned parameters are applied as
input of Multi Layer Perception (MLP) models. The results of the study showed
that using MLP methods obtained precise predictions for COD parameter and also
the rate of effects of each input parameters in water pollution. Hence,
prediction of water quality using ANN model may be useful for water quality
planning and management.
Keywords: Utrification, Anzali wetland, artificial neural network, multi layer perception
