Abstract:
In view of the protection of Gaza coastal and marine environment from further
deterioration and reverse the process from polluted to clean marine environment, it
was necessary to analyse the seawater quality data and develop several artificial
neural network (ANN) models to predict various water quality parameters. Hence,
this study was undertaken with this objective. Even though there have been a few
monitoring and evaluation literatures on the status of seawater pollution, however to
the utmost of facts this work is the initial effort to use (ANN) technology for the
prediction of seawater quality along Gaza coast. Additionally the land-based sources
causing pollution along Gaza coastal waters have been assessed and also a
management plan for Gaza coastal and marine environment protection proposed.
All seawater quality monitoring data were inserted into Minitab statistical software
and analysed by different tools including: min, max, mean, standard deviation and
geometric mean. In addition, the Pearson correlation coefficient and paired sample t-
test were used to perceive significant water quality differences at various sampling
locations.
The core objective of this work was to investigate whether it is possible to predict the
next two weeks values of water quality parameters such as pH, electrical conductivity
(EC), dissolved oxygen (DO), biological oxygen demand (BOD 5 ), Total kjeldahl
nitrogen (TKN) and orthophosphate (Ortho-P). The data used was measured by
several water quality monitoring programs in the Mediterranean Sea along Gaza. At
the initial stage of the prediction of water quality variables, real water quality data
along Gaza coast, over a period of four years beginning from 1997 to 2001, were
collected from the Environment Quality Authority, the Environmental and Rural
Research Centre-Islamic University of Gaza, Ministry of Health and some other
organisations. The training and testing of the developed ANN assessment models was
carried out using neural network toolbox in the MATLAB. Two types of feedforward
networks have been used. They are Multilayer Perceptron (MLP) and Radial Basis
Function (RBF) neural networks.
xxiFour different MLP neural networks have been trained and developed with reference
to water temperature, wind velocity and turbidity parameters to predict pH and EC
fortnight’s values. MLP and RBF neural networks have been used for predicting the
next fortnight’s dissolved oxygen concentrations. Both networks are trained and
developed with reference to the five oceanographic variables including water
temperature, wind velocity, turbidity, pH and conductivity. MLP neural network has
also been trained to predict BOD 5 level. MLP and RBF neural networks have been
trained with five parameters to predict nutrients (TKN and Ortho-P) level along Gaza
coast. Prediction results prove that both types of networks are highly satisfactory for
predicting water quality parameters in the Mediterranean Sea along Gaza coast.
Results of the developed networks have also been compared with the statistical model
and found that ANN predictions are better than the conventional methods.
The human activities including: urbanisation in coastal areas, wastewater pollution,
coastal activities, agricultural loads, industrial pollution, influence of fisheries and
solid waste as well as debris exert pressure on the marine and coastal environment.
All these are land-based sources causing pollution along Gaza coastal environment. In
order to reduce the load of pollution in the coastal waters of Gaza, it is recommended
that the Palestinian Environment Quality Authority (EQA) shall play a major and
active role in implementing the Gaza coastal environmental management plan
(GCEMP). Therefore, EQA shall prepare the short term and long term action plans
through coordination with all the relevant stakeholders.
It is hoped that ANN developed models and the proposed management plan will help
in assisting the local authorities in developing plans that are necessary to minimise the
sources that cause pollution along Gaza coastal waters.