Improving the predictive equation for the dispersion coefficient D0 in an
estuary mouth
Many people in the world live close to alluvial estuaries. Salt intrusion can be a problem for
agriculture and drinking water. It is therefore important to have a preferably simple analytical model
to understand the processes that influence salt intrusion, which allows us to analyze the effect of
environmental change or human impacts. Such a model was developed by Savenije (1986) and
adjustments were proposed recently by Kuiper & Van Rijn (2011).
Salt intrusion is determined by advective and dispersive processes. The dispersive processes, or
mixing processes, are described in the models with the help of a dispersion coefficient. The
dispersion coefficient D0 in the mouth of an estuary is often used to calibrate the model. With a
predictive equation for this coefficient, information about just the geometry of an estuary can be
enough to say something about the salt intrusion.
To reach the goal of this research data of different estuaries is needed. Old data is available from for
example Western Scheldt, Pungue and Maputo. However, new data will be gathered during a
measurement campaign in Malaysia. Salt intrusion will be measured by a moving boat method in two
estuaries. The boat will move with the tide inland from the estuary mouth and on several locations
the salinity will be measured. Geometry of the estuary will also be measured during this campaign.
Genetic programming is a technique that can help in processing the large amount of data. This
technique, also known as symbolic regression, is based on natural selection. Possible relations are
defined and a selection is made based on an objective function, like the root mean squared error.
The best relations will survive and will create new offspring. The old relations can mutate, combine
or survive completely and after a few generations the best ones will have survived. This technique
will be used in deriving empirical relations and hypotheses. These relations can be used in deriving
more physically based relations, for example by the use of energy balances in an estuary. Genetic
programming will be used as an hypotheses generating tool.
The goal of this research is to present in the end a new predictive equation. In this way the
applicability of the analytical models should be improved and salinity intrusion can be assessed. A
good predictive equation for this coefficient can be an improvement in the management of estuaries.
Literature
Kuiper, K., & Van Rijn, L. (2011). Analytical and numerical analysis of tides and salinities in
estuaries;part II:Salt distributions in prismatic and convergent channels. Ocean Dynamics.
Savenije, H.H.G. (1986). A one-dimensional model for salinity intrusion in alluvial estuaries. Journal of
Hydrology, 85(1-2), 87-109.