 |
Physico-chemical water quality models are
developed and tested through implementation of the Mike11 river
modelling software of DHI Water & Environment. The models
aims to describe and predict concentrations of dissolved oxygen
(DO), organic pollution (by the biochemical oxygen demand BOD),
nutrients (nitrogen in the form of ammonium NH4-N and nitrate
NO3-N) and total dissolved solids (TDS), taking into consideration
advection, dispersion and the most important biological, chemical
and physical processes. All significant pollution sources are
considered. Applications include case-studies in Belgium, the
Dender and Nete basins, and in the Nile Delta.
For the case-studies in the Flanders region of
Belgium, the nutrient input loads in the surface waters are predicted
by the SENTWA model. The SENTWA model describes the spatial variation
in the nutrient loads and concentrations, their contributing sources
and their dependence on rainfall and catchment characteristics
through a conceptual model. Because this modelling tool covers
the whole Flanders territory, traditional physically-based modelling
techniques (e.g. SWAT) cannot be used. More simplified or conceptual
models are needed, which try to relate nutrient concentrations
to river discharges, catchment characteristics and statistical
information on the agricultural activities. In a research project
for the Flemish Environment Agency (VMM), the SENTWA model has
been refined such that monthly nutrient load estimates can be
disaggregated to the daily and hourly time scales, and can be
splitted in discharges and concentrations. For the discharge estimates,
a regional rainfall-runoff model has been set up for the entire
Flanders region. It is based on lumped conceptual model calibrations
for a large number of gauged subbasins.
In the Nile Delta area, a full hydrodynamic model
and a physico-chemical water quality model have been set up for
the Rosetta Branch. This has been done through joint research
activities by the Hydraulics Laboratory of K.U.Leuven and the
National Water Research Center (NWRC), Cairo (Egypt). A NATO Collaborative
Linkage Grant, given to both institutes for the period 2003-2005,
financially supported these activities. This Grant was part of
the NATO Science Programme - Cooperative Science and Technology
Sub-Programme, and allowed links to be set up between the research
at both institutes. |
|
To model the flow and water quality of surface
waters accurately, the different sources of water and pollution
have to be considered in an integrated way. It requires the different
water systems from which the pollution originates (sewer systems,
waste water treatment plants WWTPs, and runoff catchment) to be
linked with the surface water model. Most frequently applied system
models cannot be used for that purpose. They are too sophisticated
and the linking of many sophisticated models leads up to unacceptable
high calculation times and memory needs. This problem is solved
by simplified conceptual models used in a way complementary to
the sophisticated models, which are often available at the authorities
responsible for the different individual water subsystems involved.
The simplified conceptual models are to be identified (model structure
identification) and calibrated based on simulation results with
the more detailed models. |
|
The uncertainty in
integrated surface water modelling is often
high because of the large scale of the model. It is therefore
important to take the uncertainties involved in the modelling
into account. A step-wise procedure based on "variance decomposition"
has been worked out to quantify the different uncertainties sources,
which are classified in input uncertainties, parameter uncertainties
and model-structure uncertainties. Because the rainfall input
is a major source of uncertainty, the rainfall input uncertainty
analysis drew special attention in the study. After quantification,
all uncertainty sources were represented by stochastic terms,
which transfer the deterministic model to a probabilistic one.
In the case-study of the Witte Nete river in Belgium, an integrated
and probabilistic immission model has been derived for the full
river basin. Comparison was made of the different sources of uncertainty
and priorities defined for future model refinement. |
