URBS HYDROLOGICAL MODEL

4.4.1.1 Use of Acquired Knowledge and Knowledge Learning

The application of the knowledge from previous work and/or from experts into DSSFCMR is illustrated in this section through the system design, development and application. However, the collected knowledge on the flood mitigation in the studied river basin is limited. The reasons are: (1) there is limited knowledge available on flood forecasting and warning other than what was obtained from the current flood warning system, which was established in 1975; (2) there is some knowledge other than flood forecasting and warning, which is difficult to incorporate into this project, since this knowledge is not directly relevant to flood forecasting and warning. Therefore, the collected knowledge is not sufficient to build a knowledge base to support decision making. However, some collected was built into DSSFCMR, as described below.

Based on the collected knowledge on the flood mitigation in the study river basin, the application of knowledge for decision support are designed through two approaches: (1) application of acquired knowledge directly and (2) knowledge learning and development. The latter is obtained through the functionality developed in DSSFCMR and provides very powerful knowledge on flood events in terms of peak discharge, flood levels and flood inundated areas. The above two methods are detailed below:

• Application of acquired knowledge directly -

There is limited acquired knowledge on hydrological model parameters relevant to the Maribyrnong catchment, although it is acknowledged that such acquired knowledge could improve the power of the DSS. One such acquired knowledge was the method of

7-day mean daily Q used as an alternative method in DSSFCMR to calculate the Initial Loss from the parameter of 7-day mean daily Q (Flow). The 7-day mean daily Q is the average daily flow over 7 days prior to the storm. This method has been used for the Maribyrnong River and the user is referred to Crapper (1989) for theory of the method. However, the equation relating the Initial Loss to the 7-day mean daily Q is given below:

(4-1) where: LC = Initial Loss(mm)

Q = 7-day mean daily flow ( m3/s) 133 7 . 23 24 . 1 2− + =

Q

Q

LC

• Knowledge learning and development -

The model parameters play a very important role in the hydrological model application. They can be different for different catchments and for different storm events. The DSSFCMR allows the user to learn and accumulate the knowledge of hydrological model parameters and their effect on flood hydrographs. The user can add, update, delete or refresh these parameters in the database based on the learnt knowledge and their appropriateness for the application.

The design on model parameters is based on selecting one model parameter at a time in forming the parameter set rather than selecting a set of parameters at a time, which meant several values of each model parameter are stored in a database having different database tables for different parameters. The main advantage of this design is the flexibility of the design to build new parameters sets for the application. However, the major disadvantage is that the user has to have a good knowledge of hydrology to determine the appropriate set of parameters.

Figure 4.12 shows the design for preparing the Catchment Definition file required by the URBS hydrological model. The central and right part of the figure shows how the acquired knowledge is incorporated in the design.

4.4.1.2 Model Definition and Parameters

As outlined in Section 3.2.2, the DOS-based URBS model required the user to prepare parameter data files such as Catchment Definition file, Rainfall Definition file, and rainfall (data) files, as ASCII files, outside the URBS software tool. Therefore, the user needs to have a good knowledge of the structure and format of these data files. However, the design in DSSFCMR allows the user to easily build these data files through the developed Windows interface. The structure (or definition) of subcatchments and most physical parameters of subcatchment such as L, F, and Area in the Catchment Definition file are fixed for a catchment and they are pre-defined and stored in the Catchment Definition file in DSSFCMR (These parameters are explained in Section 3.2.1). This means the user cannot interact with the definition of subcatchments and their physical variables from the interface. However, the design in DSSFCMR allows the user to directly view the subcatchment definition information from the interface.

o Yes

Modify database tables f relevant parameters View its values for

Catchment Definition file Select a Subcatchment No Save change ? No Yes Is the parameter Initial Loss? No Yes Continue ? Select and act a

command Start Close No Yes Compute Initial Loss?

Compute Initial Loss based on knowledge in

study area Enter or select value

for one of 5 key parameters

Figure 4.12 Hydrological Model Parameters Module in Modelbase

The user can edit the model parameters in the Catchment Definition file (i.e. alpha, m, initial loss, continue loss and base flow) through DSSFCMR interface. The parameters alpha and m

are catchment specific, while IL, CL and baseflow are storm specific. However, alpha and m

may be slightly different for different storms depending on the magnitude of storm. These parameters require calibration.

4.4.1.3 Calibration

A Windows based interface was designed to calibrate the DOS-based URBS model. The purpose of this design was to make the DSSFCMR more flexible and convenient for the user to interact with URBS. However, the user does not have to calibrate the URBS model each time during the forecasting stage and the results from a previous calibration of this event can be used instead. Figure 4.13 shows the design of the interface for interactive calibration of the URBS model.