Determining the need
Before we build a model, we must know what we want the model to do.
We may need the model for multiple reasons, including to:
- calculate our compliance with a sustainable diversion limit or annual permitted take, in line with a water resource plan
- prepare or update a water sharing plan
- calculate long-term average annual environmental limits in line with a water sharing plan
- inform a regional water strategy.
A computer model represents the real world, but no model can replicate the real world perfectly. Every model has limitations.
Models at large scales allow us to understand region-scale processes. They can help us make regional policy decisions. However, they aren’t suited to supporting local decision making, such as how large a road culvert should be.
Defining design criteria
We apply model design criteria to our river system models. These ensure our models are fit for purpose for river management and water-sharing decisions. We report each model’s success against these criteria.
- represent the key physical and management processes that affect water availability and sharing within the river system. This ensures our model reflects:
- spatial and temporal variations in water availability
- volumes and patterns of water demands
- water management arrangements and practices operating in the river system.
We document important assumptions we make when building a model and describe the impact they may have on how we interpret results - maintain the water balance. This ensures the model conserves mass. All inflows, outflows and changes in stored water must balance over all scales and any time step from one day to 100 years. Water should not appear in or disappear from the system without evident reason
- be applicable over climatically variable periods. The model must perform satisfactorily across any climate conditions likely to occur in the system. We must be confident the model is suitable for scenario modelling and will represent the effects of climate change and variability
- generate results for specific locations, across river reaches and for whole-of-river systems. The model must simulate processes at a spatial resolution that allows us to check performance and report behaviour of model components both locally and across the river system
- allow reporting across timescales from daily to annual. The model must simulate daily processes to represent flow variability at a resolution useful for ecosystem processes, water management rules, water access needs (eg irrigator access to high flows) and other statutory reporting requirements. It must also allow aggregation for monthly and annual reporting
- capture historical seasonal water use at reach and valley scales. To meet statutory requirements, including annual permitted take assessments for Murray Darling Basin Plan reporting, the model must simulate annual water use under a range of climatic conditions
- be updateable and extensible. We must be able to update and expand the model as better data and methods become available.
Model concept and scope
We conceptualise and scope our models before we build them to be sure they meet our needs. This includes asking questions about the planned model, such as:
- what scenarios might we run?
- what kinds of data do we want the model to output?
- what are the geographical boundaries of the river system, and what features does it include (eg wetlands)?
- what processes do we want to model in detail, and do we have enough data to do so?
- will we model irrigators individually or as groups? What kinds of irrigation processes will we model?
- what water management processes will we model? How will we represent environmental water management?
Data considerations
When we plan the model we assess data availability, existing models and related studies. This includes determining:
- the quality of available data
- whether we can use any existing models of the river system or related systems (eg vegetation or bird response studies) to inform the new model.
If possible we visit the river system, to meet with key stakeholders, river operations staff and data collectors, and to help us understand the data in context.
We also assess catchment maps for a better understanding of the river system’s geography.
Learn more about data types and the data components we use to build models.
Model schematic
When we plan the model, we create a model schematic. A schematic is a ‘map’ of the model that describes the relative location and types of system water flows, use and management.
We start with a simple schematic we develop by looking at maps and talking to experts. We include the final schematic in our reporting so the scope, resolution, and location of data output locations used in calibration can be independently reviewed.
We develop a node-link diagram as a guide for building the model in our modelling software. Find out more about submodel choice.
Our model schematic also highlights any limitations that might affect the model’s fitness for purpose.
Selecting submodels
River system models contain many smaller models, to represent specific relationships or components within the wider model.
These are called submodels.
River system submodels might include:
- a catchment runoff model
- a single-river flow-routing model
- a model of irrigation extraction by a group of irrigators
We choose what submodels to include based on data availability, system characteristics, and the kinds of information outputs we need.
Project management
Building a successful model is a project. It requires good planning and sound record-keeping.
We create a project plan for every model build. The plan will:
- identify the appropriate modelling method
- identify the model’s objectives and use
- set a project budget
- apply a project schedule and timeframes
- identify stakeholders and maps planned engagement
- assign ownership of the completed model
- document quality assurance processes and identify an independent reviewer.
Good project planning helps ensure that our work is transparent and can be audited.
If we find a problem with the model or learn something new that requires a change of approach, we update our project plan to reflect this.
Stakeholder engagement
When we build a model, we engage with relevant government and non-government organisations and individuals. Open, respectful engagement helps build confidence in our models and their results.
Engagement also helps us conceptualise a model by helping us understand stakeholder interests and questions.
Engagement can provide context for model data. For example, stakeholders might be able to tell us how drought or structural adjustments have affected crop areas and types.
We might establish a technical reference group or expert reference group to guide the creation of the new model. For scenario modelling or smaller scale modelling projects, we might develop a stakeholder engagement plan.
We might also work with stakeholders to design model interrogation tools, such as data visualisations, that suit their needs.
We document stakeholder engagement and outcomes.
Managing the model
Good model management supports transparency around our modelling practices. It also helps ensure consistency in modelling practice nationally and, in particular, across the Murray Darling Basin states.
Learn more about model management from the Australian Modelling Practice Notes.
Scenario modelling
Scenario modelling helps us understand river system outcomes in response to particular conditions. We use scenario models to ensure we meet Murray-Darling Basin Plan requirements, and assess compliance with water sharing regulations.
Find out more about the scenario models we use to assess compliance