Practice note: Modelling a reach water balance

Practice note: Modelling a reach water balance

This practice note is one of a set developed to provide consistency and transparency of river system models being used within the Murray-Darling Basin. The notes cover modelling practices, such as naming conventions for folder structures, to model methods, such as for flow routing and residual inflow estimation, and have been developed through a collaboration between the MDBA and Basin States.

Produced in collaboration with:


This practice note, 'Modelling a reach water balance', describes the general principles and a high-level method for calibrating a reach water-balance model. It details the different components of the reach water balance that should be considered during calibration and aims to:

  1. Highlight the importance of reach conceptualisation.
  2. Provide advice about the components of the water balance that should be considered during reach calibration.
  3. Detail methods for determining acceptable values for modelled reach water balance components (e.g. residual inflows).

Background

River systems models are water balance models designed to evaluate alternative water management strategies or policies for water sharing. The components of the water balance that are important vary between reaches and how these fluxes are included in a reach water balance will be determined by data available to model these fluxes and how important these fluxes are for water management decisions the reach. Figure 1 shows the range of the fluxes that may need to be considered when modelling a reach water balance.

The conceptualisation of the reach is a particularly important component of the reach calibration process, as it allows you to decide on the key features and processes of the reach that are essential for water management and the data available to model these processes.

In most river systems, the data available to explain various components of the water balance are limited and will not cover the full range of conditions. For example, the available streamflow record may not cover the full range of climate variability, or there may be diversions (such as floodplain harvesting) that cannot be directly metered and require estimation via modelling.

Figure 1: Fluxes that may occur on a river reach

General principles

  1. Reaches should be defined based on the location of long-term gauges if available. Intermediate gauges should be used to break fluxes (e.g. losses or routing) upstream or downstream of the gauging point if this is considered necessary. The Flow Data practice note provides some advice about the selection of primary gauging stations for flow calibration.
  2. The quality of available data, key catchment processes and uses of the model should be considered when conceptualising a reach water balance model (see the Reach Conceptualisation practice note).
  3. The Source model should contain explicit modelling of the fluxes that are important on the reach. The representation of these fluxes should be realistic while being kept as simple as possible. At a minimum, the reach water balance model should include any known diversions and an estimate of net evaporation. In most cases, the reach water balance will also include routing, residual inflows and a lumped loss relationship.
  4. A reach model will be a collection of models representing the different processes (fluxes) occurring along the reach that are important for water management within that reach. These flux models may be calibrated to observed data or may be set to reference values when data is not available or not suitable.
  5. A broad range of information (multiple lines of evidence) should be used to inform the conceptualisation and development of a reach water balance model and the demand models used in the reach.
  6. A systematic approach should be used when calibrating a reach water model. The calibration should focus on periods with minimum data uncertainty for the different model components. Not all reach flux models need to have the same calibration period.
  7. Where breakouts from the main stem of the river are required for modelling either irrigation take or environmental assets or just significant losses to the landscape, these should be included in the model.
  8. Uncertainty should be tested by using a reliable calibration and validation procedure so that the model is appropriate for use in water planning and management.
  9. The assessment of the final calibration of a reach water balance model is based on analysing flows at the downstream gauge or gauges. Representations of individual flux components may be calibrated to other data sources if available. However, the overall combination of fluxes should still be tested against an ability to replicate the flows at the downstream gauge(s).
  10. A reach water balance should be presented, detailing the fluxes and their magnitude for the reach in the reach calibration report.
  11. Reporting on the calibration of the reach water balance should include, an assessment of how suitable the model is for its intended purpose, and, how well the modelled flow matches the recorded flows at the downstream gauge (as defined by appropriate metrics).

Figure 2 is an example of a typical layout for a regulated river reach with no breakouts or gauged tributary inflows, with nodes representing known diversions and losses, a flow correction function and residual inflows on the reach between gauges. Reaches within a regulated river system vary in complexity, particularly in the flatter parts of the landscape, where breakouts and returns move water across the floodplain, allowing for unmetered diversions such as floodplain harvesting during wet periods.

Figure 2: Typical layout for a regulated river reach

Recommended high-level method

Reach model conceptualisation

  1. Define the reach extents based on the location of long-term gauging stations (see the Flow Data practice note which provides some advice about the selection of long-term gauging stations).
  2. An assessment of the quality of the data available for calibrating the different components of the reach water balance model. The conceptual model of the reach should be commensurate with the data availability and knowledge of the fluxes that are important for water management.
  3. Identify the most appropriate climate station or stations for the reach. The Climate Data practice note provides advice about the selection and source of climate data.
  4. Identify the main fluxes within the Reach. When conceptualising a river reach the following needs to be considered:
    1. The location of routing, demands, residual inflows and flow correction node (lumped loss node).
    2. The magnitude of the flux compared to other fluxes in the reach (e.g. all river reaches may have some surface-groundwater interactions. However, these will only be modelled where they significantly impact on the behaviour of the surface water in that reach).
    3. How irrigation demands will be represented in the full calibration model (time series, crop model, etc.) and during scenario modelling.
    4. How other demands (urban, environmental) will be represented in the complete flow calibration model (recorded, estimated, ignored) and in the simulation (long-term) model (urban demand models, environmental demand models, patterns, functions, time-series). If demand models need to be calibrated before they are tested in the full calibration model, the appropriate practice notes should be consulted on how to undertake this calibration.
    5. How and if seepage needs to be included.
    6. Any breakouts and returns that need to be included in the flow calibration and modelling of diversions from floodwaters.
    7. Identification of any floodplain harvesting farms and their different water sources.
  5. Collate data required for each component of the water balance. In some cases, multiple lines of evidence may be required to determine if the final estimates of the flux being modelled are within an acceptable range.
  6. Select the calibration period for the models representing the different processes and fluxes. Selection of the calibration period will be based on available data and consideration of the range of conditions that the model may be used for.
  7. The output of the model conceptualisation will be a node-link schematic that should consider the interactions between the different fluxes on the reach. For example, the location of a residual inflow will impact the water available for water users, or the location of breakouts and the number of breakouts modelled may be important for floodplain harvesting. Where possible, the location of fluxes in the model should represent their relative physical locations in the real world.

Model build for flow calibration

  1. The model in Source should be based on the node-link schematic developed during the conceptualisation phase.
  2. The initial model build should include all known losses, breakouts, returns etc.
  3. For flow calibration, metered diversions should be used where these are available to represent demands on the main stem of the river.
  4. The Source model should be developed to contain the appropriate nodes and links for modelling of the fluxes that are required on the reach (e.g. town water supplies, irrigation demands (including Floodplain Harvesting), surface-groundwater exchange, breakouts, tributary inflows)
  5. The Source model should follow the naming conventions and data structures detailed in the appropriate practice notes (see the Naming Nodes and Links, Naming and storing functions and variables, Folder Structure for Data Sources practice notes).
  6. Following the development of the Source model, model parameters should be systematically determined for the different processes and fluxes that are required for the reach.

Routing calibration (Routing Calibration)

  1. Routing should be undertaken using the piecewise linear storage model unless there is a good reason to do otherwise. When there is a good reason, this should be documented and justified.
  2. Piecewise relationships should be limited to 10 index flows in the absence of a compelling reason for more.
  3. Travel time for high flow events should be set based on observed data.
  4. Routing relationships should be flat at the start and the end.
  5. Optimisation programs can be used to determine the travel time for appropriate index flows.
  6. An assessment of the rating table and cross section should be used to determine the index flows.
  7. Routing calibration should aim to achieve the best match of the timing and attenuation of flows and the downstream flow gauge.


Estimating reach net evaporation

  1. Each routing link should contain an estimate of net evaporation, a length and rating curve for the reach should be included in the model to allow the average surface areas for different discharges to be determined.
  2. Where estimation of the net evaporation is limited due to the difficulty in determining a rating curve for the reach, it may be more appropriate to incorporate the net evaporation component into the lumped loss.

Determining breakout relationships

  1. Initial estimates of the flow/effluent relationship should be made before the calibration of the reach.
  2. Sources of information on flow/effluent relationships may include:
    1. Hydrodynamic Model.
    2. Understanding the flood runner network within the reach.
    3. Local knowledge.
    4. Remote Sensing.
  3. How flow paths could be amalgamated should be considered in the conceptualisation phase. When deciding on the level of detail/complexity in the representation of the flood runner network, the following should be considered in the decision to combine flow paths:
    1. Are flow paths in the same direction?
    2. Are there significant connections along the length of the flow paths?
    3. Can floodplain harvesters access the flow paths?
    4. Do the flow paths carry sufficient water to warrant them being modelling individually?

      If the answer to Questions a, and, b above, is Yes, and Questions c, and d, is No, then amalgamation of the flow paths is recommended.

  4. The decision to represent flow paths specifically, or in an amalgamated form, or at all, should also consider what is required to support management decision making.

Determining returns from breakouts

  1. Location of return from breakouts is required in the conceptualisation, where there is a breakout in a reach that returns further down the system, this should be documented in the conceptualisation.
  2. The breakout may not return to the river system, in which case it is the equivalent to the loss, but water in the breakout may need to be accessible for floodplain harvesting.

Estimating surface-groundwater interactions

  1. Where surface-groundwater interactions are required consideration should be given to the best approach for modelling these interactions.
  2. Where groundwater models are available, these should be used to inform the development of the model for representing the surface-groundwater interactions in the Source Model.


Estimating Residual Inflows (Estimation of residual Inflows and Bounds on mean annual inflow estimates for residual inflows)

  1. Residual inflows should represent the flow of the indirectly gauged area between the upstream and downstream gauges.
  2. Residual inflows can be determined using a number of methods, including the use of a rainfall-runoff model.
  3. Residual inflow estimates should be checked to ensure that the value is plausible on a mean annual basis when compared to nearby similar catchments or independent runoff yield datasets and calculation techniques. Figure 3 provides a useful starting point for determining if your residual inflow is within the plausible range.
  4. The estimate of the residual inflow may need to be revised following calibration of the flow- loss relationship.


Figure 3: Estimated mean annual runoff across the basin, based on Teng et al. (2012)

Estimate lumped outflow - loss relationship

  1. The reach loss node (or nodes) will include all losses that have not been explicitly modelled.
  2. A relationship between flow and loss should be developed based on periods with minimal inflows and extractions.
  3. The loss should not decrease with increasing flow. In a reach with multiple explicitly represented fluxes, the loss node becomes more equivalent to an unaccounted difference adjustment similar to that used by river operators. It is theoretically possible for decreasing unaccounted differences to occur due to the flux methodologies. However, Source requires that loss nodes do not go backwards for arithmetic reasons. If this occurs, the use of an average loss across a broader range of flows, or no loss at all, should be considered.


Incorporation of demand models into the reach water balance

Following flow calibration using observed diversions, if demand models are going to be used to represent some of the key fluxes, these should be calibrated and incorporated into the final calibrated reach water balance model.

Companion Practice Notes