Practice Note: Estimation of Losses

Owned by SysAdmin eWater
Last updated: 25 Aug 2024 by Alice Brown
6 min read

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, Estimation of Losses, describes the general principles that should be applied when incorporating losses into reach calibration. It is a companion note to the modelling a reach water balance practice note.

The Modelling a Reach Water Balance practice note outlines the high-level approaches to model known reach losses. This Practice Note aims to expand on the high-level reach water balance note by providing additional details on estimating known loss fluxes and lumped losses (or unidentified flux). 

Background

Reach losses are complex and may vary based on flow and antecedent conditions. The practice note, Data Requirements for Modelling Reach Losses, details the data required to model known losses. These losses include evaporation, seepage, initial losses following extended dry periods, fluxes to floodplains and wetlands, and groundwater. 

eWater Source has the functionality to allow the modelling of evaporation on Storage Routing links. This is the easiest of the known reach losses to model, and the modeller can include this flux on routing links if required. The modelling of other known losses depends on the data available to model the flux and the importance of the flux to the water management questions being addressed by the model. For example, it may be important to include an estimate of seepage in the model due to a desire by the stakeholders to see it included. The fluxes that will and will not be included in the reach water balance should be considered during model conceptualisation. 

Following the incorporation of methods to represent known reach losses, one or more loss nodes are added to the reach to account for losses that have not been explicitly represented. In some cases, these loss nodes may represent different things. For example, the modeller might use a loss node to represent dry period losses, while a second loss node accounts for any other unexplained flux. The modeller would normally place the loss node representing the unexplained flux upstream of the gauge representing the reach's downstream boundary. However, in some cases, it may be appropriate to have loss nodes at other locations along the reach or represent them as a loss/gain on a routing link.

The note aims to cover the representation of known reach losses and the calibration of the loss node that represents the unidentified flux or lumped flow-loss relationship.   Known reach losses include evaporation, seepage losses, losses to dead storage/transmission losses after extended dry periods, and losses to groundwater.

General Principles

  1. During reach conceptualisation, the documentation may cover how the known loss processes will be represented in the model.

  2. The conceptualisation may consider stakeholder expectations around which known reach losses should be included in the reach water balance.  

  3. Estimating known reach losses identified in the conceptualisation may be undertaken before calibration of unaccounted losses/unidentified flux (lumped flow-loss relationship).

  4. The time spent modelling known reach losses (e.g. evaporation, seepage, losses to dead storage) should be appropriate to the impact of the loss on the water management question being addressed.

  5. The lumped flow-loss relationship is estimated after all other fluxes in the reach have been represented. When determining the lumped flow-loss relationship, the modeller should consider the feedback between the estimation of residual inflows (see Estimation and Assessment of Inflows) and the lumped flow-loss relationship. It is important that inflows are not inflated beyond physically defensible levels and compensated for by a large lumped flow-loss relationship.

Suggested Methods for Estimating Losses

Known Reach Losses

Not all known losses will be represented in every reach. This section of the practice note provides high-level guidance about how some known losses may be modelled if they had been identified as important during the reach conceptualisation phase of the model build. 

Evaporation

  1. Evaporation can be included on routing reaches (Storage Routing) by defining rainfall, evaporation, reach length and average reach width for different flow rates.
  2. The Selecting Climate Data Practice Notes defines the appropriate rainfall and evaporation data to be used for estimating net evaporation on the reach. 
  3. The reach length should be defined based on the modeller's best estimate of the distance between the two gauges defining the upper and lower end of the routing link. 
  4. In eWater Source, the rating curve defines the relationship between flow and average river width for the routing links. This needs to be estimated by the modeller using the best available information. Several methods are available to determine the relationship between flow and river width. The modeller should consult with a senior modeller in their organisation to determine the best way to estimate different reach widths for different flow rates. 

Seepage Losses

  1. Seepage loss refers to a more localised and slower rate process than groundwater interactions (covered below).
  2. The model conceptualisation phase should identify reaches and locations where seepage is a significant component of the water balance or where the representation of seepage is important to the stakeholders. In most cases, seepage will typically be set to zero as it is a small component of the water balance as heavy clays typically dominate stream beds. 
  3. Where a seepage loss needs to be defined, the soil type and typical seepage rates would need to be investigated with direct field measurements or the adoption of reference values.
  4. The seepage loss can be incorporated into eWater Source using a function on a loss node or included on routing links where the modeller can define the seepage rate for different flow rates.

Losses to Dead Storage / Transmission Losses After Extended Dry Periods

  1. Dead storage or transmission losses after extended dry periods may be included in the model if appropriate data are available to support the modelling. 
  2. There are multiple options for the incorporation of antecedent conditions into eWater Source. The modelling of losses to dead storage aims to:
    1. Capture the high transmission losses that occur after an extended dry period
    2. Improve the rising limb of the hydrograph for events following extended dry periods
  3. Dead storage can be incorporated into the rating curve of the routing links (Storage Routing). This will allow for evaporation from these links and require dead storage to be filled before flow commences. 
  4. In some systems, larger water holes can be modelled using the storage/weir functionality. 
  5. To adequately model the losses to dead storage or waterholes, the following data are required:
    1. Level, volume area relationship (can be obtained from bathymetric surveys or water observations from space during dry periods).
    2. Flow data for events following extended dry periods.

Channel System Losses

  1. Channel system losses refer to the water lost when delivering water to irrigation districts.
  2. These may be included as a known reach loss or a loss component in the demand model.
  3. How these are represented should be discussed with a senior modeller and agreed upon during reach conceptualisation.

Wetlands

  1. Significant wetlands should have their own representation, and the modeller should avoid incorporating losses to these wetlands into the lumped flow-loss relationship.

Losses to Groundwater and bed sands

  1. In locations where groundwater interactions are insignificant, the flux to groundwater can be ignored.
  2. In places with known connections to groundwater systems, they are preferably included in the conceptualisation and not in the lumped flow-loss relationship. 

Lumped Flow-Loss Relationship

In most reaches, a single loss node will be used to account for any reach losses that have not been explicitly modelled. In some reaches, having more than one loss node may be appropriate. 

Typically, the lumped loss node would be calibrated using a reach model calibrated for routing and incorporating any known losses (e.g. evaporation) and observed diversions. From this model, the modeller would:

  1. Extract the modelled and observed flows at the downstream gauge.
  2. Calculate the loss required at each flow rate to align the modelled flow with the observed flows and to create a loss table. The probability of exceedance curve can provide a useful tool to inform the loss table.
    1. Note that if the flow data is being used directly, the difference between the modelled and observed flows may result in negative values. These must be removed or adjusted before the lumped loss relationship is determined. 
  3. Simplify the loss table to the minimum number of points to achieve the required shape. It is suggested that the curve consists of between 5 and 20 points. The loss table should ensure that (1) the loss is not negative, (2) the loss is either constant or increasing as flow increases, and (3) the increase in the loss is not more than the increase in flow (i.e. the slope of the loss curve not more than 45 degrees).
  4. Ensure that the highest flow rate in the loss table exceeds the maximum flow expected in the reach (this would exceed the biggest flow recorded and maintain the percentage loss for the largest calibration point).

References

Barma D and Varley I 2012a, Hydrological modelling practices for estimating low flows –stocktake, review and case studies, National Water Commission, Canberra.

Barma D and Varley I 2012b, Hydrological modelling practices for estimating low flows – guidelines, National Water Commission, Canberra.

Costelloe, J.F., Grayson, R.B., Argent, R.M., and McMahon, T. A. (2001) Modelling flow routing and transmission loss processes in arid zone floodplain rivers. MODSIM 2001 International Congress on Modelling and Simulation.

JH_Where does all the water go_Partitioning water transmission losses.pdf