Practice note: Storage Inflow Derivation

This practice note is one of a set developed to help provide consistency and transparency in the river system models used within the Murray-Darling Basin. The collection of practice notes aims to cover, data requirements, setting up and building a model in eWater Source, model calibration, and, reporting. The MDBA and the Basin States are working collaboratively to develop this collection of practice notes to improve the transparency of river system models developed in eWater Source across the MDB.

Produced in collaboration with:

This practice note, 'Storage Inflow Derivation', describes the general principles that you should follow and a high-level method that you may choose to adopt when estimating storage inflows during model calibration. These inflows then become part of the data for inclusion in the model and should be used in model calibration as outlined in the Estimation and Assessment of Inflows practice note. 

Background

A back-calculation technique referred to as storage inflow derivation (SID) can be used to determine storage inflows. A SID uses a mass balance method based on the recorded water level, rainfall, evaporation, seepage, direct extractions from the storage and storage outflow to determine the inflows. 

For any given storage, the modeller collates and prepares data to allow the inflow to the storage to be derived taking into account any changes in storage configuration, significant changes to the operation of the storage, or changes in the measurement of storage level or outflow data. This practice note covers the following:

  1. General principles that should be applied when calculating storage inflows.
  2. The data that should be collated and checked before deriving the storage inflows.
  3. The steps the modeller may use in determining storage inflows using eWater Source.

During the derivation of storage inflows, solving the following equation determines the inflow for a given timestep:

StorageVolumeEndOfTimestep = StorageVolumeStartOfTimestep + StorageInflowVolume + RainfallVolume - EvaporationVolume  - SeepageVolume - ExtractionsDirectlyFromStorage - StorageOutflowVolume

This leads to:

StorageInflowVolume = StorageVolumeEndOfTimestep - StorageVolumeStartOfTimestepRainfallVolume + EvaporationVolume  + SeepageVolume + ExtractionsDirectlyFromStorage + StorageOutflowVolume


Figure 1: Fluxes on a storage

General principles

  1. Data should be collated from all appropriate data sources
  2. QA checks should be undertaken of the different data sources to facilitate a consistent set of data that best represents the historical releases, extractions directly from storage, storage volumes, rainfall, evaporation, and, level-volume-area relationships. 
  3. Derive the storage inflows using:
    1. the level-volume-area relationships applicable over time
    2. the climate data that the modeller intends to use in the final model
    3. the seepage information that the modeller intends to use in the final model
  4. Remove any negative inflows from the inflow time-series using an approach that
    1. preserves the overall mass balance
    2. provides realistic inflow estimates
  5. Undertake validation of the storage inflows.

The suggested method for the derivation of storage inflows

Data collation and assessment

The modeller should collate and assess the following data. Note that there may be multiple different sources of information on storage levels, storage outflows, direct extractions from storage, and storage infrastructure.

Storage data

  1. Different sources of information should be compared, and a decision should be made on the Level-Volume-Area relationship/s that will be used in the model. This includes:
    1. Understanding the dates of significant changes in the storage configuration and the level-volume-area relationship that would have applied before and after the change.

    2. Deciding if different level-volume-area relationships are required for the different periods identified. If periods of different level-volume-are relationships are identified, then the storage inflow derivation will need to be run in multiple parts.

  2. Observed Level/Volume:
    1. Data from different data sources (e.g. recorded data, operational sheets) will need to be compared and a decision made on which data will be used (or the combination of data that should be used).

    2. Data should be obtained for 9 am to align with SILO climate data.
    3. Data should be visually assessed to identify any anomalies.

    4. Missing data should be infilled where possible.

  3. Storage release/spill and downstream flow data 

    1. Data may be obtained from more than one data source (e.g. downstream gauging sites, recorded releases from operations).
    2. Data should be compared and the impact of any residual area between the dam wall and the downstream gauge should be understood.

    3. The spill and release volume from storage should be accumulated over the same period as the climate data
    4. Missing data should be infilled where possible.
  4. Seepage data - any available information should be obtained on seepage rates from the storage. If no information is available it should be assumed to be zero. 
  5. Any remaining periods of missing storage data should be infilled to obtain a continuous time series. After the inflows have been calculated,  periods where data are missing, are removed from the calculated storage inflows and infilled using the methods outlined in Estimation and Assessment of Inflows.

Climate data

Extractions directly from storage

Estimation of Storage Inflows

  1. Estimation of storage inflows can be undertaken in eWater Source or externally to eWater Source (e.g. in Excel). The steps to use eWater Source are described below. Both approaches involve solving the equation shown in the background information. 

  2. Once the initial estimates are made, any periods where there is known missing data (levels, releases, pond extractions) should be removed.
  3. Following this the time series of storage inflows should be assessed for:

    1. Large negative values - these may be indicative of a timing problem in your input data
    2. Extended periods of negative values - these may be the result of missing releases or extractions.
  4. When negatives are not the result of data issues that can be resolved, the negative inflows should be removed using an agreed smoothing approach that maintains the overall mass balance, and, produces a realistic inflow hydrograph.
  5. The final inflow time series should be run through the model with forced outflows and the observed and modelled storage levels should be compared to ensure they match.

Extension of Storage Inflow Data for Scenario Modelling

Methods for extending storage inflows to cover the period of flow calibration or for scenario modelling are outlined here: Practice note: Estimation and Assessment of Inflows

Using Source to Derive Storage Inflows

The following steps will allow you to estimate inflows using eWater Source. Note that when inflows are greater than available storage volume, this method will underestimate inflows.  This underestimation of inflows may occur when determining inflows to reregulating weirs. 

  1. Build your SID model network. This will consist of a storage and the minimum flow node downstream of your storage.
  2. Set the storage table (the level-volume-area table). If more than one storage table is required, set up scenarios for each storage table.  
  3. Set the full supply level, initial level (based on the first observed level for your period for SID) and dead storage (can be set to zero).
  4. Add climate data (rainfall and evaporation).
  5. Do not configure a spillway.
  6. Configure the storage outlets to have large capacity.
  7. Add the gauged storage level data and tick the "apply unaccounted difference to storage level" checkbox
  8. Record the following outputs:
    1. Storage downstream flow
    2. Storage level
    3. Storage gauged level
    4. Storage volume
    5. Storage unaccounted differences.
  9. Set the minimum flow requirement to the observed outflow from the storage (controlled releases + spills + extractions directly from storage).
  10. Configure the model to run over the appropriate date range.
  11. Run the model.
  12. Check the output:
    1. Storage level - check that the modelled storage level matches observed level
    2. Outflows  - check the storage downstream flow matches the observed downstream flow
    3. Recorded storage unaccounted difference - these are the storage inflows and should be checked for:
      1. Large negative values - these may be indicative of a timing problem in your input data
      2. Extended periods of negative values - these may be the result of missing releases or extractions.

Small negative values are likely to be unavoidable and should be addressed by smoothing the data to give the final storage inflows.

Links to the relevant section of the eWater Documentation

Storage node