SourcePlugin.CSIRO.dSedNet Dynamic SedNet

Plugin was written by Andrew Freebairn, it is based on Scott Wilkinson's et al paper "Development of a time-stepping sediment budget model for assessing land use impacts in large river basins".  It also borrows from the Dynamic SedNet plugin developed at DERM by Robin Ellis (Unlicensed) and Ross Searle (Unlicensed) (now at CSIRO).

General Info

Published

Cuddy, Susan; Weber, Tony; Cetin, Lydia; Wilkinson, Scott; Gonzalez, Dennis; Freebairn, Andrew; Coleman, Rhys; Gamboa Rocha, Antonia; Thew, Peter; Rahman, Joel. Exploring options to manage sediment loads to Western Port: further development and application of dSedNet in an urban-rural dominated catchment. Canberra, Australia: CSIRO; 2019. https://doi.org/10.25919/5e8628274263a

S.N. Wilkinson, C. Dougall, A.E. Kinsey-Henderson, R. Searle, R. Ellis, R. Bartley
Development of a time-stepping sediment budget model for assessing land use impacts in large river basins
Sci Total Environ, 468–469 (2014), pp. 1210–1224: http://www.sciencedirect.com/science/article/pii/S0048969713008176

LT Cetin, A Freebairn, S Easton, M Sands, P Pedruco
Application of daily SedNet for modelling catchment-scale sediment generation and transport: New Zealand case study
37th Hydrology & Water Resources Symposium 2016:  Water, Infrastructure and the Environment: https://search.informit.com.au/documentSummary;dn=683996186142861;res=IELENG

A Freebairn, N Fleming, L van der Linden, Y He, SM Cuddy, J Cox, R Bridgart
Extending the water quality modelling capability within eWater Source–developing the dSedNET plugin
Goyder Institute for Water Research Technical Report Series: http://www.goyderinstitute.org/_r425/media/system/attrib/file/397/15~42_WQ_dSedNETReport_RAC.pdf

Plugin Description

The plugin is a collection of sediment and nutrient models and their associated management tools (e.g. parameterisers).

 * Work completed and\or released 

Using the plugin

Load using the Plugin manager.  Constituent generation models will appear in the Constituent Model Configuration control, Edit–>Constituent Models... Temporal parameterisation is found under the "Tools" menu, "dSedNet Gully Model Parameteriser" and "dSedNet Hillslope Model Parameteriser".  The "Spatial Parameteriser" is part of the Plugin it can be found in the "Edit" menu, Edit–>Spatial Parameteriser...

The following is a list of steps to set up a Source scenario to use dSedNet models.

1. Load the “dSedNet” plug-in using the 'Plugin Manager'
2. Use the dSednetDerivedLayers to derive useful data.  This tool removes the need to have a DEM saved within the project file.  It also generates many of the spatial parameter layers used to parameterise the dSednet models and to construct the base scenario.  Inputs required are, a hydrologically sound DEM, stream threshold (50km^2 is default), the easting and northing of the outlet cell (if one is not given all outlets will be produced at the edge of the data provided) and the file path to save results.
3. Define a Source Catchments scenario using the “Geographic Wizard for catchments”.  When defining the 'Network' use the option to 'Draw Network' and use the layers generated from the above process. 
4. Define FU areas with the use of a land-use map (raster) which covers 100% of the catchment (This is a prerequisite for using the Spatial Parameteriser)
5. Select a rainfall-runoff model and assign parameters you can calibrate it later
6. Define constituents (e.g. “Fine” and “Coarse”).
   1. You may need to define multiple sources of constituents for a given functional unit (e.g. Hill slope and Gully)
7. Assign models for catchment see constituent generation - (“Hillslope Model – dSedNet” and “Gully Model - dSedNet”) and for network models (eg stream bank model) see storage processing models
   1. then enter static model parameters
8. Assign spatial parameters to selected constituent generation models using the Spatial Parameterisation tool.
9. Execute the temporal parameterisers for associated constituent generation models

Defining FU areas

Fu areas need to be define with a spatial layer (land use) Edit → Functional Units → Assign Area Via Raster...

• The land use layer has the same number of land use codes and they are all mapped to a corresponding FU in the scenario, ie there is a corresponding values for each catchment cell

• If the land use layer changes the scenario will be made redundant.       (Tip - copy the scenario once you initial define it )

• Select the check box to 'Save spatial FU data'

Hill slope erosion component model

“The fine sediment supplied from hillslopes in each FU to the stream network is the product of gross erosion rate, FU area and a hillslope sediment delivery ratio HSDR. Hillslope supply from each subcatchment is then the sum of the contributions from all FUs in the sub‐ catchment” (Wilkinson et al. 2014). Gross daily hillslope erosion in each FU is estimated using the Modified Universal Soil Loss Equation (MUSLE) (parameters R*K*L*S*C*P), FU area is provided by Source, and HSDR is a ratio [0…1] set by the expert user. 

R is based on Yang et al (2015) - Yang Xihua, Yu Bofu (2015) Modelling and mapping rainfall erosivity in New South Wales, Australia. Soil Research 53, 178-189.

Beta base on on Yang et al (2015) - Yang Xihua, Yu Bofu (2015) Modelling and mapping rainfall erosivity in New South Wales, Australia. Soil Research 53, 178-189.

The parameter KLSC (static) (or KLSCdynamic, dynamic) can be provided to the model in a number of different ways.

1. As a played timeseries (KLSCdynamic), generated by the spatial parameteriser from a series of spatial layers
2. As a constant in full (KLSC)
3. In parts where KLSC is static and C is dynamic (KLS static and C played as a input).  Note here KLSC is only made up of that components K, L and S, the C component is the variable 

If you have previously applied a value to KLSC (or C) as a constant and are changing to use play a timeseries to KLSCdynamic (option 1 above) you will need to set the static value back to 0 (zero)

Gully erosion component model

“Gully erosion represents ongoing incision and enlargement of hillslope drainage lines and streams which have smaller contributing areas than the upstream extent of the model stream network. It also represents erosion of ‘badland’ areas of deep soil or alluvium (e.g. Brooks et al. 2009). Such erosion processes are usually caused by land use intensification” (Wilkinson et al. 2014). An input map of the current areal density of gullies, their age and cross‐section, together with relevant soil properties are used to calculate volume.

Streambank sediment supply component model

Streambank erosion is modelled along the model stream network, while channel erosion upstream of the network is represented by gully erosion. Thus, the threshold catchment area used to define the upper limit of the stream network should include all streams having significant streambank erosion that are not represented in the gully density grid. The suspended sediment supply from streambank erosion along a link (t/day) is derived by multiplying the mean-annual SedNet function of stream power and bank erodibility (Wilkinson et al.,2009).

Floodplain deposition component model

The mass of fine sediment deposited on floodplains adjacent to a link is estimated as a proportion of the incoming load, based on the proportion of discharge flooding overbank and the likelihood of settling on the floodplain considering particle size and floodplain residence time (Prosser et al., 2001b)

Reach component model

This component is a combination of the streambank component and the floodplain component as there can only be on constituent generation model assigned to a link.

Mass Transformation model

The mass transformation model is a simple scalar model that multiplies the total daily constituent mass of a link by a factor.

TotalDailyConstsituentMass = InitialStoredMass + UpstreamFlowMass + CatchmentInflowMass + AdditionalInflowMass;
ProcessedLoad = TotalDailyConstsituentMass* Factor;

Parameterisation

Gully erosion model parameters that require a value before executing the temporal parameteriser

Hillslope and gully erosion model parameters that are assigned by the temporal parameteriser

Temporal parameterisation

Temporal parameterisation has been developed within the dSedNet plug-in. Its function is to parameterise model parameters with values obtained via analysis of a model’s output time-series.  For example the long term annual average of runoff from a FU can be used as a parameter value for a model allocated to that particular FU.

The temporal parameterisation for particular models in the dSedNet plug-in has been implemented as a black box. The user only needs to set initial model parameter values and then run the parameteriser. The tool has been configured to record the required time-series while the model executes one full run (Note that changing the time period of the model run will produce different values. This may be a problem if the model is later executed over a different time period, e.g Drought vs Normal season). Finally the desired statistic is calculated and the result applied to the correct model parameter for each FU with that model.

Precondition requirements

• Temporal parameterisation is the last step of parameterisation
• Make sure the simulation period is define.  Altering the period will produce differing results
  
• The hydrological model has be calibrated

Spatial parameterisation 

Can be used to parameterise FU, Catchment and Link models, either single parameters or played time series (similar to the climate input tool)

Precondition requirements

• FU areas must be defined with a raster and the raster values must have the same number of categories as there are FUs
• If using the spatial parameteriser the layers used must be comparable with the landuse layer used to define the FU areas. Same geometry (cell sizes, number of rows and columns, lower left corner coordinates)
  
• If FU areas change the any previous parameteriation of models will be deemed incorrect and will need to be defined again
  

• There is at least one layer to process for spatially assigning inputs

• Layers are labeled ddMMyyyy

• Each layers represent continuous days (no gaps)

Source Code

Source code available https://bitbucket.org/ewater/sourceplugin.csiro.dsednet

Appendix

UI elements

Main Menu elements

Temporal Parameterisers

Plugin Models

Spatial Parameteriser

Spatial Parameteriser

FU Model Parameterised static parameter (Hillslope Model - Parameter KLSC)

FU Model Parameterise temporal inputs (Hillslope Model - Parameter C)

On completion the time series data will be found in "Data Sources" under the subheading of "Spatial Data to TS Import"

Link Model Parameterise static parameter (Reach Model - Parameter Link Slope)

Data pre-processing (TIME spatial analysis - dSednetDerivedLayers model)

Which can be found in Tools-->Plugins→SourcePlugin.CSIRO.dSedNet→dSednetDerivedLayers

Data is dragged and dropped onto the spatial component of the widget.

Note: The Beta factor is calculated from latitude, ensure that your DEM projection is 'MGA'

Inputs 

• DEM raster (The geometry of this raster should be replicated for all others used, e.g. Landuse raster)
• Stream threshold (Contributing area above a stream cell)
• Easting and Northing of outlet and 
• Directory path to save outputs

Outputs

• Stream raster
• Reach slope raster (used to parameterise Gully or Reach model), at the subcatchment scale
• Reach length raster (used to parameterise Gully or Reach model), at the subcatchment scale
• Sub catchment raster (used to define the scenario)
• Network shape file (used to define the scenario)
• Slope raster
• Steepness factor raster (used to parameterise Hill slope model)
  
• Beta factor raster (used to parameterise Hill slope model)

Inputs 

Outputs