Sediment Retention Model: Calibration Constant

SPardoSPardo Member
edited July 2014 in Terrestrial Models
Hi All,

I've completed one run of the sediment retention model and while I'm still working to refine several of the inputs, I'm also collecting data that I'm hoping will be of use when it comes to calibrating the results of my final run.

The user's guide mentions that the model allows for a calibration constant to be applied and adjusted via the Sediment Delivered Output, and I'm wondering if someone might help me brainstorm about the steps involved here.

First of all, when you say "Sediment Delivered" do you mean the "sed_export" output?

In any case, could someone sketch out what a step-by-step calibration of the sediment retention model would look like?

For example:
1) Collect total suspended solids (TSS) data
2) Redraw watersheds so that TSS collection points are the new pour points
3) Divide sed_export value by TSS for each watershed
4) etc, etc

That's just a quick example of the rough outline that I'm looking for and that will help me wrap my head around the process at this stage.



  • RichRich Administrator, NatCap Staff
    Hi Sam, I'm not sure how to do the calibration, maybe one of the hydrologists can help. But yes the sed_export shows the sediment delivered from the watershed.
  • SPardoSPardo Member
    Thanks Rich. I have heard from other users that the process I quickly outlined above, which results in a calibration constant, does not always work.

    I'm wondering if there is a longer, more complex process.

    Thanks all.
  • PerrinePerrine Moderator, NatCap Staff
    Dear Sam,

    I've just replied to your other questions:, pointing to some limitations of the models.
    These need to be considered when trying to "calibrate" the model.

    One major limitation explained in the User's guide is that the sediment model only accounts for sheetflow and rill erosion, while gullies, bank erosion, or landslides may be significant. In other words, the TSS you observe in the stream is more than what the model captures, so calibration may not be relevant.

    To conduct a "reality check", i.e. comparing what the model predicts to observed data, I'd suggest the following steps:

    1) Convert TSS data to annual loads: there are software like FLUX32 ( or LOADEST (, that help you do that

    2) Estimate the contribution from other sediment sources (gullies, stream bank and mass erosion) to the sediment budget, to isolate the load coming from sheetwash and rill erosion (which is the only source represented in the model)

    3) Draw watersheds so that their outlet is the sampling point corresponding to your TSS data

    4) Compare the InVEST sed_export (ton/yr/watershed) to your observed loads.

    A mismatch between the two values could be due to: unit conversion (always good to double check!), errors in observed values (large uncertainties in the conversion of point TSS data to annual loads), errors in parameter values, or limitations in the processes represented by the model.
    For parameter values, it helps to estimate the possible ranges and conduct simple sensitivity analyses to see if these could explain the mismatch.

    I hope this helps!

  • SPardoSPardo Member
    Helps a lot. I haven't tried to calibrate yet, I was just preparing to.

    Being made aware of the conversion methods you list in #1 above is tremendous.

Sign In or Register to comment.