sediment retention 3.0 mass balance equation clarification REDUX

SPardoSPardo Member
edited July 2014 in Terrestrial Models
I have a follow-up to a post string from this past spring that has since been closed
http://ncp-yamato.stanford.edu/natcapforums/discussion/220#Item_4

I am also finding that the mass balance is not adding up as it should:
usle_tot = sed_export + upret_tot

Morever, I ran the model using a bunch of HUC12s as the watersheds (n=92) and my error ranged from +20% to -26%, though the average error was only 0.5%.

Is there any update or further feedback anyone can share?

- Sam

Comments

  • adrianvogladrianvogl Member, NatCap Staff
    Hi,
    In previous versions of InVEST, usle_tot = sed_export + upret_tot + on-pixel retention. The last term was not in the final output, so there was always a discrepancy in the mass balance. Is this no longer the case for the newest InVEST versions? Can the software superheroes please clarify this for me? It comes up a lot and I want to make sure we are giving the right guidance for checking the mass balance.

    Thanks,
    Adrian
  • SPardoSPardo Member
    Adrian,

    Your question begs the follow-up question:
    How is "on-pixel retention" different than "upret_tot"?

    I'm not familiar with on-pixel retention.
  • RichRich Administrator, NatCap Staff
    upret_tot is the sum of all the nutrient retention over the watershed. The model simulates nutrient flow across the landscape and models the process of natural landscapes to retain that nutrient. Since we abstract the landscape to pixels, we call it on pixel retention. upret_tot is the sum of all the on pixel retention in the landscape.

    Also here's a bonus innapropriate video of Santa Claus teaching Tiger Woods about begging the question: ;)
  • SPardoSPardo Member
    Rich,

    I was referring to sediment retention not nutrient retention, but I don't think that matters when talking about the basic concept here.

    Your video about begging the question begs another question for me -- are you admitting that you are Tiger Woods and you knew that I was going to follow-up and say that you didn't really answer the ultimate question about how to correctly calculate mass balance?

    Because I think that question still remains:
    What is the correct way to calculate the mass balance for the outputs delivered with the sediment retention model?

    Adrian suggested that on-pixel retention might be part of the equation but that doesn't make sense to me because the other 3 outputs are all summed by watershed. Comparing those to on-pixel retention is like comparing apples and oranges. The on-pixel retention when summed by watershed IS the upret_tot output, no?

    Why would I be getting such a wide range of errors deviating in similar positive and negative magnitudes?

    I actually used the nutrient retention model last fall and did not experience the same problem.

    Please have someone take a closer look and inquire further with me if need be. The error in the mass balance coupled with my unusual and seemingly flawed v_stream output is very concerning.

    Thanks,
    Sam

    p.s. The v_stream post:
    http://ncp-yamato.stanford.edu/natcapforums/discussion/308/v_stream-pixel-mask-looks-more-like-a-watershed-mask
  • SPardoSPardo Member
    After taking a closer look at several of the on-pixel outputs, particularly usle.tif and on_pixel_retention.tif, I'm thinking something went very wrong.

    The results are the opposite of what I would expect as most of the retention follows along the stream network. It's almost as if I reconditioned the DEM and added 100m elevation at flowlines, instead of subtracting 100m, thereby creating a high slope "mountainous" watershed where all the ridge lines were made up of highly erodible silt and loam.

    However, I checked the DEM and it is reconditioned properly with the stream network burned 100m lower than adjacent cells.

    Any thoughts? I may try to use v2.6 if I can't get this straightened out soon.
  • adrianvogladrianvogl Member, NatCap Staff
    Hi Rich and Sam,
    Let me clarify what I was referring to as "on-pixel retention." And in this case, the sediment model IS different than the nutrient model.

    In previous (and current?) versions of InVEST sediment model, there were two types of "retention" calculated:
    on-pixel retention, which is for each pixel RKLS - RKLSCP
    upstream retention, which is the amount of sediment generated upstream in the flow path that is retained on each pixel.

    From the User Guide: "The total retained sediment (sret_x) is equal to the sum of the sediment retained on the pixel itself due to the C and P factors as well as the sediment removed through routing filtration."

    ** this is the part I am not sure that I have right, and so I was asking for clarification -
    1) how does the model calculate the "sediment retained on the pixel itself due to the C and P factors" (is it still RKLS - RKLSCP)? and
    2) is this retained sediment now included in the upret_tot output, summed by each subwatershed?

    If it is not included (as I thought it was not in the past), then shouldn't the mass balance be

    rkls = sed_export + upret_tot

    ?
  • SPardoSPardo Member
    Ya'll can ignore my previous post -- there are some typos and I guess I understand a bit more now after continuing to scrutinize my results.

    Let's focus on the mass balance (thank you for your input Adrian!).
Sign In or Register to comment.