SDR model

Hi, There is a doubt on the results obtained after running the Sediment Delivery Ratio model. SDR model was run to obtain results for ten number of sub-watersheds and out of that for seven number of sub-watersheds, the sediment retention and sediment export values were obtained as zero values.  
I will be very thankful if any comment can be made on whether there is any issue regarding to this results and any reasons regarding to this matter. The results have shown below:





























































































ID,N,10,0 GRIDCODE,N,10,0 ws_id,N,5,0 sed_retent,N,24,15 sed_export,N,24,15 usle_tot,N,24,15
1 9 10 2746743.96792412 5474.4909853687 154440.675036703
2 8 9 3963749.40184605 26213.8036206842 462011.900261286
3 7 8 513417.816145003 954.1657695714 18019.290820431
4 6 7 0 0 620748.306815928
5 5 6 0 0 67181.3941711401
6 1 2 0 0 44373.0021965461
7 2 3 0 0 165970.855639844
8 3 4 0 0 300009.314385469
9 0 1 0 0 693109.551537711
10 4 5 0 0 466055.987595842

Comments

  • swolnyswolny Member, NatCap Staff
    Hi @ramonadias -

    In your table, the subwatersheds that are missing sed_export values do have USLE values. I've seen that before when a watershed does not have any streams defined. The model works by first calculating potential erosion from a pixel (USLE) and then calculates export by considering the path between that pixel and the nearest stream (sed_export.) If there is no stream defined, there is nowhere for the potential erosion to flow to, and so sediment export is also undefined.

    So look at the stream.tif output from the model and see if there is a stream defined within the subwatersheds that have 0 values for sed_export and sed_retent. If there is not, then you'll need to think more about the Threshold Flow Accumulation input to the model. If there are streams defined, then it's a different problem that we'll need to figure out.

    ~ Stacie
  • Hi, 
    In the stream.tif output from the model only shows a stream in the area which I have obtained sed_retent and sed_export values. And also in the sed_retention.tif and sed_export.tif output maps only show the watershed areas which I have obtained  sed_retent and sed_export values.

    Therefore what should be done inorder to define streams and to obtain missing  sed_retent and sed_export values.

    Thank You.
  • swolnyswolny Member, NatCap Staff
    As I mentioned before, the Threshold Flow Accumulation (TFA) input to the model is used, along with the DEM, to create the stream network. It defines how many upslope grid cells must flow into any particular cell for it to be considered part of a stream. Higher values of TFA mean that a larger upslope area is required to create a stream, and it will produce a stream network with fewer tributaries. Similarly, lower values of TFA produce a stream network with more tributaries. 

    The "correct" value of TFA is the value that produces a stream network that looks as close as possible to the real-world streams in your watershed. This will be different for each watershed. So find a real-world stream map, and try out different values of TFA until the stream.tif looks as close as possible to your real stream map. You can use the InVEST tool RouteDEM if you want to do this outside of the SDR model.

    Now, even after doing that, it may be that some of your sub-watersheds have a very flat area where streams are not easy to define, and where you would need to use a very small TFA, which produces unrealistic streams in your other sub-watersheds. This is a little harder to work with, and we can talk about it if needed.

    ~ Stacie

  • Hi, Thank you very much for your all  valuable comments made to tackle the issues. All those comment were really helpful to success the work.

    Thank You.
  • Hi,
    I could be able to run the SDR model successfully and now I hope to run the SDR model with different scenarios. Under that i need to run the model by hypothetically changing the LULC types in the study area. For an example, I need to determine the soil erosion rates, if the forest cover is 100%,50%,25% of the total area, like that for different percentages of LULC types how much will be the soil erosion rates. Is it correct if I change the C and P factor values which were assigned to different LULC types in the Biophysical table?
    I will be very thankful if any guidance can be provided on how these C and P factor values should be changed. 

     Here, finally I am trying to determine the forest percentage which should be maintained in order to achieve  tolerable soil erosion rate for the interested area.
    Thank You.
  • swolnyswolny Member, NatCap Staff
    Hi @ramonadias -

    When creating scenarios, it doesn't only matter how much forest cover there is, but it matters *where* the forest cover is in the watershed. For example, reforesting in a buffer along a stream is likely to be more effective at reducing erosion in the stream than reforesting on a mountaintop far away from the stream.

    So you need to create a new LULC raster for each scenario, where you change some non-forested pixels to forest. If you do this, then you do not need to change the values in your biophysical table, the model will apply the Forest C and P factors to all pixels that are Forest in your new LULC raster. For example, if a pixel is Agriculture in your current LULC, the model will apply the Agriculture USLE C value to that pixel, but if that same pixel is changed to Forest in your scenario LULC, the model will apply the Forest USLE C value.

    There are other types of scenarios where you could need to change the C/P values, such as if you are doing a scenario where there is still Forest, but perhaps the forest is older and more dense, or has been logged and is less dense. Or, if you are doing restoration to the forest, or doing new management practices to agricultural land. In these cases, those pixels remain Forest or Agriculture, but the different practices will require new C or P factors. 

    ~ Stacie


  • Hi.
    Thank you. I understood the scenario case of creating new LULC raster maps and to obtain results.
    But, it will be really supportive, if further clarifications can be made on other types of scenario which you have mentioned in your comment. As you have mentioned, if I need to restoration to the forest or doing new management practices for agricultural lands, how the C/P values should be changed?
    Thank You.
  • swolnyswolny Member, NatCap Staff
    It depends on the changes that you are making between scenarios. The easiest is if you are doing restoration such that one existing LULC type turns into another. For example, agriculture is restored to forest. Then you can just use the USLE C/P values for forest, as mentioned before.

    But, there are other changes that you might want to make that are less straightforward. Perhaps in your scenario LULC you want to do agroforestry in agricultural lands, so in the scenario LULC you have a new class that is still agriculture, but now with trees added. In this case, you can do an average (or weighted average) of the USLE C values for agriculture and forest.

    If you want to do something entirely different, like add terraces to agricultural land, then you will need to do a literature search to find out how terraces impact USLE C and P values. This will probably be different for each different management practice that you're modeling.

    Sorry if most of my responses come down to "do a literature search", but really, that's what I spend a large amount of my time doing for each project, to come up with coefficient values that are "best available" for the area of interest, vegetation type, local management practices, etc.

    ~ Stacie
  • Hi, Please can I know the extent(area) of one pixel of resulted maps in SDR model.

    Thank You 
  • swolnyswolny Member, NatCap Staff
    The pixel size of the output should be the same as the pixel size of the DEM that you used as input, since the model resamples all of the other inputs to be the same cell size before doing its calculations.

    ~ Stacie
Sign In or Register to comment.