Spatial Resolution - from which layer? - for Water Yield and Nutrient/Sediment Retention efficiency

JessieWellsJessieWells Member
edited January 2014 in Freshwater Models
Dear InVEST developers,

could you please clarify which raster resolution is used for the pixel-level calculations in the Water yield, Sediment and Nutrient models?

The pixel resolution is crucial for setting nutrient and sediment retention coefficients (since eff_n and eff_p are expressed as % per pixel, not a value per hectare, and so will vary with the size of the pixel)

Could you please let me know if this is set by the a specific layer, e.g. the DEM or the LULC? (or by the finest resolution of any input layer?)

e.g. DEM?
Richard Sharp posted information to the InVest-users e-list, and refers to the DEM resolution as being critical:
https://mailman.standford.edu/pipermail/invest-users/2014-Janauary/000272.html

e.g. LULC?
Doug posted on this NatCap forum about the 'output resolution', and says that this is set by the LULC map.
(in versions before InVEST 2.5.6, it was instead set by the finest resolution of any input layer)
http://ncp-yamato.stanford.edu/natcapforums/discussion/96
I don't know if that means we can assume that the eff_n and eff_p should be set based on the LULC map?

(I have a 30 m DEM, and a 50 m LULC raster, so although these are both quite fine scale, they are proportionally very different from each other)

Lastly, Jon Fisher (TNC Spatial scientist) sent a really helpful post to the Invest-Users e-list, that I would like to use but am uncertain which raster to refer to for the pixel size:

Source Jon Fisher (TNC Spatial scientist, 22/01/2013) Invest-users e-list:
https://mailman.stanford.edu/pipermail/invest-users/2014-January/000273.html

" we used this equation:
100*(1-((1-eff_n100m/100)^(PixelSize2/10000)))
Where:
eff_n100m is the starting filtration coefficient (from the literature or InVEST sample data) in values from 1 to 100
PixelSize is the length (in m) of your pixels (e.g. we used 30)

So for example, since we used 30m pixels a coefficient of 80 becomes:
100*(1-((1-80/100)^(900/10000)))=13.48

Also please note that while you have to manually adjust your filtration coefficients, InVEST automatically adjusts the load values taking your pixel size into account so you can leave those as is.
"

Thanks very much for any clarification on this,
best wishes
- Jessie
Post edited by JessieWells on

Comments

  • DougDoug Administrator, NatCap Staff
    Hey Jessie,

    You are correct in that we have not been well documenting spatial resolution. We are currently working on using a universal approach of DEM followed by LULC for resolution priority.

    Currently in the Water Yield model the LULC is used for determining resolution.

    It also looks like Sediment and Nutrient (stand alone models) are using the LULC pixel resolution for outputs. I will double check with the model developer to confirm this.

    Thanks for bringing this issue to our attention, it will be something we work to make uniform and documented in future releases.
  • Thanks Doug!

    thanks for clarifying this for the Water yield model, and it would be great if you're able to double check which layer is being used in the Sediment and Nutrient models.

    ( I had a bit of a panic thinking that the sediment and nutrient retention coefficients I have used might all be wrong by factors of about 0.9 to 2.4, since that seems huge - i.e. applying the above formula to 50 m cells versus 30 m cells)

    A universal priority of DEM then LULC for all models sounds great for future releases. If it's not possible to implement this across models for the next release, I guess the main thing is just to make sure the details are clear for each individual model.

    Alternatively, perhaps the retention coefficients could be entered in the biophysical table on a standardised basis e.g. as % per hectare, and automatically converted within the program, based on the pixel size? (just to prevent user errors in calculations, or anyone not realising that they need to apply conversions)

    Much appreciated,
    - Jessie
  • Hi Doug,

    I wondered if you have been able to ask the developer of the Sediment & Nutrient models (stand alone) which layer's pixel resolution should be used as a reference when setting the eff_n, eff_p and sedret_eff? (i.e. to set the correct pixel size to express these values as % per pixel)
    Thanks
    - Jessie
  • DougDoug Administrator, NatCap Staff
    Hey Jessie!

    Thanks for writing back and sorry for my delay, I will get that to you by the end of the day. Thanks for your patience and again, sorry for the feedback gap.

    - Doug
  • DougDoug Administrator, NatCap Staff
    @JessieWells

    Just checked with the developer, looks like they are both based on LULC at the moment but he commented that in a future release they will be determined by the DEM. So, as of right now LULC, but in the future DEM and we'll make a note in the User's Guide when that changes. Eventually, and hopefully soon, we'll be able to confidently conform to a standard like you mentioned.

    Thanks for your help and use of our tools!

    - Doug
  • elenaelena Member
    Hi Doug and NatCap team

    About the output resolutions, I am checking in the user's guide and former posts in the forum but yet I cannot find an answer.

    For the InVEST 3.3.0 water yield, SDR and NDR models, is the output resolution still defined by the LULC input resolution?

    Thanks a lot!

    Looking forward to your news!

    Best,
    Elena





  • jdouglassjdouglass Administrator, NatCap Staff
    Hi @elena, in InVEST 3.3.0, Hydropower ("Annual") Water Yield uses the LULC pixel size, but Seasonal Water Yield, SDR and NDR all use the resolution of the DEM.

  • elenaelena Member
    Thank you so much Doug!!

    Best,
    Elena
Sign In or Register to comment.