Sed_export's units - SDR model

[BiancaBendito]

Hello. Good night everyone.

I read all the topics about interpretation and units of SDR model outputs. However, I need to confirm that my interpretation is correct regarding the units and scales of analysis of the results.

Considering the main results of the model (rkls.tif, sed_export.tif, sediment_deposition.tif, usle.tif, sed_retention.tif and watershed_results_sdr.shp). And considering that the results have two units: tons/pixel for pixel scale and tons/basin for basin scale my question is:

For pixel scale analysis (in-site) I must convert the sediment_deposition.tif, usle.tif and rkls.tif files from tones/pixel/year to tones/hectare/year for analysis (such as observing in absolute terms the differences between different land uses in the basin).
So for basin-scale analysis (eg for scenario comparisons) I use Arcgis’ zonal tool to calculate the sum for the entire basin. And the sed_retention I should not convert because the analysis is relative. Are these statements correct?

My main question is about sed_export.tif.
For pixel scale analysis should I convert it from tones/pixel/year to tones/hectare/year as well?
And for off-site analysis should I convert too?

I am comparing the model result with an observed data in tons/year from a fluviometric station (exp-obs).

The model generates the watershed_results_sdr.shp file which shows a sed_export in tons/watershed (exp-mod) . Can I compare these results directly (exp-obs and exp-mod)??
Or do I need to convert the sed_export.tif file from tones/pixel to tones/hectare to tones/hectare, then use Arcgis’ zonal tool to acquire the sum for the entire basin and only then compare that sum with exp-obs??

I hope I managed to demonstrate my doubt. My English is not very good.

[swolny]

Hi @BiancaBendito -

Converting sediment_deposition.tif, usle.tif and rkls.tif to tons/hectare/year may indeed be appropriate for your purposes, but it is not required. For example, if you are comparing sediment export between different sub-watersheds, you might want to use tons/hectare/year, so that you have taken into account the area of those watersheds, since it’s common that larger sub-basins have higher export simply because they are larger in area.

On the other hand, if you are comparing pixel values between two scenarios, it is probably not necessary to convert to tons/hectare/year, if the pixel sizes are the same in both scenarios. If you are comparing against observed data in tons/year, you should also be able to just sum the sed_export.tif values, without converting to tons/hectare. If you do convert to tons/hectare, you’ll just need to do an extra step of multiplying the sum by area to get the total for the basin, which should be unnecessary.

For the case where you’re quantifying the difference between different land uses, it does seem useful to calculate tons/hectare/year, if you want to understand which land use classes are contributing more or less sediment on average, again while controlling for the fact that some land use classes cover more area than others.

Yes, you can use ArcGIS’ Zonal Statistics tool to calculate the sum. The model also calculates the sum for several parameters within the watershed layer that you provide as input, and provides that in the output watershed_results_sdr.shp.

And correct, the sed_retention values are relative, not absolute values. It’s funny that you ask about these, as we are getting ready to release an updated version of the SDR model that removes these relative outputs (that are not well documented), and includes two new results that quantify the sediment retention service in a more useful way. For now, using sed_deposition is the output to use for quantifying how much sediment is retained on a pixel, but note that it only includes the erosion from upslope, not the erosion originating on the pixel itself.

Your English is great, and your questions very clear. Hopefully my responses are too!

~ Stacie