NDR 3.7.0 Export vs. Retention

Hi there,

If I wanted to calculate total retention over my watershed, would I take the total load minus total export? In the new version (3.7.0) I do not get a total retention output like in the sediment model.


1 Like

I’m also very interested in the same question. After reading through the equations in the user manual, there doesn’t look to be an easy way to quantify nutrient retention on a cell-by-cell basis, is that correct?

You could of course subtract N or P exports from loads. But that would give you the amount of N or P from a given cell that gets absorbed somewhere between that grid cell and the stream, not the amount absorbed by each grid cell. Is there any way to quantify the latter using the NDR model? Or is the assumption that there’s too much uncertainty in the underlying biophysical processes to accurately quantify this without using more complex models?

Many thanks!

1 Like

Hi @audiae -

The retention output in the sediment model does not actually calculate the total amount of sediment retained, it just calculates the difference in erosion between the current land cover map, and a hypothetical one where all land cover types were turned to bare ground. This does not include any deposition from upslope, only values the erosion difference due to current vegetation. I believe that the software team is working on an actual sediment retention function, that will include retention from upslope pixels, but that’s not ready yet, and I’m not sure if it is also being planned for NDR or not. @rich, can you comment?

To value the service, we usually take the difference in nutrient export between the current landscape and a scenario. I don’t see an obvious reason why you couldn’t take load minus export as a proxy for retention, although I’d be careful presenting absolute values for this, as Ken notes.

~ Stacie

Thanks Stacie. Just to be 100% clear, is it correct then that the NDR model cannot calculate total retention by a given grid cell?

If that’s the case, then potentially useful nutrient retention calculations could be the amount of a given grid cell’s nutrient load that is retained somewhere downstream (i.e., grid cell-based load - export), or the same summed e.g., at watershed scale, correct?


Hi @swolny, yes, we’ll have an SDR model on the next release that calculates per-pixel sediment deposition. I don’t have anything planned for NDR, but nobody’s asked me either. :slight_smile:

And yes, @kjbagstad, if I understand your question correctly, you can always calculate per-watershed nutrient retention with the “sum of total load - sum of export” calculation.

1 Like

Thanks @rich . FYI, per-pixel nutrient retention would indeed be useful for ecosystem accounting, if the opportunity arises to tweak the NDR model to enable that calculation in the future.

Hi, I have a similiar question. @rich

In Annual Water Yield model, how can I get water retention cell-by-cell? I want to study on how much water the land can help intercept and infiltrate, rather than let the water flow quickly to the stream and to the sea. It seems I cannot. Because in the userguide it seems water yield = runoff + groundwater recharge? And it says it should NOT be interpreted at the pixel level. I just ask for sure. Maybe I should use other methods like Precitipation - AET - Runoff? Or is there any posibility in the latest version or future version?

And thank you for the new release v 3.8. Just ask for sure. Is sediment retention cell by cell the tif file named sed_retention_[Suffix].tif? (type:raster; units: tons/pixel): Map of sediment retention with reference to a watershed where all LULC types are converted to bare ground.

Thank you!

@Tookie, In the Annual Water Yield model, per-pixel retention can be found in your output workspace under output/per_pixel/wyield_[Suffix].tif, as described in the user’s guide section on interpreting the results of the AWY model.

In the InVEST SDR model, yes, the raster with pixel values representing sediment retention in tons/pixel (relative to bare ground) is sed_retention_[Suffix].tif.

Thank you for answering! But I’m still confusing. Does the output\per_pixel\wyield_[Suffix].tif (mm) mean the water yield for every pixel? If yes, why the user guide says it cannot be interpreted at the pixel level? And if I want to calculate the water retention for each pixel, can I use Precipitation.tif (mm) – output\per_pixel\aet_[Suffix].tif (mm) – output\per_pixel\wyield_[Suffix].tif (mm)?

Oh, I just realize the water yield is calculate from Precipitation - AET, so if I do Precipitation.tif (mm) – output\per_pixel\aet_[Suffix].tif (mm) – output\per_pixel\wyield_[Suffix].tif (mm), I will get zero, am I right?
So I might not use Invest models to get water retention right? :joy:

Hi @Tookie,

It’s not that you can’t use the per-pixel water yield raster, it’s just that you should not use them for something they do not represent. From the user’s guide:

We do continue to provide pixel-scale representations of some outputs for calibration and model-checking purposes only. These pixel-scale maps are not to be interpreted for understanding of hydrological processes or to inform decision making of any kind.

Regarding your math (precip - aet - wyield) and whether that will get you values of 0, the equation for calculating water yield is slightly more elaborate than that. Feel free to review the equations at http://releases.naturalcapitalproject.org/invest-userguide/latest/reservoirhydropowerproduction.html

I’m not a hydrologist, so I’ll need to defer to one on whether you could reasonably determine retention from some of the outputs of the Annual Water Yield model. @RafaSchmitt?

Hi @Tookie,
indeed you can use the month per month rasters. However, as James mentioned, we do not recommend to interpret results on a single pixel level. Commonly, we suggest that you divide your area in some smaller sub-areas (e.g., subwatersheds, or smaller management units of interest) and calculate the mean values for quickflow, baseflow, etc. for these units. The base-flow contribution raster can be used as a proxy for the annual water infiltration (rather than direct runoff). Your water budget calculations seem ok, too, but I would do them only on the sub-watershed level and judge them very critically before using.

1 Like

Hi, you mean the Seasonal water yield model right? I just found it in the userguide. The base flow might be what I need. I will read it carefully. And thank you all so much!