Hello @ReissM -
These are all good questions. I’m much more familiar with Nutrient Delivery Ratio (NDR) than Urban Stormwater (US), but I’ll offer a few thoughts, and hopefully others will too.
Are they fundamentally doing the same thing, or are they conceptually different?
They are conceptually different. NDR first determines the nutrient load leaving each pixel, then calculates the amount of that load that reaches a stream (called export) based on the “nutrient delivery ratio” which is a function of the upslope area and downslope flow path.
Urban Stormwater first calculates the volume of water retained on a pixel, and running off of a pixel, then applies the “event mean concentration” nutrient value to determine the amount of nutrient pollution avoided by retention, or contributed by runoff. It does not trace the downstream flowpath. As the User Guide notes:
Retention on a given pixel, at least when considered at an aggregated sub-watershed scale, should be affected by retention on upstream and downstream pixels. (For example, the current InVEST nutrient delivery ratio model takes this into account through a flowpath delineation and retention length framework, which is not as feasible in an urban drainage network because of how storm drains alter surface and sub-surface flow paths.)
The nutrient loading from the pixel that NDR starts with is akin to the load result from US. But US stops at calculating the per-pixel load contribution, it does not determine how much reaches a stream.
When should one be used over the other — especially in mixed landscapes (e.g., peri-urban areas)?
NDR was designed for landscapes that have at least mostly natural drainage (so a digital elevation model can be used to determine upslope and downslope properties, and where streams are located). US was designed for urban areas where drainage is mostly via infrastructure, so natural drainage doesn’t apply so much. NDR will tell you how much nutrient arrives at streams, US does not. Both will give an indication of where nutrient runoff is being produced on the landscape. If you are working in a peri-urban area, and are most interested in understanding where nutrient pollution is likely to be retained or generated, then US seems like an appropriate choice.
Should the results from each model be interpreted differently in terms of ecosystem service supply?
Oddly enough, NDR doesn’t currently provide an output that quantifies the service of nutrient retention (although we’re working on it). You can calculate a proxy for it as described in the User Guide. US provides the avoided loading result, which takes into account the landscape properties that contribute to infiltration versus runoff, and can be thought of as a service.
If I run both models in the same location, how do I appropriately characterize the differences in what they’re capturing? Are the results additive?
I’ve never done this, but would say that the results are not additive. The loading results are similar, so that would be double-counting. And the NDR export results rely on there being a more or less natural drainage to route nutrient to, which urban areas do not generally have. This is a question where I’d really like to hear the experience of others.
~ Stacie