discover_nhd_plus_id accuracy question

[jgrabowski-usgs]

Hello,

I have a question about the discover_nhdplus_id function in the nhdplustools package that I'm hoping you can help me with. I have a Python script based around the Hydrolink module that I have been using to snap points to the NHD. The script uses both stream name and proximity to snap points to a NHD reach. Part of the script is identification of the NWIS site stream name through a few different rules, including looking for words like "near," "below," and "above." The script works well, but I've been worried that there may be an existing, vetted database of comids for NWIS sites. This lead me to check out the discover_nhdplus_id function from nhdplus tools, and compare the comid results to the results from my script.

I ran discover_nhd_plus_id for all NWIS sites with field observations in HUC17 to the NHD, and found that the results only matched my comids 85% of the time. When taking a look the ones that have different values, it is frequently the case that the nhdplustools function has placed the point on the next reach over, either upstream or downstream, from the reach closest to the site coordinates. Here's a screenshot example-- site 13302500, Salmon River at Salmon ID. The hollow green point is the site coordinates (neatly on top of the USGS topo basemap gage symbol), the filled red point is where I snapped it on the NHD, and the selected, bright blue line is the reach where discover_nhdplus_id placed the point. The reach where my script snapped the point is the closest, and the reach that discover_nhdplus_id identified is the next one downstream. My question is, is this instance a miss from discover_nhdplus_id, or does the tool know something I don't know?

Here's another example that I believe is a clear miss by discover_nhdplus_id: 12116100 - Canyon Creek Near Cedar Falls, WA. Canyon Creek is a small stream that is not in NHD. Because Canyon Creek did not match the name of NHD reach nearby, my script did not snap the point. discover_nhd_plus_id returned a reach on the nearby Cedar River. Additionally, the new NWIS page displays the site drainage area as the drainage area of the Cedar River.

I'm hoping that you can explain how the tool works, and what might result in an outcome like snapping to the not-nearest NHD reach in the first instance, and the wrong stream in the second instance. I'm a big fan of nhdplustools and I appreciate any guidance you are able to provide. Thanks!

Julia

[dblodgett-usgs]

Thanks for reach out out Julia --

If you are providing an nldi feature to discover_nhdplus_id, it calls get_nldi_feature() and returns the comid associated with that feature. In the case of USGS gages, those COMIDs have been established in this project: https://github.com/internetofwater/ref_gages

If you are passing discover_nhdplus_id the point location of the gage, then it does a point in polygon query for the comid of the catchment the point falls within. Based on your question, it's not clear which of these methods you are using.

Either way, I am realizing I need to document this better and will take care of that when I get a moment.

If you can provide a little more specific information about how you are using the function, I can help chase down why things might be different.

[jgrabowski-usgs]

Dave,

Thanks for getting back to me so quickly! I used the discover_nhd_plus_id function with the sites as ndli_features. To get the ndli_features, I concatenated the agency and site code of the sites, like "USGS-10366000," for example. The data I am working with is all sites with field observations in NWIS within the study area. Most are USGS gages, but some are not. Maybe my question is more relevant for the ref_gages project, rather than nhdplusTools?

Julia

[dblodgett-usgs]

Yes -- I think your question is likely more related to the site locations in ref_gages. ref_gages is composed of the SWIMS dataset and automatic gage snaps based on similar logic to what you described above. I'm not surprised that you are finding differences but would appreciate hearing about anything that seems problematic.