NWI habitat data

[jbeaulie]

Integrate NWI habitat data provided by Mark Mitchell. Probably need a readNWI.R script. These data can then be merged into the lake-aggregated data set in mergePredictorsPrepForStatistics.R.

Mark Mitchell aggregated the NWI data to the lake-scale, but site specific information could also be useful for modeling that uses data from all measurements sites (i.e. sites within lakes). Our direct depth measurement should be better than NWI estimates of profundal vs littoral, however; perhaps we can recode our depth data into littoral vs profundal via NWI criteria. This would enable us to use our measurements and modeling and NWI for upscaling. Investigate NWI documentation for profundal vs littoral criteria.

[brdeemer]

Littoral is defined generally as <2.5m within the permanently flooded zone of the waterbody in the NWI, but there is some ambiguity around emergent vegetation. Found documentation here: https://www.fws.gov/sites/default/files/documents/Classification-of-Wetlands-and-Deepwater-Habitats-of-the-United-States-2013.pdf Helpful diagram on page 17 & definition on page 18: "Littoral. This subsystem includes all wetland habitats in the Lacustrine System. It extends from the shoreward boundary of the System to a depth of 2.5 m (8.2 ft) below low water, or to the maximum extent of nonpersistent emergents if these grow at depths greater than 2.5 m."

[jbeaulie]

Will use data down to Class level of hierarchy. Concerned that level of detail may vary across states. Check to see how many lakes have data at System, (subsytem when relevant), and class levels. If any missing, look for state by state patterns in missingness.

[brdeemer]

Edited: Created an object, "nwi" that has data for 148 reservoirs. Adjusted percentages of system and subsystem information to be relative to the total sum of habitat available. We need to think through how identify the "class" information since there are repeat classes in different system types. All reservoirs have data to the class level.

[jbeaulie]

I looked at Unconsolidated Bottom as an example of a class value that shows up under multiple System/Subsystem types. Per the NWI documentation, Unconsolidated Bottom has the same definition, regardless of what System/Subsystem it is nested within. In this case, I think we should aggregated all Unconsolidated Bottom for each reservoir. Same would apply to other class values, assuming they are defined uniformly across System/Subsystems.

[brdeemer]

I checked and the "class" values add to a median of 195... a lot more than 100% (e.g. it looks like the datafile is providing relative surface areas for the system/subsystem that they pertain to). I think we can correct this by multiplying the class values by the percentage of the lake that was in that particular system/subsystem, but we need to know which classes pertain to which systems/subsystems which isn't clear to me from the input file. The input metadata has some of the class info (with ".x" appended) labeled as "NWI Class Name" and other class info (with ".y" appended) labeled as Splitclass name. There are repeated class names across all 3 included System types (for example there is an "aquatic bed" category in Riverine, Lacustrine, and Palustrine system types), but there are also cases where the class only appears for one System (e.g. there is only a streambed category for the Riverine systems). Any thoughts?

[brdeemer]

Ok, I re-read Mark Mitchell's comments and I think I understand better. I've corrected all percentages by dividing by the total lacustrine+palustrine+riverine in the polygon. Classes are collapsed across the riverine and lacustrine systems. I think we might want to decide on a combination of classes to represent percent shallow depth, but I'm not sure which classes are best for this. Only 16 out of 118 surge sites have any littoral habitat classified.

[jbeaulie]

@jbeaulie provide Mark with a list of habitat codes that equal "shallow" and see if he can generate proportion of each waterbody that is shallow based on those codes.

[brdeemer]

For the email to Mark: habitat codes that correspond with shallow are:

1. all palustrine habitat (P)
2. lacustrine littoral habitat(L2)
3. riverine unconsolidated shore(R1US, R2US, R3US)
4. riverine aquatic bed (R1AB, R2AB, R3AB)
5. riverine emergent (R1EM, R2EM, R3EM)
6. riverine intermittent (R4)

Rusty Griffin also mentioned he was happy to hop on a call to help talk through the query if there are questions along the way. rusty_griffin@fws.gov

[jbeaulie]

emailed Mark on 9/25/24

[jbeaulie]

e-mailed Mark 1/31/25

[jbeaulie]

Bridget will incorporate the habitat aggregated data into the lake-scale data file for the data paper.

[brdeemer]

I have added the habitat aggregated data into the lake-scale data file, but we are still waiting on Mark for the breakdown by "shallow" habitat types

[jbeaulie]

NWI data for the SuRGE lakes are read from SuRGE_Sharepoint/data/siteDescriptors/nwi/SURGE_AllAttributes_NWI_20240124.xlsx. This file contains data for 12 lakes that were not sampled (9, 15, 56, 183, 185, 194, 212, 213, 244, 285, 318, 325) and is missing data for another 13 (238, 239, 249, 253, 263, 265, 287, 302, 308, 323, 331, 999, 1000). Lake 1000 is in Puerto Rico, and perhaps NWI doesn't extend to that region, but not sure why the others are missing.

I modified code to exclude the extra lakes (currently carried into lake_scale.csv for the data paper) but we need to track down the missing lakes.

[jbeaulie]

emailed Mark on 8/8/2025