U.S. Data and Package Basics

TL;DR

Pick an outlet location and download some data.

# Uncomment to install!
# install.packages("hydrogeofetch")

library(hydrogeofetch)
library(sf)

start_point <- st_sfc(st_point(c(-89.362239, 43.090266)), crs = 4269)
start_comid <- discover_nhdplus_id(start_point)

flowline <- navigate_nldi(list(featureSource = "comid", 
                               featureID = start_comid), 
                          mode = "upstreamTributaries", 
                          distance_km = 1000)

subset_file <- tempfile(fileext = ".gpkg")
subset <- subset_nhdplus(comids = as.integer(flowline$UT$nhdplus_comid),
                         output_file = subset_file,
                         nhdplus_data = "download", 
                         flowline_only = FALSE,
                         return_data = TRUE, overwrite = TRUE)

flowline <- subset$NHDFlowline_Network
catchment <- subset$CatchmentSP
waterbody <- subset$NHDWaterbody

## Or using a file:

flowline <- sf::read_sf(subset_file, "NHDFlowline_Network")
catchment <- sf::read_sf(subset_file, "CatchmentSP")
waterbody <- sf::read_sf(subset_file, "NHDWaterbody")

plot(sf::st_geometry(flowline), col = "blue")
plot(start_point, cex = 1.5, lwd = 2, col = "red", add = TRUE)
plot(sf::st_geometry(catchment), add = TRUE)
plot(sf::st_geometry(waterbody), col = rgb(0, 0, 1, alpha = 0.5), add = TRUE)

Or fetch an NWIS site as the starting point and generate a plot. Data is returned and/or stored in a local file for later use.

# ?plot_nhdplus for more
plot_data <- plot_nhdplus(
  outlets = list(featureSource = "nwissite", featureID = "USGS-05428500"), 
  gpkg = subset_file, overwrite = TRUE)

This vignette covers a range of utilities the hydrogeofetch package offers for working with data in a U.S. context.

The first thing you are going to need to do is go get some data to work with. hydrogeofetch provides the ability to download small subsets of the NHDPlus directly from web services. For large subsets, greater than a few thousand square kilometers, you can use download_nhdplusv2().

If you are working with the whole National Seamless database, hydrogeofetch has some convenience functions you should be aware of. Once you have it downloaded and extracted, you can tell the hydrogeofetch package where it is with the nhdplus_path() function.

This vignette stages its data in a working directory. Set work_dir to wherever you want downloaded and sample data to live.

work_dir <- file.path(hydrogeofetch_data_dir(), "nhdpt_v_cache")

dir.create(work_dir, showWarnings = FALSE, recursive = TRUE)

nhdplus_path(file.path(work_dir, "natseamless.gpkg"))

basename(nhdplus_path())

If you don’t need or want all the geometry for the flowlines, consider using get_vaa(). It allows you to retrieve specific NHDPlus attributes with very little overhead. It also supports access to an updated set of network attributes that incorporate numerous network updates from national hydrologic modeling projects.

vaa <- get_vaa()
names(vaa)
nrow(vaa)

NHDPlus HiRes

NHDPlus HiRes is an in-development dataset that introduces much more dense flowlines and catchments. In the long run, hydrogeofetch will have complete support for NHDPlus HiRes. So far, hydrogeofetch will help download and interface NHDPlus HiRes data with existing hydrogeofetch functionality. It’s important to note that hydrogeofetch was primarily implemented using NHDPlusV2 and any use of HiRes should be subject to scrutiny.

For the demo below, a small sample of HiRes data is staged into work_dir (the working directory set above) and read from there. The first line shows how you can download additional data (just change download_files to TRUE).

download_nhdplushr(nhd_dir = "download_dir", 
                   hu_list = c("0101"), # can mix hu02 and hu04 codes.
                   download_files = FALSE) # TRUE will download files.

out_gpkg <- file.path(work_dir, "nhd_hr.gpkg")
hr_data <- get_nhdplushr(work_dir, 
                         out_gpkg = out_gpkg)
(layers <- st_layers(out_gpkg))
names(hr_data)
unlink(out_gpkg)

hr_data <- get_nhdplushr(work_dir, 
                         out_gpkg = out_gpkg, 
                         layers = NULL)
(layers <- st_layers(out_gpkg))
names(hr_data)

Discovery and Subsetting

One of the primary workflows hydrogeofetch is designed to accomplish can be described in three steps:

what NHDPlus catchment is at the outlet of a watershed,
figure out what catchments are up or downstream of that catchment, and
create a stand alone subset for that collection of catchments.

Say we want to get a subset of the NHDPlus upstream of a given location. We can start with discover_nhdplus_id() First, let’s look at a given point location. Then see where it is relative to our flowlines.

lon <- -89.36
lat <- 43.09

start_point <- sf::st_sfc(sf::st_point(c(lon, lat)),
                          crs = 4269)

plot(sf::st_geometry(flowline))
plot(start_point, cex = 1.5, lwd = 2, col = "red", add = TRUE)

OK, so we have a point location near a river and we want to figure out what catchment is at its outlet. We can use the discover_nhdplus_id() function which calls out to a web service and returns an NHDPlus catchment identifier, typically called a COMID.

If we set raindrop = TRUE, we can also get a elevation derived downslope trace to the nearest flowline and some additional info. See get_raindrop_trace() for more on this functionality.

start_comid <- discover_nhdplus_id(start_point, raindrop = TRUE)
start_comid

plot(sf::st_geometry(start_comid))
plot(sf::st_geometry(flowline), add = TRUE, col = "blue", lwd = 2)
plot(start_point, cex = 1.5, lwd = 2, col = "red", add = TRUE)

If you have the whole National Seamless database and want to work at regional to national scales, skip down to the Local Data Subsetting section.

Web Service Data Subsetting

hydrogeofetch supports discovery and data subsetting using web services made available through the Network Linked Data Index (NLDI) and the Water Mission Area pygeoapi. The code below shows how to use the NLDI functions to build a dataset upstream of our start_comid that we found above.

The NLDI can be queried with any set of watershed outlet locations that it has in its index. We call these “featureSources”. We can query the NLDI for an identifier of a given feature from any of its “featureSources” and find out what our navigation options are as shown below.

dataRetrieval::get_nldi_sources()$source

nldi_feature <- list(featureSource = "comid", 
                     featureID = as.integer(start_comid$comid)[1])

get_nldi_feature(nldi_feature)

We can use get_nldi_feature() as a way to make sure the featureID is available for the chosen “featureSource”. Now that we know the NLDI has our comid, we can use the “upstreamTributaries” navigation option to get all the flowlines upstream or all the features from any of the “featureSources” as shown below.

flowline_nldi <- navigate_nldi(nldi_feature, 
                               mode = "upstreamTributaries", 
                               distance_km = 1000)

plot(sf::st_geometry(flowline), lwd = 3, col = "black")
plot(sf::st_geometry(flowline_nldi$origin), lwd = 3, col = "red", add = TRUE)
plot(sf::st_geometry(flowline_nldi$UT), lwd = 1, col = "red", add = TRUE)

The NLDI only provides geometry and a comid for each of the flowlines. The subset_nhdplus() function has a “download” option that allows us to download four layers and all attributes as shown below.

output_file_download <- file.path(work_dir, "subset_download.gpkg")

output_file_download <-subset_nhdplus(comids = as.integer(flowline_nldi$UT$nhdplus_comid),
                                      output_file = output_file_download,
                                      nhdplus_data = "download", return_data = FALSE,
                                      overwrite = TRUE)

sf::st_layers(output_file_download)

flowline_download <- sf::read_sf(file.path(work_dir, "subset_download.gpkg"), 
                                 "NHDFlowline_Network")

plot(sf::st_geometry(dplyr::filter(flowline_download, 
                                   streamorde > 2)), 
     lwd = 7, col = "darkgrey")
plot(sf::st_geometry(flowline_nldi$UT), 
     lwd = 3, col = "red", add = TRUE)

This plot illustrates the kind of thing that’s possible (filtering to specific stream orders) using the attributes that are downloaded.

Before moving on, one more demonstration of what can be done using the NLDI. Say we knew the USGS gage ID that we want NHDPlus data upstream of. We can use the NLDI to navigate from the gage the same as we did for our comid. We can also get back all the nwis sites the NLDI knows about upstream of the one we chose!

nldi_feature <- list(featureSource = "nwissite", 
                     featureID = "USGS-05428500")

flowline_nldi <- navigate_nldi(nldi_feature, 
                               mode = "upstreamTributaries", 
                               distance_km = 1000)

output_file_nwis <- file.path(work_dir, "subset_download_nwis.gpkg")

output_file_nwis <-subset_nhdplus(comids = as.integer(flowline_nldi$UT$nhdplus_comid),
                                  output_file = output_file_nwis,
                                  nhdplus_data = "download",
                                  return_data = FALSE, overwrite = TRUE)

sf::st_layers(output_file_download)

flowline_nwis <- sf::read_sf(output_file_nwis, 
                                 "NHDFlowline_Network")

upstream_nwis <- navigate_nldi(nldi_feature,
                               mode = "upstreamTributaries",
                               data_source = "nwissite", 
                               distance_km = 1000)

plot(sf::st_geometry(flowline_nwis), 
     lwd = 3, col = "blue")
plot(sf::st_geometry(upstream_nwis$UT_nwissite), 
     cex = 1, lwd = 2, col = "red", add = TRUE)

Local Data Subsetting

While web service data access is very convenient, some use cases make working with web services impossible or cumbersome such that working with local data is preferable. hydrogeofetch supports such workflows with hybrid, web-service and local, workflows.

With the starting COMID we found with discover_nhdplus_id() above, we can use hydroloom’s navigate_hydro_network() to retrieve a collection of COMIDs along the upstream-with-tributaries network path.

UT_comids <- hydroloom::navigate_hydro_network(flowline, start_comid$comid[1], "UT")
UT_comids

If you are familiar with the NHDPlus, you will recognize that now that we have this list of COMIDs, we could go off and do all sorts of things with the various flowline attributes. For now, let’s just use the COMID list to filter our fline sf data.frame and plot it with our other layers.

plot(sf::st_geometry(flowline))
plot(start_point, cex = 1.5, lwd = 2, col = "red", add = TRUE)
plot(sf::st_geometry(dplyr::filter(flowline, comid %in% UT_comids)),
     add=TRUE, col = "red", lwd = 2)

Say you want to save the network subset for later use in R or in some other GIS. The subset_nhdplus() function is your friend. If you have the whole national seamless database downloaded, you can pull out large subsets of it like shown below (this queries for data from the local geodatabase without loading the whole thing into memory). If you don’t have the whole national seamless, look at the second example in this section.

output_file <- file.path(work_dir, "subset.gpkg")

output_file <-subset_nhdplus(comids = UT_comids,
                             output_file = output_file,
                             nhdplus_data = nhdplus_path(), 
                             return_data = FALSE, overwrite = TRUE)

sf::st_layers(output_file)

Now we have an output geopackage that can be used later. It contains the network subset of catchments and flowlines as well as a spatial subset of other layers as shown in the status output above. To complete the demonstration, here are a couple more layers plotted up.

catchment <- sf::read_sf(output_file, "CatchmentSP")
waterbody <- sf::read_sf(output_file, "NHDWaterbody")

plot(sf::st_geometry(flowline))
plot(start_point, cex = 1.5, lwd = 2, col = "red", add = TRUE)
plot(sf::st_geometry(dplyr::filter(flowline, comid %in% UT_comids)),
     add=TRUE, col = "red", lwd = 2)
plot(sf::st_geometry(catchment), add = TRUE)
plot(sf::st_geometry(waterbody), col = rgb(0, 0, 1, alpha = 0.5), add = TRUE)

Indexing

Indexing functions have moved to the hydroloom package. Use hydrogeofetch to fetch flowline data and hydroloom to index points to those flowlines.

hydroloom::index_points_to_lines(flowline, start_point)

index_points_to_lines() works with a list of points too. For demonstration purposes, we can use the gages in our subset from above.

gage <- sf::read_sf(output_file, "Gage")

hydroloom::index_points_to_lines(flowline, sf::st_geometry(gage), precision = 10)

See the hydroloom indexing documentation and the migrating from nhdplusTools article for more.

Mainstems across networks

A mainstem (a river from its headwater to its outlet, identified by a geoconnex ref/mainstems uri) is carried by NHDPlusV2, NHDPlusHR and 3DHP, but each network breaks it into different flowlines with different identifiers. get_mainstem_flowlines() returns the flowlines of one or more mainstems from any of the three, with mainstemid set to the mainstem uri and the topology and length attributes hydroloom expects. With include_confluences = TRUE it also returns the outlet flowline of each identified tributary, which is what hydroloom::get_confluences() needs to find where tributaries join.

new_hope <- "https://geoconnex.us/ref/mainstems/2408056"

nh_3dhp <- get_mainstem_flowlines(new_hope, "3dhp", include_confluences = TRUE)

confluences <- hydroloom::get_confluences(nh_3dhp)

confluences[grepl("/ref/", confluences$hydro_location_id), ]

Because the mainstem uri is shared, a point or flowline indexed on one network can be moved to another while staying on its own river — a gage just above a confluence is not pulled onto the tributary. translate_hydrolocations() does this for points in one of two ways. method = "geometry" indexes each point to the target flowlines of its mainstem. method = "hydro_location" records each point as a distance from the nearest confluence that both networks carry and lays that distance out on the target. The second is approximate in absolute terms, since each network puts the confluence in its own place, but it keeps the point on the same side of the confluence. translate_flowlines() does the same for flowlines, returning the pieces of target flowlines that cover each source flowline.

Both functions read the ref_rivers lookup and updated network attribute tables for NHDPlusV2, which are large downloads on first use, so the calls below are not run here.

# a point on New Hope River just above the Northeast Creek mouth
pt <- sf::st_sf(comid = 8894344,
                geometry = sf::st_sfc(sf::st_point(c(-78.9763, 35.8355)),
                                      crs = 4326))

translate_hydrolocations(pt, "nhdpv2", "3dhp", method = "hydro_location")

# a Morgan Creek flowline
translate_flowlines(8896344, "nhdpv2", "3dhp")

Two properties of the data bound these results. 3DHP identifiers are not persistent across releases, so a placement on 3DHP holds for the release recorded in measured_on. The NHDPlusHR lookup in ref_rivers v3.2 covers only part of the country (New Hope is not in it), so "nhdphr" returns nothing for many mainstems.