The goal of this vignette is to demonstrate a simple and lightweight approach to building maps with NHDPlus data.
For this example, we'll start from an outlet NWIS Site. Note that other options are possible with discover_nhdplus_id and discover_nldi_sources.
library(sf)
library(nhdplusTools)
nwissite <- list(featureSource = "nwissite",
featureID = "USGS-05428500")
flowline <- navigate_nldi(nwissite,
mode = "upstreamTributaries",
data_source = "")
nhdplus <- subset_nhdplus(comids = flowline$nhdplus_comid,
output_file = file.path(temp_dir, "nhdplus.gpkg"),
nhdplus_data = "download",
overwrite = TRUE)
flowline <- read_sf(nhdplus, "NHDFlowline_Network")
upstream_nwis <- navigate_nldi(nwissite,
mode = "upstreamTributaries",
data_source = "nwissite")
basin <- get_nldi_basin(nwissite)
Now we have a file at the path held in the variable nhdplus and three sf data.frames with contents that look like:
st_layers(nhdplus)
names(flowline)
names(upstream_nwis)
names(basin)
class(st_geometry(flowline))
class(st_geometry(upstream_nwis))
class(st_geometry(basin))
Our file has four layers: network flowlines, simplified catchments, nhd area features, and nhd waterbody features.
The flowlines have a large set of attributes from the NHDPlus dataset. And the nwis sites have a few attributes that came from the NLDI. Attributes for NWIS sites can be found using the dataRetrieval package.
See the NHDPlus user guide linked here for more on what these layers are and what the flowline attributes entail.
First, a side note on bounding boxes. With the ongoing transition from the sp package to the sf package, there are a few stumbling blocks. Bounding boxes are one of them. As shown below, the sf bbox format is a named vector of class “bbox”. The sp bbox format is a matrix with named dimensions. Many packages expect the sp format. the ggmap package expects yet another bbox format, much like sf but with different names.
library(sp)
sf_bbox <- st_bbox(basin)
sf_bbox
class(sf_bbox)
sp_bbox <- sp::bbox(sf::as_Spatial(basin))
sp_bbox
class(sp_bbox)
# Or without the sp::bbox
sp_bbox <- matrix(sf_bbox,
byrow = FALSE,
ncol = 2,
dimnames = list(c("x", "y"),
c("min", "max")))
sp_bbox
ggmap_bbox <- setNames(sf_bbox, c("left", "bottom", "right", "top"))
ggmap_bbox
In order to maximize flexibility and make sure we understand what's going on with coordinate reference systems, the demonstration below shows how to use base R plotting with the package prettymappr and rosm.
In this example, we have to plot just the geometry, extracted with st_geometry and we need to project the geometry into the plotting coordinate reference system, EPSG:3857 also known as “web mercator”. The reason we have to make this transformation is that practically all basemap tiles are in this projection and reprojection of pre-rendered tiles doesn't look good. We do this with a simple prep_layer function.
The prettymapr::prettymap() function isn't strictly necessary, but it gives us nice margins, a scale bar, and a north arrow. The rosm::osm.plot and base plot commands put data onto the R plotting device so the first to be plotted is on the bottom. A couple hints here. lwd is line width. pch is point style. cex is an expansion factor. Colors shown below are basic R colors. the rgb function is handy for creating colors with transparency if that's of interest.
prep_layer <- function(x) st_geometry(st_transform(x, 3857))
prettymapr::prettymap({
rosm::osm.plot(sp_bbox, type = "stamenwatercolor", quiet = TRUE)
plot(prep_layer(basin),
lwd = 2, add = TRUE)
plot(prep_layer(flowline),
lwd = 1.5, col = "deepskyblue", add = TRUE)
plot(prep_layer(dplyr::filter(flowline, streamorde > 2)),
lwd = 3, col = "darkblue", add = TRUE)
us_nwis_layer <- prep_layer(upstream_nwis)
plot(us_nwis_layer,
pch = 17, cex = 1.5, col = "yellow", add = TRUE)
label_pos <- st_coordinates(us_nwis_layer)
text(label_pos[,1],label_pos[,2],
upstream_nwis$identifier,
adj = c(-0.2, 0.5), cex = 0.7)
}, drawarrow = TRUE)
Below is a very similar example using ggmap and ggplot2 geom_sf. Note that ggmap takes case of projections for us, which should either make you happy because it just works or very nervous because it just works.
library(ggmap)
library(ggplot2)
upstream_nwis[c("lon", "lat")] <- sf::st_coordinates(upstream_nwis)
basemap_toner <- get_map(source = "stamen", maptype = "toner", location = ggmap_bbox, zoom = 11)
basemap_terrain <- get_map(source = "stamen", maptype = "terrain-lines", location = ggmap_bbox, zoom = 11)
toner_map <- ggmap(basemap_toner)
terrain_map <- ggmap(basemap_terrain)
toner_map
terrain_map + geom_sf(data = basin,
inherit.aes = FALSE,
color = "black", fill = NA) +
geom_sf(data = flowline,
inherit.aes = FALSE,
color = "deepskyblue") +
geom_sf(data = dplyr::filter(flowline, streamorde > 2),
inherit.aes = FALSE,
color = "darkblue") +
geom_sf(data = upstream_nwis, inherit.aes = FALSE, color = "red") +
geom_text(data = upstream_nwis, aes(label = identifier, x = lon, y = lat),
hjust = 0, size=2.5, nudge_x = 0.02, col = "black")
Hopefully these examples give a good head start to plotting NHDPlus data. Please submit questions via github issues for more!! Pull requests on this vignette are more than welcome if you have additions or suggestions.