How can I delineate multiple watersheds in R using the streamstats package? - r

Question

There is an R package in development that I would like to use called streamstats. What it does is delineate a watershed (within the USA) for a latitude & longitude point along a body of water and provides watershed characteristics such as drainage area and proportions of various land covers. What I would like to do is extract some watershed characteristics of interest from a data frame of several lat & long positions.
I can get the package to do what I want for one point
devtools::install_github("markwh/streamstats")
library(streamstats)
setTimeout(120)
dat1 <- data.frame(matrix(ncol = 3, nrow = 3))
x <- c("state","lat","long")
colnames(dat1) <- x
dat1$state <- c("NJ","NY","VA")
dat1$lat <- c(40.99194,42.02458,38.04235)
dat1$long <- c(-74.28000,-75.11928,-79.88144)
test_dat <- dat1[1,]
ws1 <- delineateWatershed(xlocation = test_dat$long, ylocation = test_dat$lat, crs = 4326,
includeparameters = "true", includeflowtypes = "true")
chars1 <- computeChars(workspaceID = ws1$workspaceID, rcode = "MA")
chars1$parameters
However what I would like is to be able to give the delineateWatershed function several watersheds at once (i.e., all 3 locations found in dat1) and combine the chars1$parameters output variables DRNAREA,FOREST,LC11DEV, and LC11IMP into a data frame. Maybe this could be achieved with a for loop?
The ideal output would look like this
state lat long DRNAREA FOREST LC11DEV LC11IMP
1 NJ 40.99194 -74.28000 160 66.2 26.20 5.50
2 NY 42.02458 -75.11928 457 89.3 2.52 0.18
3 VA 38.04235 -79.88144 158 NA 4.63 0.20

Answer 1

I would put what you have in a function then use purrr::pmap_df() to loop through each row in dat1 then bind all the results together. See also this answer
library(dplyr)
library(purrr)
library(tidyr)
library(streamstats)
setTimeout(120)
dat1 <- data.frame(matrix(ncol = 3, nrow = 2))
colnames(dat1) <- c("state", "lat", "long")
dat1$state <- c("NJ", "NY")
dat1$lat <- c(40.99194, 42.02458)
dat1$long <- c(-74.28000, -75.11928)
dat1
#> state lat long
#> 1 NJ 40.99194 -74.28000
#> 2 NY 42.02458 -75.11928
Define a function for catchment delineation
catchment_delineation <- function(rcode_in, lat_y, long_x) {
print(paste0("Processing for lat = ", lat_y, " and long = ", long_x))
ws <- delineateWatershed(xlocation = long_x, ylocation = lat_y, crs = 4326,
includeparameters = "true", includeflowtypes = "true")
ws_properties <- computeChars(workspaceID = ws$workspaceID, rcode = rcode_in)
# keep only what we need
ws_properties_df <- ws_properties$parameters %>%
filter(code %in% c("DRNAREA", "FOREST", "LC11DEV", "LC11IMP")) %>%
mutate(ID = ws$workspaceID,
state = rcode_in,
long = long_x,
lat = lat_y)
return(ws_properties_df)
}
Apply the function to each row in dat1 data frame
catchment_df <- pmap_df(dat1, ~ catchment_delineation(..1, ..2, ..3))
#> https://streamstats.usgs.gov/streamstatsservices/watershed.geojson?rcode=NJ&xlocation=-74.28&ylocation=40.99194&includeparameters=true&includeflowtypes=true&includefeatures=true&crs=4326https://streamstats.usgs.gov/streamstatsservices/parameters.json?rcode=NJ&workspaceID=NJ20210923064141811000&includeparameters=truehttps://streamstats.usgs.gov/streamstatsservices/watershed.geojson?rcode=NY&xlocation=-75.11928&ylocation=42.02458&includeparameters=true&includeflowtypes=true&includefeatures=true&crs=4326https://streamstats.usgs.gov/streamstatsservices/parameters.json?rcode=NY&workspaceID=NY20210923064248530000&includeparameters=true
catchment_df
#> ID name
#> 1 NJ20210923064141811000 Drainage Area
#> 2 NJ20210923064141811000 Percent Forest
#> 3 NJ20210923064141811000 Percent Developed from NLCD2011
#> 4 NJ20210923064141811000 Percent_Impervious_NLCD2011
#> 5 NY20210923064248530000 Drainage Area
#> 6 NY20210923064248530000 Percent Forest
#> 7 NY20210923064248530000 Percent Developed from NLCD2011
#> 8 NY20210923064248530000 Percent_Impervious_NLCD2011
#> description
#> 1 Area that drains to a point on a stream
#> 2 Percentage of area covered by forest
#> 3 Percentage of developed (urban) land from NLCD 2011 classes 21-24
#> 4 Average percentage of impervious area determined from NLCD 2011 impervious dataset
#> 5 Area that drains to a point on a stream
#> 6 Percentage of area covered by forest
#> 7 Percentage of developed (urban) land from NLCD 2011 classes 21-24
#> 8 Average percentage of impervious area determined from NLCD 2011 impervious dataset
#> code unit value state long lat
#> 1 DRNAREA square miles 160.00 NJ -74.28000 40.99194
#> 2 FOREST percent 66.20 NJ -74.28000 40.99194
#> 3 LC11DEV percent 26.20 NJ -74.28000 40.99194
#> 4 LC11IMP percent 5.50 NJ -74.28000 40.99194
#> 5 DRNAREA square miles 457.00 NY -75.11928 42.02458
#> 6 FOREST percent 89.30 NY -75.11928 42.02458
#> 7 LC11DEV percent 2.52 NY -75.11928 42.02458
#> 8 LC11IMP percent 0.18 NY -75.11928 42.02458
Reshape the result to desired format
catchment_reshape <- catchment_df %>%
select(state, long, lat, code, value) %>%
pivot_wider(names_from = code,
values_from = value)
catchment_reshape
#> # A tibble: 2 x 7
#> state long lat DRNAREA FOREST LC11DEV LC11IMP
#> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 NJ -74.3 41.0 160 66.2 26.2 5.5
#> 2 NY -75.1 42.0 457 89.3 2.52 0.18
Created on 2021-09-22 by the reprex package (v2.0.1)

Answer 2

Since you mentioned the use a for loop I thought why not make a solution of it.
Here is your data:
library(dplyr)
library(purrr)
library(tidyr)
library(streamstats)
setTimeout(120)
dat1 <- data.frame(matrix(ncol = 3, nrow = 2))
colnames(dat1) <- c("state", "lat", "long")
dat1$state <- c("NJ", "NY")
dat1$lat <- c(40.99194, 42.02458)
dat1$long <- c(-74.28000, -75.11928)
dat1
Create an empty list to store the watershed characteristics:
water_shed <- list()
Loop through the dat1 and return the properties for each respective longitude and latitude:
for(i in 1:nrow(dat1)){
water_shed[[i]] <-
delineateWatershed(xlocation = dat1$long[i], ylocation = dat1$lat[i], crs = 4326,
includeparameters = "true", includeflowtypes = "true")
}
Now create a list to store the watershed properties:
ws_properties <- list()
Loop through the water_shed returning the parameters of each location:
for(i in 1:length(water_shed)){
ws_properties[[i]] <- computeChars(workspaceID = water_shed[[i]][[1]], rcode = dat1$state)
}
Finally, create a dataframe for your desired outputs then append the properties for each location looping through the list of watershed properties:
# data frame:
ws_properties_df <- data.frame(state=character(),long=integer(), lat=integer(),
DRNAREA = integer(), FOREST = integer(), LC11DEV = integer(), LC11IMP = integer(),
stringsAsFactors=FALSE)
#append properties for eact location
for(i in 1:length(ws_properties)){
ws_properties_df[i,] <- ws_properties[[i]]$parameters %>%
filter(code %in% c("DRNAREA", "FOREST", "LC11DEV", "LC11IMP")) %>%
mutate(state = dat1$state[i],
long = dat1$long[i],
lat = dat1$lat[i]) %>%
select(state, long, lat, code, value) %>%
pivot_wider(names_from = code,
values_from = value)
}
Desired Output: