added entero_stationdata and associated docs

R/entero_stationdata.R

@@ -0,0 +1,16 @@
+#' Enterococcus station location metadata
+#'
+#' @format A data frame with 50 rows and 5 columns:
+#' \describe{
+#'  \item{station}{character, sample station, as named in the Water Quality Portal}
+#'  \item{latitude}{numeric, latitude in decimal degrees}
+#'  \item{longitude}{numeric, longitude in decimal degrees}
+#'  \item{bay_segment}{character, full name of Bay Segment in which the station lies}
+#'  \item{bay_segment_abbrev}{character, abbreviated version of Bay Segment in which the station lies}
+#' }
+#' @details Latitude, Longitude, and Bay Segment information for each of the identified key Enterococcus stations. Generated by \code{data-raw/entero_stationdata-raw.R} ( \link[https://github.com/tbep-tech/tbeptools/blob/master/data-raw/bsmap-raw.R]{view on github})
+#'
+#' @source Lat/Long information from Water Quality Portal, https://waterqualitydata.us
+#'
+
+"entero_stationdata"

data-raw/entero_stationdata-raw.R

@@ -0,0 +1,65 @@
+#' compile metadata on key Enterococcus stations
+#' primarily so bay segments are easily available
+
+library(dplyr)
+library(sf)
+library(tbeptools)
+
+
+# load stations
+stns <- unique(catch_precip$station)
+
+
+# download data to get lat/longs
+entero_names <- c('Enterococci',
+                  'Enterococcus')
+startDate <- as.Date('2013-01-01')
+endDate <- as.Date('2023-12-31')
+args <- list(
+  siteid = stns,
+  characteristicName = entero_names,
+  startDateLo = format(startDate, '%m-%d-%Y'),
+  startDateHi = format(endDate, '%m-%d-%Y')
+)
+
+dat_in <- read_importentero(args = args) %>%
+  dplyr::select(-qualifier,
+                -LabComments)
+
+
+# refine the data frame
+# there are a few with multiple lat/longs, but from eyeballing it nothing
+# is obscenely different - so take an average
+dat <- dat_in %>%
+  select(station, Latitude, Longitude) %>%
+  distinct() %>%
+  summarize(.by = station,
+            latitude = mean(Latitude),
+            longitude = mean(Longitude))
+
+# three of our original stations don't show up in the last 10 years
+# make sure it's no mistake but otherwise we can ignore - if
+# no longer sampled, they won't show up in anything else anyway
+stns[which(!(stns %in% dat$station))]
+
+
+
+# intersect with spatial object
+# containing bay segment delineations
+
+# first make the spatial object valid
+tbsegshed2 <- st_make_valid(tbsegshed)
+
+# intersect and refine the data frame
+entero_stationdata <- st_as_sf(dat,
+                               coords = c("longitude", "latitude"),
+                               crs = "EPSG:4326",
+                               remove = FALSE) %>%
+  st_intersection(tbsegshed2) %>%
+  st_drop_geometry() %>%
+  rename(bay_segment_abbrev = bay_segment,
+         bay_segment = long_name)
+
+# save it
+save(entero_stationdata, file = 'data/entero_stationdata.RData', compress = 'xz')
+