fix up mainstem uris

R/gage_locations.R

@@ -178,26 +178,21 @@ get_hydrologic_locations <- function(all_gages, hydrologic_locations, nhdpv2_fli
 }
 
 add_mainstems <- function(gage_hydrologic_locations, mainstems, vaa) {
-  mainstems <- mainstems[,c("id", "uri"), drop = TRUE]
-  mainstems$id <- as.integer(mainstems$id)
-
-  vaa <- data.table::as.data.table(vaa)
-
-  vaa <- data.table::merge.data.table(unique(vaa), unique(mainstems), 
-               by.x = "levelpathi", by.y = "id", 
-               all.x = FALSE, all.y = TRUE)
+  mainstems <- mainstems[,c("head_nhdpv2_COMID", "uri"), drop = TRUE]
+  mainstems$head_nhdpv2_COMID <- as.integer(gsub("https://geoconnex.us/nhdplusv2/comid/", "", 
+                                                 mainstems$head_nhdpv2_COMID))
 
-  vaa <- vaa[,c("comid", "uri")]
-
-  names(vaa) <- c("comid", "mainstem_uri")
-
-  vaa <- vaa[!is.na(comid)]
-
-  gage_hydrologic_locations <- data.table::as.data.table(gage_hydrologic_locations)
-
-  out <- data.table::merge.data.table(gage_hydrologic_locations, vaa, 
-                               by.x = "nhdpv2_COMID", by.y = "comid", 
-                               all.x = TRUE, all.y = FALSE)
-
-  sf::st_sf(as.data.frame(out))
+  mainstem_lookup <- group_by(vaa, levelpathi) |>
+    filter(hydroseq == max(hydroseq)) |>
+    ungroup() |>
+    select(head_nhdpv2_COMID = comid, levelpathi) |>
+    distinct() |>
+    left_join(mainstems, by = "head_nhdpv2_COMID") |>
+    filter(!is.na(uri)) |>
+    select(-head_nhdpv2_COMID) |>
+    right_join(select(vaa, comid, levelpathi), 
+               by = "levelpathi") |>
+    select(-levelpathi, comid, mainstem_uri = uri)
+    
+  dplyr::left_join(gage_hydrologic_locations, mainstem_lookup, by = c("nhdpv2_COMID" = "comid"))
 }

_targets.R

@@ -30,7 +30,7 @@ list(
   tar_target("nhdpv2_fline_proc", select(st_transform(nhdpv2_fline, 5070),
                                         COMID, REACHCODE, ToMeas, FromMeas)),
   tar_target("mainstems", get_all_mainstems("data/mainstems/")),
-  tar_target("vaa", get_vaa(atts = c("comid", "levelpathi"),
+  tar_target("vaa", get_vaa(atts = c("comid", "levelpathi", "hydroseq"),
                            updated_network = TRUE)),
   # This function downloads all NWIS sites from the site file
   tar_target("nwis_gage", get_nwis_sites()),