Coordinate the request and retrieval of STAC calls
Examples
# \donttest{
monitoring_location_id <- c("USGS-01435000", "USGS-05406360")
features <- search_stac(
monitoring_location_id = monitoring_location_id,
collections = c("continuous"))
#> Requesting:
#> https://api.waterdata.usgs.gov/stac/v0/search?collections=continuous&filter=monitoring_location_id%20IN%20%28%27USGS-01435000%27%2C%20%27USGS-05406360%27%29%20OR%20&filter-lang=cql2-text&limit=50000&f=json&lang=en-US
#> Remaining requests this hour:587
###################################################
# No download, no load: Returns lazy table
con_1 <- initial_connection()
data_tbl <- open_stac(features = features,
con = con_1)
head(data_tbl)
#> # A query: ?? x 11
#> # Database: DuckDB 2.0.0 [unknown@Linux 7.0.0-1017-azure:R 4.6.2/:memory:]
#> time_series_id monitoring_location_id parameter_code statistic_id
#> <chr> <chr> <chr> <chr>
#> 1 696c70dd239f4e5ab750c95e54… USGS-05406360 00060 00011
#> 2 696c70dd239f4e5ab750c95e54… USGS-05406360 00060 00011
#> 3 696c70dd239f4e5ab750c95e54… USGS-05406360 00060 00011
#> 4 696c70dd239f4e5ab750c95e54… USGS-05406360 00060 00011
#> 5 696c70dd239f4e5ab750c95e54… USGS-05406360 00060 00011
#> 6 696c70dd239f4e5ab750c95e54… USGS-05406360 00060 00011
#> # ℹ 7 more variables: time <dttm>, value <chr>, unit_of_measure <chr>,
#> # approval_status <chr>, qualifier <list>, last_modified <dttm>,
#> # method_category <chr>
summarize_data <- data_tbl |>
dplyr::mutate(value = as.numeric(value)) |>
dplyr::summarize(count = dplyr::n(),
start = min(time, na.rm = TRUE),
end = max(time, na.rm = TRUE),
max_val = max(value, na.rm = TRUE),
.by = c(time_series_id, monitoring_location_id, parameter_code)) |>
dplyr::collect()
summarize_data
#> # A tibble: 61 × 7
#> time_series_id monitoring_location_id parameter_code count
#> <chr> <chr> <chr> <dbl>
#> 1 0be5a4c566f145998e159920fe6e90f5 USGS-01435000 00010 159825
#> 2 af4996712fbf4d63810d6c133391de71 USGS-01435000 62611 166027
#> 3 f22a2084703c45578e9fb1fcc3880b1f USGS-01435000 00065 497980
#> 4 f9a3961b281441d9bb408f0bca322d56 USGS-01435000 00095 141554
#> 5 fc7dfde1701b41b1b6740b0cda1614ee USGS-01435000 00095 83385
#> 6 e8a4982eb7434d6ea94286bf80d24583 USGS-01435000 00045 151149
#> 7 3a7de90ef9824ec7ab7370b4e0b1340b USGS-01435000 00020 240764
#> 8 6885cbcdc52e4776a7962634cc95f12d USGS-01435000 00095 225810
#> 9 f5079a8831044133a78f2ff27c8a0418 USGS-01435000 00010 11774
#> 10 701417f906ee47399e262abf7522620a USGS-01435000 00010 11774
#> # ℹ 51 more rows
#> # ℹ 3 more variables: start <dttm>, end <dttm>, max_val <dbl>
DBI::dbDisconnect(con_1)
#########################################################
# Download data locally, no load: Returns lazy table
raw_data_store <- file.path(tempdir(), "test_1")
dir.create(raw_data_store, showWarnings = FALSE)
local_path <- download_stac(
features = features,
file_directory = raw_data_store)
#> Downloading USGS-05406360: continuous
#> Downloading USGS-01435000: continuous
#> Downloading USGS-01435000: continuous
con_2 <- initial_connection()
data_tbl <- open_stac(features = local_path,
con = con_2)
head(data_tbl)
#> # A query: ?? x 11
#> # Database: DuckDB 2.0.0 [unknown@Linux 7.0.0-1017-azure:R 4.6.2/:memory:]
#> time_series_id monitoring_location_id parameter_code statistic_id
#> <chr> <chr> <chr> <chr>
#> 1 1b2e9bc4974a4d5b8bfa436093… USGS-01435000 00010 00011
#> 2 0be5a4c566f145998e159920fe… USGS-01435000 00010 00011
#> 3 a14ffc218c0647fb8554708eaf… USGS-01435000 00010 00011
#> 4 af4996712fbf4d63810d6c1333… USGS-01435000 62611 00011
#> 5 1e9b356375e44e5ba4a9fae83b… USGS-01435000 62611 00011
#> 6 1f10ef6fce32427bb908396225… USGS-01435000 00095 00011
#> # ℹ 7 more variables: time <dttm>, value <chr>, unit_of_measure <chr>,
#> # approval_status <chr>, qualifier <list>, last_modified <dttm>,
#> # method_category <chr>
DBI::dbDisconnect(con_2)
# }