The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
import_ghcn_hourly() will now re-attempt to download
data up to three times if the first attempt fails. This makes it more
robust to interrupted connections with NOAA’s data store.
import_ghcn_stations() now caches data per session,
speeding up data import.
The default provider has changed back to
OpenStreetMap from CartoDB.Voyager as issues
with OSM in Leaflet have been resolved.
getMeta(), getMetaLive(),
importNOAA(), importNOAAlite() and
exportADMS(). They continue to work, but will be removed in
a future release. Please use import_isd_stations(),
import_isd_stations_live(),
import_isd_hourly(), import_isd_lite() and
write_adms() instead.This update contains a myriad of minor bug fixes. The following may change the values returned by, or written from, existing code.
import_ghcn_hourly() now calculates total cloud
cover (cl/sky_cover) as the maximum of all
cloud layer amounts, as documented and consistent with
import_isd_hourly(). Previously, when
sky_condition did not report a total, the lowest layer’s
amount was used instead, under-reporting cloud cover in roughly 10% of
observations.
When hourly = TRUE,
import_ghcn_hourly() and import_isd_hourly()
no longer return duplicate hours for stations whose latitude, longitude
or elevation change within the requested period.
write_adms() now writes every year present in the
data. Previously only the year of the first row was written, and all
other data were silently dropped.
The following fixes are more minor and broadly make
{worldmet} more robust to variations in returned data.
import_ghcn_hourly():
minute, cloud layer heights or
sub-hourly precipitation).codes.abbr_names = FALSE.import_ghcn_daily():
append_codes = TRUE no longer errors for stations whose
attributes include an observation time (e.g., precipitation at many US
stations), and now returns flag columns when
extra = FALSE.sn**/sx**) are
now converted from tenths of degrees when only one such column is
present.import_ghcn_monthly_temp() now always reads data
flag columns as character, rather than sometimes as logical (e.g., a “T”
flag becoming TRUE).
import_ghcn_monthly_prcp() now filters the returned
data by year. Previously year was ignored, and
using it with table = "inventory" alone caused an
error.
import_isd_hourly():
pwc in the
global environment.source = "fwf": present weather (pwc)
is now returned when hourly = TRUE; fields appearing in
only one record of a file are no longer ignored; precip is
no longer derived from row positions when hourly = FALSE;
and station codes missing from the ISD metadata no longer cause an
error.import_isd_lite() now respects its
progress argument, and no longer errors when a station code
is missing from the ISD metadata.
write_met():
ext = "delim" now writes files, using the
delim argument and a .csv, .tsv
or .txt extension as appropriate. Previously no files were
written._.<ext> files are no longer written when
stations in the data do not share the same years.write_adms(interp = TRUE) no longer exits early
without writing a file when one of the interpolated variables is
entirely missing.
The styling of marker popups in
import_ghcn_stations() and
import_isd_stations() has been improved.
import_ghcn_stations() has gained
provider in line with
import_isd_stations().
The default provider has changed from
OpenStreetMap to CartoDB.Voyager due to
ongoing issues with OSM in Leaflet (https://github.com/rstudio/leaflet/issues/964).
!BREAKING: The lon argument of
import_isd_stations() has been renamed lng to
match import_ghcn_stations().
import_ghcn_hourly() and
import_isd_hourly() will now always return a complete year
of data - i.e., 8760 rows for a non-leap year. Missing values will be
padded with NA values (#59).
import_ghcn_inventory() is now less sensitive to
small changes in the format of NOAA’s ghcnh-inventory.txt
file.
import_ghcn_hourly() will now succeed even if
certain columns are not present in the raw data files.
import_ghcn_stations() now correctly show the WMO ID
(previously this was a duplicated GSN flag).
import_ghcn_hourly() now calculates
sky_cover from the sky_condition
column.
arrow has been added as a suggested
dependency.
leaflet has been downgraded to a suggested
dependency. When a user requests a met station map they will be prompted
to install leaflet if it is not already installed.
mirai and carrier have been upgrade to
imported dependencies. This is because they are always required by
purrr::in_parallel() regardless of whether parallelisation
is being used or not.
The Global Historical Climate Network (GHCN) has officially superseded NOAA’s Integrated Surface Database (ISD). Many changes in this release are related to this external change.
Added import_ghcn_stations(),
import_ghcn_inventory(), and
import_ghcn_countries() to grant access to metadata for the
GHCN.
Added import_ghcn_hourly(),
import_ghcn_daily() and import_ghcn_monthly()
to grant access to the GHCNh, GHCNd and GHCNm services.
Added import_isd_stations(),
import_isd_hourly() and import_isd_lite() to
provide a consistent API with the newly added GHCN functions.
Added write_met(), which allows for hourly ISD and
GHCN and daily GHCN data to be easily split into years and sites and
written to file. This is a more flexible alternative to using
importNOAA(path=).
Several functions have been deprecated and replaced due to the
updated worldmet API:
getMeta() ->
import_isd_stations()
getMetaLive() ->
import_isd_stations_live()
importNOAA() ->
import_isd_hourly()
importNOAAlite() ->
import_isd_lite()
exportADMS() ->
write_adms()
the path argument ->
write_met()
rlang::%||% is now explicitly imported by
worldmet to avoid issues in old version of R.worldmet no longer depends on
openair.importNOAA() now recommends trying the other
source if it fails to fetch data (i.e., if
source = "delim" fails, the user will be prompted to try
source = "fwf").Fixed source = "fwf" failing when importing data
from closed stations.
Fixed source = "fwf" when hourly = TRUE
and multiple sites are imported.
Removed present weather condition when
source = "fwf".
importNOAAlite() has gained the path
argument, in line with importNOAA().
importNOAAlite() now supports parallel processing
using the {mirai} package, in line with
importNOAA().
importNOAAlite() can now import data from years other
than 2025.Parallel importing of NOAA data is now powered by
{mirai}. This reduces the number of dependencies of
{worldmet}, and also allows users to be more flexible with
how parallel processing is achieved.
mirai::daemons() themselves. n.core will still
work for back-compatibility, but will give a once-per-session
warning.getMeta() has gained the crs argument
to search NOAA ISD stations by coordinates other than latitude and
longitude.
Added importNOAALite() to access the ISDLite
filestore.
Added the importNOAA(source=) argument to import ISD
data from different file stores. This option can be useful if one of the
file stores is not available for whatever reason.
All error and warning messages are now powered by
{cli} and are more informative.
quiet argument of importNOAA() now
also toggles the progress bar.getMeta() has gained the crs argument,
which allows getMeta(lat =, lon =) to be defined using
other coordinate systems.Fixed join problem due to {dplyr} updates.
Fixed spelling mistakes in weatherCodes
Converted the {worldmet} vignette to an article.
This should ensure it is no longer reliant on NOAA’s servers. It can now
be accessed at https://openair-project.github.io/worldmet/articles/worldmet.html.
importNOAA() will now display a progress bar when
multiple years of met data are to be imported.
Improved the formatting of the popups in getMeta()
in line with openairmaps::networkMap().
Added the provider argument to
getMeta(), allowing for users to define one or more
leaflet basemaps. Uses the same default as
openairmaps::networkMap().
exportADMS() now invisibly returns the input data,
in line with, e.g., readr::write_csv().
do not add precip to exportADMS()
Cloud cover data mostly spans 1 to 8 Oktas and no zero (clear
sky); use ceil_hgt = 220000 (unlimited) to indicate clear
sky when cloud cover is reported as missing.
Do not use ftp for site information.
precip_12 gives the sum of the
precipitation over the past 12 hours and the sum of this column should
be the annual total in mm. precip spreads the 12-hour total
evenly across the previous 12 hours. {worldmet} no longer
tries to combine 12 and 6 hour totals.importNOAA()weatherCodes so that everything works when
used by other packages through “explicit namespacing”” (e.g.
worldmet::importNOAA()) without having to
library(worldmet).fix bug when lat and lon provided in
getMeta()
fix bug when several years are selected and columns are different
lengths when n.core = 1
Significant changes due to NOAA storage formats and different storage locations
Remove options for precipitation and present weather in
importNOAA(); just return everything
Return data as {tibble}
export getMetaLive() to allow users direct access to
all meta data and easy re-use without re-downloading.
add option path to allow users to save met data as
an rds file.
deprecate fresh option in
getMeta().
Assume 9999 is missing for visibility (was
999999)
fix short WBAN codes.
make sure all meta data are used and not only sites with most recent year
query live meta data when using getMeta().
parallel processing for sites and years.
use readr::read_csv() for meta data
(read.csv() seems very slow in R 3.4.3).
downloads now from webserver rather than ftp. Should be faster and allow more downloads. Thanks to Stuart Grange.
add parallel processing using {foreach}.
closeAllConnections().default to downloading fresh meta data each time.
update meta data.
NA when missing and
not NaN.Add ability to return precipitation measurements, if available.
Add precipitation to exportADMS()
These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.