Stations#

easysnowdata.stations#

Snow-station observations from five public networks (§9).

import easysnowdata as esd

aoi = (-121.94, 46.72, -121.54, 46.99)                 # Mount Rainier
inv = esd.stations.inventory(aoi, daily_only=True)      # GeoDataFrame
obs = esd.stations.load(inv, variables=["swe", "snwd"],
                        time="2023-10/2024-06")         # (station, time)
everything = esd.stations.archive.load()                # every daily station

Three layers, as global_snow_networks’ DESIGN.md §2 lays them out:

clients: Pure access, moved here verbatim with its history: one client per network (AWDB, CDEC, DataBC, NVE, Yukon), each answering get_all_stations / get_data / get_metadata with plain dict records in metric units. Use it directly when you want flags, a native variable name or an interval the adapter does not surface.
this module: inventory() and load(), which put those records into the package’s own return types and take aoi and time like every other loader.
archive: The daily archive that repo publishes — a normalized station inventory and one pre-downloaded CSV per daily-verified station — for “everything daily since 1980” without hitting five APIs. SWE and snow depth only.

The station × time grid comes back dense with NaN where a station has no observation. That is what xarray, Dask and groupby want, and a station record is only sparse in the everyday sense — nothing here needs a sparse array backend. Station positions are latitude and longitude coordinates rather than an xvec geometry coordinate: the plan lists xvec as optional, it would be a new dependency, and float coordinates round-trip through Zarr and netCDF where a geometry object does not.

Station identity has two spellings, both accepted everywhere: the globally unique code ("679_WA_SNTL") and the network’s own station id ("679:WA:SNTL"). See easysnowdata.stations.networks.

`inventory`	Snow stations as a GeoDataFrame, indexed by their global station code.
`load`	Station observations as a `(station, time)` Dataset.
`metadata`	Full per-station metadata from the owning network's API.

easysnowdata.stations.archive#

The daily station archive: everything daily, without hitting five APIs.

global_snow_networks pre-downloads daily SWE and snow depth for every station its probe has verified as daily-or-better, and publishes two artefacts that this module reads (§9 step 3):

all_snow_stations.geojson: Every station from all five networks, periodic sites included, with the 26 normalized properties of that repo’s DESIGN.md §6.1 — including daily_or_better, which is the probe’s verdict rather than what a network advertises about itself.
data/all_station_csvs.tar.xz: One date,wteq_cm,snwd_cm CSV per daily-or-better station, bundled. About 28 MB, and the cheapest way by far to get every station’s whole record — which, for a few long snow courses, reaches back to 1896.

import easysnowdata as esd

inv = esd.stations.archive.inventory(daily_only=True)   # one request
ds = esd.stations.archive.load(aoi=(-121.94, 46.72, -121.54, 46.99))

Both artefacts are refreshed daily by that repo’s CI, so the archive is at most a day behind and covers only SWE and snow depth. For anything else — a different variable, an interval other than daily, quality flags, or today’s observation — use easysnowdata.stations.load(), which goes to the networks themselves.

`csv_url`	URL of one station's archive CSV.
`inventory`	The published station inventory as a GeoDataFrame indexed by code.
`load`	Daily SWE and snow depth for many stations, from the published archive.
`network_of`	`{code: network}` from the inventory, for codes whose shape is ambiguous.

easysnowdata.stations.clients#

`AWDBClient`	Client for the USDA NRCS AWDB REST API v1.
`AWDBError`	Raised when the AWDB API returns an error or a request fails.
`CDECClient`	Client for CDEC (California Data Exchange Center).
`CDECError`	Raised when the CDEC API returns an error or a request fails.
`DataBCClient`	Client for BC snow survey data via BC Data Catalogue.
`DataBCError`	Raised when a DataBC request fails.
`NVEClient`	Client for the NVE HydAPI (Norwegian hydrological data service).
`NVEError`	Raised when the NVE HydAPI returns an error or a request fails.
`YukonClient`	Client for the Yukon Water Data (AquaCache) API.
`YukonError`	Raised when the Yukon AquaCache API returns an error or a request fails.

easysnowdata.stations.networks#

The five snow-station networks, and how a station code maps to one of them.

Each Network ties a catalog source id to the vendored client that speaks to it (easysnowdata.stations.clients) and to the spelling of a station identifier in that network. Two spellings matter:

code: The globally unique code the inventory and the archive use, e.g. "679_WA_SNTL". Unique across all five networks.
station_id: What the network’s own API calls the station, e.g. the AWDB triplet "679:WA:SNTL". Only AWDB’s two spellings differ; for the other four the code is the station id.

split_by_network() turns a mix of codes, station ids and inventory rows into {network_id: [station_id, …]}, which is what easysnowdata.stations.load() fans out over.

`Network`	One snow-station network and the client that reads it.
`canonical_units`	The unit `easysnowdata.stations.load()` emits for type_name.
`get`	Return the `Network` called network.
`split_by_network`	Group stations by network, as `{network_id: [station_id, …]}`.
`to_code`	The globally unique code of station_id in network.
`to_station_id`	The native station id of station in network.