canwxdb-py

Python client for the wxdb.ca API — historical Canadian weather observations from Environment and Climate Change Canada, queryable by geography and time range.

wxdb.ca is a community project and not affiliated with or endorsed by Environment and Climate Change Canada.

Installation

pip install canwxdb

Optional extras for richer output formats and progress bars:

pip install canwxdb[pandas]      # DataFrame support
pip install canwxdb[geo]         # GeoDataFrame support (includes pandas)
pip install canwxdb[progress]    # tqdm progress bars
pip install canwxdb[all]         # everything above

Quick start

from canwxdb import Client

client = Client()

Look up a station

station = client.station(52)
# {'type': 'Feature', 'geometry': ..., 'properties': {'name': 'ESQUIMALT HARBOUR', ...}}

Find stations near a location

# By lat/lon + radius
stations = client.stations(lat=48.43, lon=-123.37, radius_km=50)

# By Canadian postal code
stations = client.stations(postal_code="V8W1A1", radius_km=25)

# By bounding box  (min_lon, min_lat, max_lon, max_lat)
stations = client.stations(bbox=(-124.5, 48.3, -123.0, 49.0))

Fetch observations

# Daily observations for a single station
obs = client.daily(station_id=52, start="2024-01-01", end="2024-12-31")

# Hourly observations
obs = client.hourly(station_id=52, start="2024-07-01", end="2024-07-31")

# Monthly observations
obs = client.monthly(station_id=52, start="1990-01-01", end="2024-12-31")

All three methods accept the same geographic filters as stations().

Date arguments accept ISO 8601 strings, datetime.date, or datetime.datetime objects.

Get a pandas DataFrame

df = client.daily(
    station_id=52,
    start="2024-01-01",
    end="2024-12-31",
    format="dataframe",
)

df.head()
#         date  station_id  mean_temp_c  ...  latitude  longitude
# 0 2024-01-01          52         -1.2  ...     48.43    -123.43
# 1 2024-01-02          52          0.5  ...     48.43    -123.43

Get a GeoDataFrame (GeoPandas)

gdf = client.daily(
    postal_code="V8W1A1",
    radius_km=100,
    start="2023-01-01",
    end="2023-12-31",
    format="geodataframe",
)

# Plot mean temperature across stations
gdf.plot(column="mean_temp_c", legend=True, figsize=(10, 6))

The GeoDataFrame uses WGS-84 (EPSG:4326) and has a Point geometry column built from each station's coordinates.

Progress bars (recommended for large queries)

For multi-page queries — especially in Jupyter notebooks — pass show_progress=True to see live row counts. The library automatically uses tqdm.notebook inside Jupyter and plain tqdm elsewhere.

# In a Jupyter notebook this renders a rich notebook-style progress bar.
df = client.daily(
    bbox=(-124.5, 48.3, -123.0, 49.0),
    start="2015-01-01",
    end="2024-12-31",
    format="dataframe",
    show_progress=True,
)
# Fetching: 45000 rows [00:18, 2490 rows/s]

Limit pages for exploratory queries

When exploring a new dataset, use max_pages to avoid accidentally fetching millions of rows:

# Just peek at the first page
df = client.hourly(
    bbox=(-80, 43, -76, 44),
    start="2020-01-01",
    end="2023-12-31",
    format="dataframe",
    max_pages=1,
)

Using as a context manager

with Client() as client:
    df = client.daily(station_id=52, start="2024-01-01", end="2024-12-31",
                      format="dataframe")

Connecting to a custom endpoint

client = Client(base_url="https://my-private-instance.example.com")

Output formats

Error handling

from canwxdb import Client, WxDBNotFoundError, WxDBRateLimitError

client = Client()

try:
    station = client.station(99999)
except WxDBNotFoundError:
    print("Station not found")

# The client automatically retries 429 responses with exponential back-off.
# WxDBRateLimitError is raised only if all retries are exhausted.

Data source

Weather data is sourced from Environment and Climate Change Canada's open data under the Government of Canada Open Government Licence.