Historical weather

action0.open_meteo.historical_weather · Open-Meteo docs · HistoricalWeatherClient, preset to https://archive-api.open-meteo.com

GetV1Archive serves reanalysis data (ERA5 and friends) from 1940 onwards. Unlike the forecast operation, start_date and end_date are required — both are datetime.date fields and serialize to ISO 8601:

import datetime

from action0.client.backends.requests import RequestsBackend
from action0.open_meteo.historical_weather import (
    GetV1Archive,
    GetV1ArchiveDailyItem,
    HistoricalWeatherClient,
)

with RequestsBackend() as backend:
    client = HistoricalWeatherClient(backend)
    year_1990 = client.send(
        GetV1Archive(
            latitude="48.21",
            longitude="16.37",
            start_date=datetime.date(1990, 1, 1),
            end_date=datetime.date(1990, 12, 31),
            daily=[
                GetV1ArchiveDailyItem.TEMPERATURE_2M_MEAN,
                GetV1ArchiveDailyItem.PRECIPITATION_SUM,
            ],
            timezone="Europe/Vienna",
        )
    )

assert year_1990.daily is not None
print(len(year_1990.daily.time))  # 365
print(year_1990.daily.temperature_2m_mean[:3])  # [2.9, 3.8, 1.6]

Notes

• hourly= (GetV1ArchiveHourlyItem) gives hour-resolution history; the response blocks work exactly like the Weather forecast ones (parallel lists plus *_units companions).

• Reanalysis lags a few days behind real time — for the most recent days combine with the forecast API’s past_days=.

• Long ranges return long lists; one request per year keeps payloads comfortable.