Climate change projections

action0.open_meteo.climate · Open-Meteo docs · ClimateClient, preset to https://climate-api.open-meteo.com

GetV1Climate serves downscaled CMIP6 (HighResMIP) projections from 1950 to 2050, daily resolution. start_date and end_date are required; pick one climate model per request (see the note below):

import datetime

from action0.client.backends.requests import RequestsBackend
from action0.open_meteo.climate import (
    ClimateClient,
    GetV1Climate,
    GetV1ClimateDailyItem,
    GetV1ClimateModelsItem,
)

with RequestsBackend() as backend:
    client = ClimateClient(backend)
    projection = client.send(
        GetV1Climate(
            latitude="48.21",
            longitude="16.37",
            start_date=datetime.date(2040, 1, 1),
            end_date=datetime.date(2040, 12, 31),
            daily=[GetV1ClimateDailyItem.TEMPERATURE_2M_MAX],
            models=[GetV1ClimateModelsItem.MRI_AGCM3_2_S],
        )
    )

assert projection.daily is not None
print(projection.daily.temperature_2m_max[:3])  # [3.2, 5.0, 6.7]

Notes

• One model per request. With a single entry in models= the API answers with the plain variable names the typed model declares. With several models it switches to per-model suffixed keys (temperature_2m_max_MRI_AGCM3_2_S, …) that the upstream schema does not declare — those keys are dropped by the generated model, so comparing models means one request per model. (Upstream schema gap; tracked in schemas/README.md.)

• Projections show climate, not weather: analyze distributions over years, never a single simulated day.

• disable_bias_correction=True returns the raw model output.