Bug: Stations reporting in the future

[aschl]

The Meteostat's bulk data API (https://bulk.meteostat.net/v2/) returns stations that report weather data in the future.
What is the reason for this?

The data was downloaded on Jan. 6th, 2025. Some stations report weather up to a week later Jan 15th (which is in the future relative to the download date).

[clampr]

Hi @aschl,

This is expected behavior. Meteostat includes weather forecast models like DWD's MOSMIX in its time series. If you want actual observations only, please use the model parameter.

[aschl]

Thanks for your reply @clampr. I went through my code to double-check and I am not sure this is the case.
I downloaded the data in bulk and used the following URL: https://bulk.meteostat.net/v2/hourly/{station_id}.csv.gz
I think no model data should be contained or am I misunderstanding something here?

[aschl]

https://bulk.meteostat.net/v2/stations/full.json.gz contains the metadata for each station:
and looks something like

...
{
        "id": "00FAY",
        "name": {
            "en": "Holden Agdm"
        },
        "country": "CA",
        "region": "AB",
        "identifiers": {
            "national": "32395",
            "wmo": "71227",
            "icao": "CXHD"
        },
        "location": {
            "latitude": 53.19,
            "longitude": -112.25,
            "elevation": 688
        },
        "timezone": "America/Edmonton",
        "inventory": {
            "model": {
                "start": "2021-07-13",
                "end": "2025-01-21"
            },
            "hourly": {
                "start": "2020-01-01",
                "end": "2024-12-07"
            },
            "daily": {
                "start": "2002-11-01",
                "end": "2024-03-13"
            },
            "monthly": {
                "start": 2003,
                "end": 2022
            },
            "normals": {
                "start": 1991,
                "end": 2020
            }
        }
    },
        ...

Shouldn't the URL https://bulk.meteostat.net/v2/hourly/{station_id}.csv.gz only contain the hourly data from "2020-01-01" to "2024-12-07" in this case?
Instead the csv seems to also contain the model data?!

[...]
2025-01-20,00,-24.5,-27.0,80,,,168,18.0,,1038.4,,
2025-01-20,06,-28.0,-30.4,80,,,180,11.9,,1035.6,,
2025-01-20,12,-29.6,-31.7,82,,,321,12.6,,1035.0,,

The csv file does not flag which entries are modelled and which are actual measurement. Would it maybe make sense to clearly differentiate the hourly and model data so the pure hourly station data are contained in https://bulk.meteostat.net/v2/hourly/{station_id}.csv.gz and model data can be accessed via https://bulk.meteostat.net/v2/model/{station_id}.csv.gz?

[clampr]

Hi @aschl,

Inventory data is updated with delay. If you want to differentiate between the different data sources, please use source maps.