Source of data in /stations/{stationId}/observations/latest endpoint

[DeanDavis]

Just starting to mess with this API and I'm always trying to reconcile the data I'm getting from the endpoint with some secondary source on a NWS website to see if they match.
I'm focused on the KMKG station for now.
The METAR reports from the station are once an hour at the :55 mark.
But, I'm getting more frequent data on the endpoint.
Where is this data coming from?
The closest I seem to find that reports what appears to be the same data is
https://www.weather.gov/wrh/timeseries?site=KMKG

I think the data I see there should match what I'm seeing on the endpoint.
My concern is, more often then not, the data returned in the endpoint isn't always complete.
For instance a call to /stations/KMKG/observations/2025-10-21T16:25:00+00:00
Will yield a windDirection and windSpeed values of NULL
But on the weather.gov page those value do exist for the same time slot.
More unusual is the gust value is present in the API and does match the value on the website.
So just trying to reconcile what I'm getting on the API with it's source and clear up my confusion on the values.

[rr0ss0rr]

I'll explain this in a very basic view (since I have very basic knowledge) ..

As you found out, METAR reports come out hourly (usually at the :55 mark). You can get the latest from "aviationweather.gov" via API or "tgftp.nws.noaa.gov" via download through curl or wget.

The observations you see via stations, I believe are from AWOS (Automated Weather Observing System) that are updated every 20 minutes. In the past, there were large delays to the API endpoint (fixed now), so at the time I became more reliant on the METAR record. I've now incorporated a hybrid mode using data from both.

[DeanDavis]

Thanks, yes, I'm doing the same. I grab the METAR as a start then override any more recent data I get from the /stations/{stationId}/observations/latest endpoint

What I'm trying to do is true up the data.
Right now I'm assuming the data I see at
https://www.weather.gov/wrh/timeseries?site=KMKG
Is also AWOS data since the time stamps will match up with time stamps from the API but I was curious why, for example, the data I see on the weather.gov site always has data for wind speed yet the API will many times report NULL as wind speed for the same time stamp.

[rr0ss0rr]

That I can’t answer .. I also grab the wind data from METAR as well as temperature and dewpoint as that combo is almost guaranteed in METAR .. dewpoint is sometimes missing from the API.

[scadergit]

The API uses Madis for the upstream source. It may take up to 20m for Madis to finish QA after the observation is issued, and returns partial data during the process. The legacy sites--anything not API--use a direct source that is only AWOS.

[scadergit]

For forecast or observations?

[bcbeta]

For observations (snotel, etc)

[scadergit]

As an example, here is a snotel site in the API, which has a 3-hour precip property (currently null for various reason, but should typically be provided if the station is sending it): https://api.weather.gov/stations/COLC2/observations.

[bcbeta]

That is great, last winter i don’t think they were sending snowfall data but I’ll look for it this season. There are also other non-snotel MADIS stations that have precip reported by synoptic but not the api. I’ll try to find some examples.

edit- will snotel have snowfall and depth or just liquid precip amounts?

[sullynole]

There is a known issue with precip in API that is mentioned on the updates tab here: https://www.weather.gov/documentation/services-web-api

There also appears to be an issue with wind gust reports sometimes going to null after initially showing up. Please pass examples of this to nco.ops@noaa.gov