API says it's cloudy when it's sunny about open-meteo

[cmetz]

Ah, now I understand. I thought current_weather was not a forecast, but the last current weather conditions. Is there currently a possibility to query the actual current weather e.g. every 3 hours and are then "outdated" values still the forecast at that time, or are they replaced by the actual weather conditions?

It just surprised me that values in the past just significantly deviate from the weather conditions at that time, but if they are just the old forecast values then that's totally fine. :)

Hi @DonaldDuck313,

my first guess was as well, that there might be something wrong with the data. As I looked closer to the coordinates, I spotted a common mistake.

Lund is at 55° 42' N, 13° 11' E. This is a DMS coordinate in minutes and seconds. The API is using the decimal DD format. The correct decimal coordinate would be 55.70 N, 13.191 E.

32 kilometers difference could explain the wrong weather forecast.

Could you check, if the forecast is getting better with the correct decimal coordinate?

I will also think about improving the API documentation, as this is a quickly overlooked error.

Best, Jeff

[DonaldDuck313]

The app I'm making uses Android's Location.getLatitude() and Location.getLongitude() methods which seem to return the correct format (they return a double so that's the only thing that would make sense). So even though the link I sent you was incorrect, I'm quite sure the coordinates my app was using were correct.

Also at the same time as it sends a request to the API, my app also sends a request to Google's API to get the town name which says "Lund", so I'm quite sure there was nothing wrong with the location.

[cmetz]

I also have trouble with the WMO code. I also receive a 3 for cloudy, but it's sunny outside, maybe some high thin clouds, but not really cloudy. so maybe this codes just means it gets cloudier as described in:

https://artefacts.ceda.ac.uk/badc_datadocs/surface/code.html

code 4677 / 4680:

00 - Cloud development not observed or not observable
01 - Cloud generally dissolving or becoming less developed
02 - State of sky on the whole unchanged
03 - Clouds generally forming or developing

but i don't know how to determine the current cloud level.

maybe we need code 2700 for this:

WMO code 2700: Cloud cover / amount

/ - cloud is indiscernible for reasons other than fog or other meteorological
phenomena, or observation is not made.
0 - sky clear
1 - 1 okta : 1/10 - 2/10
2 - 2 oktas : 2/10 - 3/10
3 - 3 oktas : 4/10
4 - 4 oktas : 5/10
5 - 5 oktas : 6/10
6 - 6 oktas : 7/10 - 8/10
7 - 7 oktas or more, but not 8 oktas : 9/10 or more, but not 10/10
8 - 8 oktas : 10/10
9 - sky obscured by fog or other meteorological phenomena

update

it seems that you can get this values by using this options, for the hourly weather, but they are note included in the current weather option:

cloudcover         Instant 	% 	Total cloud cover as an area fraction
cloudcover_low     Instant 	% 	Low level clouds and fog up to 3 km altitude
cloudcover_mid     Instant 	% 	Mid level clouds from 3 to 8 km altitude
cloudcover_high    Instant 	% 	High level clouds from 8 km altitude

also the overall cloudcover is very "aggressive", i get nearly 100% but the mid and lower clouds are nearly at 0%, so maybe just using the cloudcover_low seems to correspond better to the impression of a person. But I am not a metrologist. :)

[patrick-zippenfenig]

The weathercode is supposed to be an instantaneous value. Of course there are forecast inaccuracies. @cmetz for which coordinates did you observe this?

While it is sunny and the weather code was indicating cloudy, did you see also high cloud cover values in low, mid and. high cloud cover values?

In the next months there will be additional weather forecast domains from more national weather services implemented. Depending on your area of interest, they can improve forecasts.

[cmetz]

@patrick-zippenfenig Yesterday i used Magdeburg (GER) as location. the current weathercode from the api was 3 and the forecast weather was reporting 100% high clouds and 0% for the others at this point in time. there were some high clouds, but all other weather sites reported the current weather as sunny e.g. https://wttr.in/

i currently use this API call, to only fetched the current_weather, but I'm not sure if this is only returning the least current forecast, or the last reported actual weather:
https://api.open-meteo.com/v1/forecast?latitude=52.12773&longitude=11.62916&current_weather=true&timeformat=unixtime

here are some screenshots from yesterday:

open-meteo api:

https://www.timeanddate.de/wetter/deutschland/magdeburg/rueckblick

[patrick-zippenfenig]

Right now, it is only returning the least recent forecast hour. In Germany updates happen every 3 hours (Data from Germany weather service DWD, model ICON-D2). I am a bit surprised that it is showing that many high clouds while it is sunny. Were there still scattered cirrus clouds?

Arguably, even at full high cloud cover it could still be fair weather. In case of ICON-D2 the weather code is directly supplied by DWD as-well. I could modify the weather code in those conditions, but I am not sure if this is worsening forecast for other locations.

There are some improvements coming in the next months:

• DWD MOSMIX for 1h updates for larger cities
• Additional weather models from other national weather services

Could you compare the weather forecast for the next couple of days and report back? So far I do not see a systematic error other than the forecast just being wrong from the German weather service

[patrick-zippenfenig]

Closing for cleanup. Feel free to reopen if you have more questions