DWD WarnApp Sensor (amtliche Warnungen des deutschen Wetterdienstes)

that worked. thanks!

After sending an email to dwd about the cert, i got an answer today that they made some modifications.
Everything works now as before, no need to copy the certs.

THX to DWD! :+1:

3 Likes

Wow, thank you for your effort :+1:

Thanks for the plugin!

Occasionally, I see some ā€œsensor xxx is taking over 10 secondsā€ log entries. Thatā€™s why I started changing the component so it uses the new WFS API instead of the JSON file. While the JSON file contains all regions, WFS allows filtering.
I looked at the data and realised that no warning level in terms of integer is provided. Nevertheless, there is a key ā€œdwd:SEVERITYā€ containing a string rating.
Does any of you know how to map e.g. ā€œModerateā€ to an integer value?

My current progress and an example API response (attached at the end) can be found here:

Thanks!

Hi everyone,

Iā€™ve found a mapping between serverity and int rating and continued working on the plugin :ok_hand:
In case anyone wants to test: Please feel free (see github):slight_smile:
Working fine for me until now :grimacing:

Feel free to create a pull request with your solution. I think it is better to use the filtered API for low power devicesā€¦

I want to reopen the discussion on this topic because i lately realized that there is something wrong with the warning level of the current implementation.

For example: For my region ā€œLandkreisā€ currently a ā€œFROSTā€ warning is issued as level 1 (yellow) on the DWD homepage. The warning level of the HA sensor is 2 for the same region. This is not a HA sensor bug because the level 2 is issued by dwd in their json (https://www.dwd.de/DWD/warnungen/warnapp/json/warnings.json).

So i think it is time to switch to their new format like @Hummel95 did it already.

After i read this thread i did some further investigations and found out that we could improve this sensor much more because the DWD offers more and more detailed informations with their new API. With their new API not only ā€œLandkreiseā€ can be queried by also ā€œGemeindenā€ or ā€œStadtteileā€ of bigger cities. The following file contains a list of all available so called warncells witch can be queried.
https://www.dwd.de/DE/leistungen/opendata/help/warnungen/cap_warncellids_csv.html
Cell starting with ā€œ1ā€ represent ā€œLandkreiseā€. Cells starting with ā€œ2ā€ represent ā€œBinnenseenā€. Cells starting with ā€œ4ā€ represent ā€œNord- und Ostseeā€. Cells starting with ā€œ5ā€ represent ā€œKĆ¼steā€. Cells starting with ā€œ7ā€ or ā€œ8ā€ represent ā€œGemeindenā€ or ā€œStadtteileā€.

With the following query one can check if the warncellid is correct

https://maps.dwd.de/geoserver/dwd/ows?service=WFS&version=2.0.0&request=GetFeature&typeName={region_type}&CQL_FILTER=WARNCELLID=%27{warncellid}%27&OutputFormat=application/json

{region_type} must be the right one of the following
dwd:Warngebiete_Kreise
dwd:Warngebiete_Binnenseen
dwd:Warngebiete_See
dwd:Warngebiete_Kueste
dwd:Warngebiete_Gemeinden

{warncellid} your desired warncellid

The query to get the warnings for your cell is

https://maps.dwd.de/geoserver/dwd/ows?service=WFS&version=2.0.0&request=GetFeature&typeName={region_type}&CQL_FILTER={cql_filter_property}=%27{warncellid}%27&OutputFormat=application/json

{region_type} must be the right one of the following
dwd:Warnungen_Kreise
dwd:Warnungen_Binnenseen
dwd:Warnungen_See
dwd:Warnungen_Kueste
dwd:Warnungen_Gemeinden

{cql_filter_proerty} must be
GC_WARNCELLID for dwd:Warnungen_Kreise
WARNCELLID for all other types

{warncellid} your desired warncellid

I already started working on an improvement of Hummel95ā€™s solution and itā€™s working fine so far but a found one problem i want to start a discussion about. In the current sensors configuration one has to enter the regionā€™s name to configure the sensor. This is fine for ā€œLandkreiseā€. The problem is that the regionā€™s name is not unique for ā€œGemeindenā€ e.g. ā€œGemeinde Oberhausenā€ exists two times in the warncell list.

Is it therefore ok to change the required configuration from ā€œregion_nameā€ to ā€œwarncellidā€ in configurations.yaml?
After an update of the sensor all users would have to update their configurations.yaml to get the sensor working again. Otherwise it cannot be guaranteed that the right cell is selected.

Hi Stephan,

Thanks for working on this! I stopped running home-assistant last year but Iā€™m planning on installing it again soon.

In my opinion changing the configuration parameter to warncellid makes sense especially since itā€™s justified by an increased prediction accuracy.

If you havenā€™t checked out my pull request please do so. The HA team requested some changes in the code to comply with their rules. https://github.com/home-assistant/core/pull/23234

Thanks!

After reading the pull request i realize that there is more work to do than i initially thought.

First step: Splitting sensor and API
Check if one of the on PyPI available DWD APIs fits our needs, or if not program a new one.

After a quick check it seems that there is no DWD API on PyPI which uses there OWS service.

I think i will start trying, never did this before, to develop a dwdows PyPI package which than can be used for the ā€œWarnwetterā€ and maybe a ā€œPollenflugā€ or ā€œBiowetterā€ sensor.

1 Like

I just finished my first version of the API, i called it dwdwfsapi because the DWD calls its underlying service Web Feature Service (WFS).

For those who are interested:
Link to PyPI
Link to GitHub

The next days i will start with step two: embedding the API in home-assistant

3 Likes

Feel free to replace my current codeā€¦

I know this is slightly off topic, but seen as darksky (which most people were using for forecasts) has been bought by apple is there any chance to upgrade the dwd integration to also pull weather data and forecasts?
Using the warnwetter app gives me the best forecasts for my area anyway and I would love to have that level of exactness in home assistant.
Of course this would be only interesting for us Germans but then I guess the warning levels currently supported are also of not much use for someone in Australia :wink:
Sadly I know nothing about coding and using the api but it looks like you are quite well versed in thisā€¦
So maybe someone is up for the task of integrating the weather data as well! Thanks!

1 Like

Try to use Yr.NO. Itā€™s very accurate in Dus and uses the same weather forecast model like DWD.