honestly … no idea … maybe to reduce traffic because the gif is quite big ?
no idea 
Fair Point 5-6 MB for every call is quite something. Then again, the load is on the client and the network, but not HA - so I take it…
Unfortunately, I have no idea how to graph this data, but I can give a direction to look for:
You would need to combine the forecast sensors, some templating and a way to show all in a graph. Hope this helps!
My hope was that I could reuse that feature in a clean way for this:
It works as is with a highly modified version of mini-graph … I guess one could also create a version that works with your data …
Oh, I didn’t know this possibility! If you need a change in the data format, let me know
This is based on this config:
- type: custom:long-term-card
data_field: 1
name: 'Precipitation Celle (5d)'
display_mode: 'abs'
formater: "--datetime:--yyyy-mm-dd-hh-ii-ss"
date_out_format: 'yyyy-mm-dd HH:MM'
data_property: 'data'
lower_bound: 0
upper_bound: 100
maxDays: 5
data_delimiter: 'value'
unit: '%'
entities:
- sensor.cloud_coverage_celle
- type: custom:long-term-card
data_field: 1
name: 'Precipitation Probability Celle (5d)'
display_mode: 'abs'
formater: "--datetime:--yyyy-mm-dd-hh-ii-ss"
date_out_format: 'yyyy-mm-dd HH:MM'
lower_bound: 0
upper_bound: 100
data_property: 'data'
maxDays: 5
data_delimiter: 'value'
unit: '%'
entities:
- sensor.precipitation_probability_celle
- type: custom:long-term-card
data_field: 1
name: 'Temperature forcast Celle (7d)'
display_mode: 'abs'
formater: "--datetime:--yyyy-mm-dd-hh-ii-ss"
date_out_format: 'yyyy-mm-dd HH:MM'
data_property: 'data'
maxDays: 7
data_delimiter: 'value'
unit: '%'
entities:
- sensor.temperature_celle
And this card:
https://raw.githubusercontent.com/KoljaWindeler/long_term_card/master/dist/long-term-card-bundle.js
This is a highly hacked version of the mini-graph-card but sadly modified so heavily that it doesn’t seem possible to integrate that back into the original project. Also the documentation is super poor as of now … but … have a look if it works for you
data_property, formater, display_mode & data_field have to be left as is to work with your data.
JKW
1 Like
that looks nice. Could you give a little hint how to install? I added long-term-card-bundle.js to the ressources as ‘/local/long-term-card-bundle.js’ and stored the file in config/www but still “no card type found”
Thanks!
Look right, can you open http://yourip:8123/local/long-term-card-bundle.js ?
- url: /local/long-term-card-bundle.js
type: module
I don’t think that there should be more to it …
JKW
http://yourip:8123/local/long-term-card-bundle.js results in 404.
Running the community package (Hass.io on Synology DSM (native package)) in docker on a synology. Can this be the problem?
Problem solved, got lost in directory structure 
works great!
Hey @FL550. I’m continuing from what I started at a github issue (don’t want to go off topic there).
I found this here on the opendata server which actually seems to contain real life data instead of predictions. Might this be “more” useful than the predictions in some cases (eg. cloud coverage)?
The problem with the data you have found is, that only a few stations are available and not all of these contain all measurements. For example your requested cloud coverage is missing on most of them.
I know hourly forecasts (which should be more precise) can be found here, but the problem is that this file contains all stations and after extraction uses over 600MB of space. Parsing this every hour puts a huge load on the storage device, because this can’t be done in RAM like with the 6 hour forecast.
I’ve also found some data here. The data seems a bit more actual then the 6-hour forecast. Problem with this data is, that it is infrequently updated.
Hi just a quick noob question: I’m using this integration and activated the sensors “weather” and “temperature”. Do I have to do anything else? Or will my weather card automatically “integrate” those sensors?
As far as I understand, activating a sensor will update the values hourly instead of every 6 hours with just the weather entity?
My card-code:
type: weather-forecast
entity: weather.dwd_weather_“my location”
name: “my location”
secondary_info_attribute: temperature
show_forecast: true
(I replaced my actual location with “my location” for the sake of securtiy…sorry I’m a bit paranoid with these things…)
Thanks a lot guys!
You don’t need to activate the sensors if you are using the weather card. The sensors only gives you access to the raw hourly data if you need them.
The underlying data is only updated every 6 hours by DWD, so no quicker updates are possible.
The configuration you shared should work.
Ok, thank you!!
Hi,
I would love to have the DWD forecast for my town but I cannot get the integration to work.
I run HA core in a docker container and I have HACS installed. I can install the “Deutsche Wetterdienst (DWD)” integration, folder ‘custom_components/dwd_weather’ is created but the setup dialog does not show up. No sensors are created (which probably is a result of not having configured my desired station_id).
I deinstalled and reinstalled a few times - no change.
Is there an other way of triggering the setup dialog or configuring the integration manually?
Thanks in advance.
1 Like
Hi, after the installation you also have to add the integration to use it. Have you followed the instructions? https://github.com/FL550/dwd_weather#installation
Only if you add the integration, the setup dialog is shown.
Thanks FL550,
I thought that would show up after installation. Sorry for the stupid question and thanks for your work on that integration.
1 Like
Hi again,
I’m still somewhat curious about those sensors…So I’m using this weather card: https://github.com/bramkragten/weather-card which has the ability to show “hourly forecast” info.
I tried to do this with the sensor.weather entity…but this doesn’t seem to work. In developer tools the sensor state only shows “ok” - however the hourly weather data values (cloudy, rainy etc) seem to be received.
I believe that the weather card doesn’t show any info as “ok” isn’t a state it can deal with. Is it somehow possible to change the sensor’s state according to the value it receives?
Or am I thinking in the wrong direction here?
Thank you!