Recommendations for forecast service

107 views
Skip to first unread message

p q

unread,
Feb 29, 2020, 2:12:58 PM2/29/20
to weewx-user
Hi,

I have been happily running Weewx for three years now. A while back the WU forecasting stopped working. Based on what I've read here, I switched over to NWS. I'm running an old version and put off getting a better source until I upgraded. I'm in the process of upgrading to 4.0 and starting to think about what to do with the forecast function. 

I'm in the US (SF Bay Area). Does anyone have any recommendations for what forecast source to use? Darksky? Windy? Give WU another shot?

vince

unread,
Feb 29, 2020, 2:26:22 PM2/29/20
to weewx-user
On Saturday, February 29, 2020 at 11:12:58 AM UTC-8, p q wrote:
I'm in the US (SF Bay Area). Does anyone have any recommendations for what forecast source to use? Darksky? Windy? Give WU another shot?

Why not NWS ?

 

p q

unread,
Feb 29, 2020, 3:09:13 PM2/29/20
to weewx-user
In a word, microclimates.

I'm using NWS now. It's good, but I found that WU was more accurate. Summer high/lows were closer to what I was actually getting. I live in a rain shadow and I think WU did a better job of taking that into account. The nearest NWS station is only 6 miles away but it could be 5 degF different from here in the summer. WU had a forecast for the zipcode. Maybe I'm deluding myself, but it seemed more accurate.

The second nearest is 15 miles away and could easily be 40 degF different in the summer. 
 I

Les Niles

unread,
Feb 29, 2020, 3:19:03 PM2/29/20
to weewx-user
Do you use the NWS spot forecasts?  I live on a ridge in the Santa Cruz mountains with no NWS stations around and weather that is very different than the valleys on either side.  The NWS forecast is pretty accurate, and much better than DarkSky which doesn’t seem to understand that we’re above the usual summer inversion.  I extended the weewx forecast module to support the NWS spot forecasts.

  -Les

--
You received this message because you are subscribed to the Google Groups "weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to weewx-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/weewx-user/3bff6d0a-2dd7-4559-b8c2-a7ea9c5f2563%40googlegroups.com.

Greg Troxel

unread,
Feb 29, 2020, 3:37:51 PM2/29/20
to Les Niles, weewx-user
Les Niles <nile...@gmail.com> writes:

> Do you use the NWS spot forecasts? I live on a ridge in the Santa
> Cruz mountains with no NWS stations around and weather that is very
> different than the valleys on either side. The NWS forecast is pretty
> accurate, and much better than DarkSky which doesn’t seem to
> understand that we’re above the usual summer inversion. I extended
> the weewx forecast module to support the NWS spot forecasts.

Agreed that NWS point forecasts (as they call them; I'm guessing that
spot is another word for the same thing) are good.

The other question is how many places are actually creating forecasts,
vs repackaging NWS forecasts without attribution. While that's legal as
they are in the public domain, being works of the United States, I don't
think it's polite. (I don't know what DarkSky is actually doing.)

It seems that NWS, and perhaps some other governments, run models and
make the results, and then NWS makes forecasts from multiple models
(applying forecaster judgement). TV stations etc. also make forecasts
from the models, and only sort of attribute the model data as
taxpayer-funded NWS outputs.

So I guess the big question is if DarkSky is

1) just publishing NWS forecasts (and perhaps other agency forecasts in
other places, like UK and the Met Office)

2) repackaging some other forecasts based on NWS models

3) using NWS data and making their own forecasts

4) repackaging some other forecast not based on NWS models

I think 1 is mostly likely, because 2 and 3 seem to involve having to
pay (or put up with TV ads) and adverse terms, and 4 I have never heard
of.

rich T

unread,
Feb 29, 2020, 3:57:49 PM2/29/20
to weewx-user

vince

unread,
Feb 29, 2020, 4:17:53 PM2/29/20
to weewx-user
On Saturday, February 29, 2020 at 12:09:13 PM UTC-8, p q wrote:
In a word, microclimates.

I'm using NWS now. It's good, but I found that WU was more accurate. Summer high/lows were closer to what I was actually getting. I live in a rain shadow and I think WU did a better job of taking that into account. The nearest NWS station is only 6 miles away but it could be 5 degF different from here in the summer. WU had a forecast for the zipcode. Maybe I'm deluding myself, but it seemed more accurate.

The second nearest is 15 miles away and could easily be 40 degF different in the summer. 



Yup.   Can't help much there.  I've found DarkSky is absurdly wrong here just south of SeaTac airport so I switched back to NWS since it was closer, although yes it's off for me too since I'm only a mile or so from the water.  The airport can be quite different in its conditions.
 

p q

unread,
Feb 29, 2020, 4:31:09 PM2/29/20
to weewx-user
With NWS, I'm using foid=MTR and lid=CAZ510 - which is the Concord Airport. Can I get it better refined than that?

I'm using the instructions posted here: https://github.com/weewx/weewx/wiki/forecasting

Is there a better way?


On Saturday, February 29, 2020 at 11:12:58 AM UTC-8, p q wrote:

mwall

unread,
Feb 29, 2020, 4:32:06 PM2/29/20
to weewx-user
On Saturday, February 29, 2020 at 3:19:03 PM UTC-5, ln77 wrote:
 I extended the weewx forecast module to support the NWS spot forecasts.

les,

could you make a merge request with your spot forecast changes so that i can incorporate that functionality into the weewx-forecast extension?


Les Niles

unread,
Feb 29, 2020, 5:29:42 PM2/29/20
to weewx-user
Yes.  It would be great to get it merged.  

  -Les

--
You received this message because you are subscribed to the Google Groups "weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to weewx-user+...@googlegroups.com.

Anton vanNwnhzn@GMail

unread,
Mar 1, 2020, 1:53:28 AM3/1/20
to weewx...@googlegroups.com

Another entry for comparison:

https://www.wxsim.com/accu.htm

Op 29-2-2020 om 22:17 schreef vince:
--
You received this message because you are subscribed to the Google Groups "weewx-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to weewx-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/weewx-user/03a850b1-cc06-45c4-a8c2-3aa0dc739847%40googlegroups.com.
-- 

===============================================================
Contactinfo voor Anton van Nieuwenhuijzen:
Email    = ton...@gmail.com
Fax2Mail = (+31/0)84.8397303 [ook Voice2Mail]   
===============================================================
Deze E-mail en eventuele aanhangende files zijn 
alleen bestemd voor de geadresseerde(n). 
Als je deze E-mail ten onrechte hebt ontvangen, 
dan aub verwijderen en de afzender informeren.
Reply all
Reply to author
Forward
0 new messages