Attilla,

Thanks for your update.  I look forward to the new features.  That would be
very exciting if someone can get a European model going so you can also
cover Europe.

I'm wondering if you will experience problems with the increased spatial
resolution in the CMC images causing individual 1 hour CSC squares to be
less representative.  If the model gets to the point of producing a region
of speckled clear sky and cloud, you will probably have to average (i.e.,
blur) the model to get a representative forecast for the 1 hour slot.
Has this been a problem in the past?  Do you think it will be a problem
with higher resolution maps?
--
Steve Koehler
koehler@...

On Fri, 7 May 2004, Stephen C. Koehler wrote:

>
> Attilla,
>
> Thanks for your update.  I look forward to the new features.  That would be
> very exciting if someone can get a European model going so you can also
> cover Europe.
>
> I'm wondering if you will experience problems with the increased spatial
> resolution in the CMC images causing individual 1 hour CSC squares to be
> less representative.  If the model gets to the point of producing a region
> of speckled clear sky and cloud, you will probably have to average (i.e.,
> blur) the model to get a representative forecast for the 1 hour slot.
> Has this been a problem in the past?  Do you think it will be a problem
> with higher resolution maps?
> --
> Steve Koehler
> koehler@...

Indeed, it will be a problem for a few Clear Sky Clocks. I have already
did some animations to see how those little holes moves within low clouds.
I have seen a few point or grid points where holes doesn't move which will
give the impression to clear sky clocks users of many clear skies hours
when it will be cloudy all around. We found the problem which seems to be
related to lakes being ice covered (which should not be a problem for
southern regions). The way the low clouds are forecast, well what I can
say... they are not advected.
With icy lakes, well there is no moisture being evaporated and tile in the
model is dryer than surrounding tiles with snow or vegetation. So, over
the lake (or within 15km around the lake) the grid point is dryer and tend
to evaporate the clouds.
There is no easy solution for now, but we should add a patch in the model
to solve this freaky stationnary holes which appear occasionnally.

I have to say, I am quite happy with the general picture of the new 15km
cloud schemes. When compared to satellite pictures, it is getting close to
satellite picture quality... and it verifies!!!!

Have a good weekend!

                                     *               *
                                                         _~~~~~~~
Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

On Fri, 7 May 2004, Stephen C. Koehler wrote:


Just want to let you know the implementation of the higher resolution
cloud-transparency-seeing forecast based on the 15km resolution models
will done on May 18th for the noon integration.

                                     *               *
                                                         _~~~~~~~
Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

Allan,

> I have to say, I am quite happy with the general picture of the new
15km
> cloud schemes. When compared to satellite pictures, it is getting
close to
> satellite picture quality... and it verifies!!!!

When you have some time, I would be interested to hear more about how
you verify your models.  Do you continuously measure the prediction
against reality, or do you spot check?  How do you quantify the
closeness of the fit to reality?

-- Steve Koehler

--- In AstronomyWeather@yahoogroups.com, Allan Rahill <

> > I'm wondering if you will experience problems with the increased
spatial
> > resolution in the CMC images causing individual 1 hour CSC
squares to be
> > less representative.


> Indeed, it will be a problem for a few Clear Sky Clocks.

It's sometimes a problem now. A clock can display a long
string of clear blocks when there is a bank of cloud very closeby
moving in a paralell direction. A tiny error in the forecasted
position of the cloud bank, or the lat/long of the clock, can
mean difference between a long string of blue or a long string
of white block. Only one is likely to be right.

It's a problem unique to the way clear sky clocks show
Allan's forecasts data.

A solution the the  problem is one of the 4 features I'm working
on.

-ad

I'm often asked to create CSCs for europe. My answer is
always 1. CMC dosent forecast astronomy-weather for europe
and 2. I dont know of another cloud forecast that's accurate
enough to make a good clock.

Now Balthasar Indermuehle of Switzerland has written me:

>I have finally located weather info that may enable you to create
>clearskyclocks for europe.

And offered me the link:
http://www.unibas.ch/geo/mcr/3d/meteo/nmm22/nmm22.htm
for the "Cloud" or "Wolken" forecast.

It would be quite a bit of coding for me to create CSCs from
these Cloud maps. Before I start, I'd like some indication
of the accuracy of the forecasts.

They are generated from a forecast model called NNM4

Anyone hear have an opinion as to how good NNM4 is?

-ad

Attilla

Just one small data point.  I have been using their cloud cover forecast
for a while - I have not being logging it, but (from a non expert user)
it appears to be more accurate than most forecasts available in UK.

We would really REALLY welcome a SCS for Europe if the data is decent.

Thanks

Adam

> -----Original Message-----
> From: AstronomyWeather@yahoogroups.com
> [mailto:AstronomyWeather@yahoogroups.com]
> Sent: 17 May 2004 14:20
> To: AstronomyWeather@yahoogroups.com
> Subject: [AstronomyWeather] Digest Number 56
>
> ------------------------ Yahoo! Groups Sponsor
>
> There is 1 message in this issue.
>    Date: Sun, 16 May 2004 17:09:12 -0000
>    From: "Attilla Danko" <danko@...>
> Subject: clear sky  clocks for Europe
>
> I'm often asked to create CSCs for europe. My answer is
> always 1. CMC dosent forecast astronomy-weather for europe
> and 2. I dont know of another cloud forecast that's accurate
> enough to make a good clock.
>
> Now Balthasar Indermuehle of Switzerland has written me:
>
> >I have finally located weather info that may enable you to create
> >clearskyclocks for europe.
>
> And offered me the link:
> http://www.unibas.ch/geo/mcr/3d/meteo/nmm22/nmm22.htm
> for the "Cloud" or "Wolken" forecast.
>
> It would be quite a bit of coding for me to create CSCs from
> these Cloud maps. Before I start, I'd like some indication
> of the accuracy of the forecasts.
>
> They are generated from a forecast model called NNM4
>
> Anyone hear have an opinion as to how good NNM4 is?

On Sun, 16 May 2004, Attilla Danko wrote:

> I'm often asked to create CSCs for europe. My answer is
> always 1. CMC dosent forecast astronomy-weather for europe
> and 2. I dont know of another cloud forecast that's accurate
> enough to make a good clock.
>
> Now Balthasar Indermuehle of Switzerland has written me:
>
> >I have finally located weather info that may enable you to create
> >clearskyclocks for europe.
>
> And offered me the link:
> http://www.unibas.ch/geo/mcr/3d/meteo/nmm22/nmm22.htm
> for the "Cloud" or "Wolken" forecast.
>
> It would be quite a bit of coding for me to create CSCs from
> these Cloud maps. Before I start, I'd like some indication
> of the accuracy of the forecasts.
>
> They are generated from a forecast model called NNM4
>
> Anyone hear have an opinion as to how good NNM4 is?
>
> -ad

Attilla,

Forget it... there is no way to produce a CSC with these maps. This is
the problem with all other country cloud forecasts. There is no
resolution. In fact CMC global model has more resolution than NNM4 cloud
outputs and I always refused to produce cloud images with this model!


                                                         _~~~~~~~
Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

The powerfailure of last summer gave an accidental experiement where it
was observed that without all the power generation, air pollution goes
down quite a bit, making for clearer skies:

Here's a link to the Montreal Gazette
http://www.canada.com/montreal/montrealgazette/news/story.html?id=8f15c677-678c-\
4dff-8c58-4e5f148a28ca


And a few others:

http://www.newscientist.com/news/news.jsp?id=ns99995038
http://www.webindia123.com/news/showdetails.asp?id=39559&cat=Science
http://news.newkerala.com/world-news/index.php?action=fullnews&showcomments=1&id\
=20242

So there you have it, less pollution will increase sky quality. It
improved the transparency from 20km to 60km

For selected clear sky clocks, I've added a page that summarsizes
the distribution of all of the past forecasts (for which I have
records).

(Some people have been asking for this. But they really wanted
climate data based on observation. Even explaining "but it's just a
forecast, not a history of observations" didnt dampen their
ardor. So I've given in. I hope I've added enough disclaimers. )

For the summaries, I use only those forecasts of Allan that are
12 hours, or less old (i.e. highest accuracy).

here is a sample:
http://www.cleardarksky.com/clmt/c/JordLkNCct.html

Other similar pages are linked from some clock's legend pages.

a-d

Attilla, the summaries are great. But July is missing in the "by month" section.

-Florian

> Attilla, the summaries are great. But July is missing in the "by month"
section.

Thanks because I dont have data from last July. My archives, 20 gigs worth,
only go back as far as August 2003.

-ad

On Sat, 26 Jun 2004, Attilla Danko wrote:

> For selected clear sky clocks, I've added a page that summarsizes
> the distribution of all of the past forecasts (for which I have
> records).
>
> (Some people have been asking for this. But they really wanted
> climate data based on observation. Even explaining "but it's just a
> forecast, not a history of observations" didnt dampen their
> ardor. So I've given in. I hope I've added enough disclaimers. )
>
> For the summaries, I use only those forecasts of Allan that are
> 12 hours, or less old (i.e. highest accuracy).
>
> here is a sample:
> http://www.cleardarksky.com/clmt/c/JordLkNCct.html
>
> Other similar pages are linked from some clock's legend pages.
>
> a-d
>

Attilla,

Quite interesting this new tool you developped. I have some questions
though.

1- What is the time period you use to produce these probabilities? Is it
    12hr forecasts?
2- For the transparency forecast, I would take out the white boxes. As you
    know, the white appears in the forecast as a default when there are
more than 3/10 of clouds for a pixel. I do that to avoid forecasting
transparency when there are clouds in the sky... which cloud be
disastrousous when it is high thin cirrus cloud types. Actually the
percentage in your bar graphiques show how many times the model forecast
a sky with more than 3/10 of clouds... which is purely abitrary. Perhaps,
you should show only the percentage on the first 4 categories for the
transparency. Two weeks ago, I did a home experiment trying to add a 5th
category (like the seeing) for the transparency forecast. I didn't like
the final results and I sticked with the current forecast
categories.
I would suggest to calculate the propabilities only for the 4 categories
from light blue to dark blue. You can keep the white box (same calculation
as actual bar graph) to show how many times we get a sky with less than
3/10 of clouds.

I found quite interesting the seasonal variation we can see in those
graphs. I thing there are some usefull informations the users can use
despite these are from forecasts and not from observations.

Thanks Attilla for your excellent work again!


                                     *               *
                                                         _~~~~~~~
Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

--- In AstronomyWeather@yahoogroups.com, Allan Rahill
<allan.rahill@e...> wrote:
> On Sat, 26 Jun 2004, Attilla Danko wrote:
>
  > 2- For the transparency forecast, I would take out the white
boxes. As you
>    know, the white appears in the forecast as a default when there
are
> more than 3/10 of clouds for a pixel.

Ok. I no longer count "white" in either transparency or seeing
forecasts. That does however make the stats difficult to interpret.

To make interpretation a little easier, I've counted up
the number of times seeing,cloud cover, and transparency
forecast conditions for good observing. Here are two examples:

http://www.cleardarksky.com/clmt/c/Torontoct.html#ObservingConditions
http://www.cleardarksky.com/clmt/c/CottCampCAct.html#ObservingConditio
ns

The real question is "why the heck are we not all living
in California?"

-ad

Attilla,

I thought of something else for your history pages.  I'm curious to know how
accurate the forecasts are for 24+ hours out, versus the one that is 0+ hours
out.  In other words, if I see a forecast for tomorrow, how likely are the
color squares for the same time to look similar when tomorrow's clock comes
out.  I'm not sure how you would do this, but it would be interesting to have
a gauge on the stability of the longer-range forecasts for an area.  I
suspect that some regions have much greater stability than others.

-- Steve Koehler
    steve_koehler@...

On Thu, 15 Jul 2004, Koehler, Steve wrote:

> Attilla,
>
> I thought of something else for your history pages.  I'm curious to know how
> accurate the forecasts are for 24+ hours out, versus the one that is 0+ hours
> out.  In other words, if I see a forecast for tomorrow, how likely are the
> color squares for the same time to look similar when tomorrow's clock comes
> out.  I'm not sure how you would do this, but it would be interesting to have
> a gauge on the stability of the longer-range forecasts for an area.  I
> suspect that some regions have much greater stability than others.
>
> -- Steve Koehler
>    steve_koehler@...
>
>

A lot of computing but not necessarely a good idea. The initialisation of
the model could be wrong for clouds since it is a diagnostic from moisture
and not the reality. The best is to use satellite data for verifications,
but again it is hard to develop a tool to know if there is cloud or not
for each pixel due to cloud tempartures being very often as cold as the
ground. Beleive me it takes a lot of computation with numeros channel from
GOES satellite to know if there is low clouds or not during the winter
season all across north america.

This kind of study will be interesting as you mention which should vary
from region to other regions but also for each seasons.



Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

> I thought of something else for your history pages.  I'm curious to know how
> accurate the forecasts are for 24+ hours out, versus the one that is 0+ hours
> out.

This would be very intersting to know. But as Allan has explained comparing
forecasts is not the best way to do this.

The best data that I know of that begins to address your question is from
Bruce L. Gary who has been systematically comparing clock forecasts to
observed weather conditions -- but for just his site.

You may find his report interesting:
http://groups.yahoo.com/group/AstronomyWeather/message/62

-ad

Allan and Attilla,

I agree that some sort of checking the forecasts against actual conditions
would be the most useful, but that sounds pretty difficult to do accurately
on a broad scale (as Allan explained).  My suggestion makes the assumption
that a near-range forecasts are a pretty good prediction of reality, and uses
that to compare how far off the longer-range forecasts are.  This could be
done easily and broadly.  This could tell you something about the stability
of the models by region, and also by time of year.

As an example, imagine piecing together a cloud cover graph for the entire
year from predictions that are as near as possible.  Usually, this will come
from 12-hour chunks, if the clocks are updated every 12 hours.  Now, make a
cloud cover prediction graph, but use forecasts that are 24 hours out to
piece this together.  Plotting these on the same graph would give some visual
indication of how good the 24 hour forecasts are.  Some parts of the year may
track pretty well, but other times would track very poorly.  This could be
reduced to numbers by some sort of comparison scheme that blurs small
variations, but captures the essence of the forecast (perhaps RMS error on a
sliding window?)

Attilla,

If you would be willing to share with me historical data for some location
(mine?), I would be happy to fool around with this idea to see if it has any
merit.

-- Steve Koehler
    steve_koehler@...

> -----Original Message-----
> From: danko(server1) [mailto:danko@...]
> Sent: Friday, July 16, 2004 3:45 AM
> To: AstronomyWeather@yahoogroups.com
> Subject: Re: [AstronomyWeather] history page idea
>
>
> > I thought of something else for your history pages.  I'm curious to know
> how
> > accurate the forecasts are for 24+ hours out, versus the one that is 0+
> hours
> > out.
>
> This would be very intersting to know. But as Allan has explained
> comparing
> forecasts is not the best way to do this.
>
> The best data that I know of that begins to address your question is from
> Bruce L. Gary who has been systematically comparing clock forecasts to
> observed weather conditions -- but for just his site.
>
> You may find his report interesting:
> http://groups.yahoo.com/group/AstronomyWeather/message/62
>
> -ad
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>

> If you would be willing to share with me historical data for some location
> (mine?), I would be happy to fool around with this idea to see if it has any
> merit.

Yes. I'm setup to extract forecast history for particular sites. I'm happy to do
this
for anyone that asks.

The data is a bunch of numbers and isnt pretty. But it can be parsed if your
comfortable
writing scripts.


Steve, remind me which site you use.

-ad

Attilla,

Minneapolis would be close enough.  I view at a few sites a bit south of
town, but Minneapolis would be central in this area.  Thanks.

-- Steve Koehler
    steve_koehler@...

> -----Original Message-----
> From: danko(server1) [mailto:danko@...]
> Sent: Friday, July 16, 2004 3:15 PM
> To: AstronomyWeather@yahoogroups.com
> Subject: Re: [AstronomyWeather] history page idea
>
>
> > If you would be willing to share with me historical data for some
> location
> > (mine?), I would be happy to fool around with this idea to see if it has
> any
> > merit.
>
> Yes. I'm setup to extract forecast history for particular sites. I'm happy
> to do this
> for anyone that asks.
>
> The data is a bunch of numbers and isnt pretty. But it can be parsed if
> your comfortable
> writing scripts.
>
>
> Steve, remind me which site you use.
>
> -ad
>
>
>
>
> Yahoo! Groups Links
>
>
>
>

Some you may be interested to have a look at these images of Alaska-Yukon
forest fires and smoke production! The loop is a bit long to download but
worth the wait!

Allan Rahill

_______________________________________________________________________
An impressively thick pall of smoke from the ongoing Alaska/Yukon fires is
beginning to move over portions of the northcentral US -- the hazy appearance of
the smoke is quite obvious on MODIS "true color" imagery:

http://tinyurl.com/44xow

It's interesting to watch the effect that the thick smoke has on reducing solar
insolation and shutting down the formation of Cu clouds as it moves across
southern Ontario and Manitoba:

http://cimss.ssec.wisc.edu/goes/misc/040716/040716_g12_vis_anim.html

The smoke is almost as effective at inhibiting Cu formation as the cold lakes in
northern MN (note the numerous "Cu shadows" downwind of the lakes, which don't
have ice floating in them because it's July).

--
Scott Bachmeier      University of Wisconsin - Madison / SSEC / CIMSS
scott.bachmeier{at}ssec.wisc.edu http://www.ssec.wisc.edu/~scottb

On Sat, 24 Jul 2004, MA wrote:

> Hi:
>
> Does the seeing forecast take into account fire and smoke?  Because the
forecast for Calgary right now
> at 4AM local time is far from what I am seeing.  The seeing is poor and smoke
is to blame.  The pic
> is the smoke and fire observation for NA.  As you can see the smoke plumes are
massive.
> Thanks.
>
> Mike

Mike,

Worldwide weather numerical models don't take account fire and smoke in
their data base. There is no such operational data base and it is
impossible for now to include this information. So, my transparency
forecast don't take account smoke from forest fire in the forecast.


                                     *               *
                                                         _~~~~~~~
Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

Allan,

That was very interesting.  Thanks!

-- Steve Koehler
    steve_koehler@...

> -----Original Message-----
> From: Allan Rahill [mailto:allan.rahill@...]
> Sent: Saturday, July 24, 2004 7:05 PM
> To: Astro-Weather-Meteo
> Subject: [AstronomyWeather] Forest fire.... smoke!
>
>
> Some you may be interested to have a look at these images of Alaska-Yukon
> forest fires and smoke production! The loop is a bit long to download but
> worth the wait!
>
> Allan Rahill

Starting tonight, July 29th 2004,

The sky transparency forecast is now showing 5 color levels instead of 4.
It will be in harmony with the seeing forecast using the same color
scale, grey to dark blue intervalls. Grey color will correspond to poor
transparency and dark blue for excellent conditions.

Attilla Danko did  modifications for all tonight Clear Sky Clocks
to show this change.

                                     *               *
                                                         _~~~~~~~
Allan Rahill                             *     *      *(     )
                                                    ~~-(__   )
Centre meteorologique Canadien                  * (____ CMC )
Canadian Meteorological Center, Dorval, Canada     (_______)
                                                       ////
ASTRO-METEO-WEATHER                                   '''
http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
Attilla web page
http://www.cleardarksky.com/csk

Allan,

Thanks for the update, and for all the work you do in getting forecasts for
us.

-- Steve Koehler
    steve_koehler@...

> -----Original Message-----
> From: Allan Rahill [mailto:allan.rahill@...]
> Sent: Thursday, July 29, 2004 2:43 AM
> To: Astro-Weather-Meteo
> Subject: [AstronomyWeather] Tranparency forecast...
>
>
> Starting tonight, July 29th 2004,
>
> The sky transparency forecast is now showing 5 color levels instead of 4.
> It will be in harmony with the seeing forecast using the same color
> scale, grey to dark blue intervalls. Grey color will correspond to poor
> transparency and dark blue for excellent conditions.
>
> Attilla Danko did  modifications for all tonight Clear Sky Clocks
> to show this change.
>
>                                     *               *
>                                                         _~~~~~~~
> Allan Rahill                             *     *      *(     )
>                                                    ~~-(__   )
> Centre meteorologique Canadien                  * (____ CMC )
> Canadian Meteorological Center, Dorval, Canada     (_______)
>                                                       ////
> ASTRO-METEO-WEATHER                                   '''
> http://www.cmc.ec.gc.ca/cmc/htmls/mainpage.html      '''
> Attilla web page
> http://www.cleardarksky.com/csk

I used to download raw GOES satellite images from
http://rsd.gsfc.nasa.gov/goeseast/

However, the server has disappeared. Anyone know if this is
temporary or if the server moved?

-ad

I use this for satellite weather images...

   http://www.goes.noaa.gov/
----- Original Message -----
From: "Attilla Danko" <danko@...>
To: <AstronomyWeather@yahoogroups.com>
Sent: Friday, September 24, 2004 5:22 PM
Subject: [AstronomyWeather] GOES download site gone?



I used to download raw GOES satellite images from
http://rsd.gsfc.nasa.gov/goeseast/

However, the server has disappeared. Anyone know if this is
temporary or if the server moved?

-ad





Yahoo! Groups Links

Hi Attilla,

I use this for satellite weather images...

   http://www.goes.noaa.gov/

also here...

   http://www.wrh.noaa.gov/satellite/

Not sure if these are similar to the site you mentioned.

-Florian


----- Original Message -----
From: "Attilla Danko" <danko@...>
To: <AstronomyWeather@yahoogroups.com>
Sent: Friday, September 24, 2004 5:22 PM
Subject: [AstronomyWeather] GOES download site gone?



I used to download raw GOES satellite images from
http://rsd.gsfc.nasa.gov/goeseast/

However, the server has disappeared. Anyone know if this is
temporary or if the server moved?

-ad

Sorry for the double post. I hadn't completed the first one when it somehow got
sent.

-Florian

Florian wrote:
> I use this for satellite weather images...
>
>   http://www.goes.noaa.gov/

Yes. But http://rsd.gsfc.nasa.gov/goeseast/
gave access to the raw IR2 and raw IR4 channels. The raw
data had no map overlays and include data to back map
the pixels to latitude and longitude. I was using that
stuff to compute my own statellite image.

It's access to the raw IR sensor data that I miss.
-ad