[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20030528: McIDAS installation at UFRJ (cont.)



>From:  David Garrana Coelho <address@hidden>
>Organization:  UFRJ
>Keywords:  200305280404.h4S44lLd024244

Hi David,

re: Currently, 5 images sectors centered on 23S, 53W are being created
on brisa (from ADDE access

>       I dont know if 53W was a typo or intended, but I suppose its the
>furthest east you could center the image, am I wrong? If I am, 43W would
>be more appropriate.=) (More comments on this below)

I chose the center point so that the areal extent of the images
at their maximum resolution would not include blank space on the
eastern flank.  After I get you some HOWTOs for bringing up McIDAS
and loading the full images yourself, you will see that the GOES-12
scans I chose go as far east as possible, and so cover your modeling
domain as well as can be hoped for.

0.65 um (visible) - resolution - 1 km:
http://www.unidata.ucar.edu/staff/tom/meteoforum/sectors/goeseast_vis_resolution.gif

0.65 um (visible) - area coverage - 2 km:
http://www.unidata.ucar.edu/staff/tom/meteoforum/sectors/goeseast_vis_extent.gif

3.9 um (short infrared) - area coverage/resolution - 4 km:
http://www.unidata.ucar.edu/staff/tom/meteoforum/sectors/goeseast_39_resolution.gif

6.5 um (water vapor) - area coverage/resolution - 4 km:
http://www.unidata.ucar.edu/staff/tom/meteoforum/sectors/goeseast_wv_resolution.gif

10.7 um (thermal infrared) - area coverage/resolution - 4 km:
http://www.unidata.ucar.edu/staff/tom/meteoforum/sectors/goeseast_ir_resolution.gif

13.3 um (long infrared/CO2) - area coverage/resolution - 4 km:
http://www.unidata.ucar.edu/staff/tom/meteoforum/sectors/goeseast_co2_resolution.gif

>       The links above aren't working,

Oops!  I was frantically typing the message yesterday afternoon while
trying to get out of the office as soon as possible.  The 'gifs'
directory is where I normally put images for remote users to look at.
I put the new sectors in the meteoforum/sectors directory.  The URLs as
listed above should now be correct.  I verified the display of the CO2
image on my home machine (through a 24K dial-up, ugh!).  Sorry for the
typo!

>but I tried mcidasx on brisa from
>home (bashing my head against the monitor a few times trying to discover
>the right commands to display the imagery...lol), and it appears to be
>cutting exactly Rio de Janeiro state...:=/.

When there is no southern sector from GOES, you will end up with either
of two situations:

- Northern Hemisphere scan which extends just south of Belem

- Northern Hemisphere extended scan which extends to RJ

>The way it is, it display a
>large portion of central south america, including the Andes and a part of
>Pacific Coast. Rio de Janeiro is exactly off the eastern frontier of
>image.

This may be caused by your McIDAS session having small frame sizes (the
default is 480x640) and you loading the image in the default manner
which puts the upper left corner of the image in the upper left corner
of the display frame.

Try the following:

1) start McIDAS with:

mcidas -config

2) from the GUI that pops up, set the frame sizes to 700x950.  This will
   be done by dragging the slider for the number of Lines and Elements
   from left to right

3) click on the save options radio button near the bottom of the GUI

4) start your McIDAS session

After you have a session up with the MCGUI interface, click on the
Display option at the top of the MCGUI; select imagery, and then select
the RIO dataset.  If the RIO option doesn't show up (this is what I was
trying to setup yesterday afternoon, and it may not be finished), then
select the GOESEAST dataset.  GOESEAST will allow you to display
sectors from the images from which the RIO sectors are being created.
This is the the point I wanted to get you to anyway, so let's go for it
now.

When you are able to get the GUI selector up for the RIO and/or
GOESEAST datasets, you will be able to select the band, center Lat and
Lon for image loads of the latest image, and do a 10 frame loop.
At this point, you will be able to see the full extent of the sectors
and/or full images from GOESEAST.

>Don't know if it helps you, but to give a visual cue, Rio is
>exactly on the single portion of Brazilian coast which is oriented mostly
>east-west, with 2 small bays along it.

The good news is that McIDAS knows where RIO is and so can move the cursor
to it.  The disconnect we are seeing at the moment is undoubtedly
being caused from you not being able to see the full extent of the
sectors being created, or your having looked at an image for which there
was no southern GOES-12 sector to composite.

>I also tried to open the files
>using the "xview" command, and it showed a missing eastern part of image
>misplaced at the west side of image (which mcidas was unable to show). It
>looks like a cut/paste operation made out of order (my guess).

I will check on this, but I don't think that this is the case.  I would
bet that xview is not correctly handling the imagery which is in
McIDAS AREA format.

re: let's put off the generation of full South American sectors until
we decide on the Rio sectors

>Okidoki

I was thinking about this last night at home, and I think that you have
plenty of bandwidth to grab the full South American portion of the
larger images for all of the IR channels.  The full images are only
4.4 MB in size, so this should not be hard.  With the full South
American view, you can create any sector you want using McIDAS.  The
visible image, on the other hand, is pretty large, 69 MB, so creation
of image sectors from it at full resolution is probably the way to go.
You could, of course, create a VIS sector of all of South America
at the same resolution (4 km) as the IR sectors and also create high
resolution sectors for wherever you want.  The script I setup
for the transfer is easily modified to create any sector that you
can imagine (to the extent of the original images, of course).

>       As I mentioned above, did my best to view the imagery. Nothing
>very fancy as animations, of course, due to bandwidth limitation here on
>my home.

I have the same problem!  I used to live in a place that had a
spectacular DSL connection: 1.2 Mbps down and 320 Kbps up.  That
was heaven!

By the way, since you are running Linux at home, you could load
and run McIDAS from there.  That way, the only thing going over
the network connection is the data.  If your connection is as slow
as mine (24 Kbps), then loading images will be very painful.  Displaying
things like observational data is very useful, however.

>I logged as mcidas and started mcidasx. After a lil head bashing,
>I figured out DSINFO and IMGDISP, and saw the images. I tried MAP BRAZIL
>and it looked somewhat distorted, maybe because of projection options, not
>sure.

To draw am map on top of an image, you use 'MAP SAT' or commands
like 'MAP FILE=OUTLHRES'.  These tell McIDAS to use the navigation
information already contained in the image.  A 'MAP BRAZIL' will
draw a map of Brazil in a conic projection regardless of the
image that is displayed.  This is why the map did not look correct
on the image you displayed.

>Amidst my (blunt) learning process, I tried the GUI (without success
>in loading/displaying any data on it) and executed mcinet2000.sh as root
>(as told on MCIDAS Installation page), hope this was something you would
>ask for me to do in the future, if not, my apologies.

Actually yes, I was going to ask you to run mcinet2002.sh today so that
I could load the images being created on brisa back on my machine at
work.  This was the final step in providing me with an end-to-end
verification that the sectors being created were correct.  Thanks!  One
thing you will need to do, however, is edit /etc/services as 'root' and
comment the first entries for port 500 (isakmp).  McIDAS incorrectly
sets itself up to use port 500 for uncompressed data transfers.  It
uses port 503 for compressed data transfers, which is also incorrect.
-- I say incorrect since they did not register the use of these ports
with Iana --  In the future (perhaps the next distribution), they will
use port 112, which has been registered with Iana for years (we did
that back in the early 1990s).

>       I am planning to try the observational data overlay and map
>options tomorrow, locally.

The first step is to get the MCGUI up and running correctly.  The next
step is to explore the options available.

>       As a small side note (and something of heads up too) I was
>thinking about submitting all this work you having as a paper in the next
>Brazilian Congress of Meteorology in the next year. It would serve as an
>open invitation to fellow institutions here in Brazil (and South America)
>to use IDD, and as a small (very small compared to all the trouble you
>having, may I add) "payment" for all the work you having with us. I was
>thinking of something like a small report of all the work done on brisa
>(and the future nameless dedicated server), history and the capabilities
>of the tool, and "testimonies" of current users (UFRJ, Buenos Aires and
>Belem). Let me know if you like the idea...:=)

I think that this is a GREAT idea!  To my mind, this is the only way
that other Brazilian and South American universities will be come
intimately familiar with the potential benefit of participating in the
Unidata community.  It has been my goal for almost 15 years to foster
the growth of our scientific community in the international arena.  I
am absolutely delighted that the movement of data to your institution
is working as well as it is.  A short year and a half ago, doing what
we are doing now was not possible, and now it is.  You have got to love
this :-)

Hopefully, a presentation at the Brazilian Congress of Meteorology
would get a number of institutions wanting to participate in
IDD-Brazil.  This would mean, of course, that the UFRJ IDD node and
internet connection would become very busy :-)

>       Thanks once more,

No worries.  I have to close here so I can eat some breakfast and then
head into work.

Cheers,

Tom
--
+-----------------------------------------------------------------------------+
* Tom Yoksas                                             UCAR Unidata Program *
* (303) 497-8642 (last resort)                                  P.O. Box 3000 *
* address@hidden                                   Boulder, CO 80307 *
* Unidata WWW Service                             http://www.unidata.ucar.edu/*
+-----------------------------------------------------------------------------+