20030109: McIDAS ADDE server for MSG imagery

>From:  "=?ISO-8859-1?Q?Marianne=20K=F6nig?=" <address@hidden>
>Organization:  EUMETSAT
>Keywords:  200301090846.h098k4t14949 McIDAS EUMETSAT MSG ADDE

Hi Marianne,

>This is probably one of my stupid mcidas questions again, but I am so
>stuck on this issue that I need some expert's advice on that:

I will help if I can.

>what I am trying to do is write an ADDE Server for our in-house format
>of MSG data (yes, we have first images!!!!!).

Wonderful!

>All that is very fine,
>and I can do an IMGLIST and IMGDISP using some sort of ..aget and
>..adir programs. How do I manage to get something like IMGPROBE
>working?

IMPROBE justs makes an AGET call, so if you have IMGDISP working,
IMGPROBE should work with very little/no effort.

>The problem I see is the filename: My servers sort of
>construct a filename from the user input of day, time and band, but of
>course IMGPROBE does not know any of that.

IMGPROBE will know all of these things _if_ you are probing an image
that you have displayed in a frame.  Have you tried that, or are you
trying to run IMGPROBE on a dataset member directly?

>For normal AREA files I see
>that IMGPROBE reads in the first AREA directory entry which is the
>position within the dataset - and constructs a filename from that.

For datasets that are not composed of AREA files or ones where the
data files do not follow AREA file naming conventions, the server
will need to organize (sort) all of the images in the dataset and
then extract information from the one that matches your request.
This request might be specified by day/time/band, or it may be
specified by dataset position number and band.

>I
>cannot see a solution for this, but I am sure that must exist as many
>people have written servers which should tackle the same problem.

The servers I wrote for radar images (the NEXRAD Level III images and
composites in formats other than AREA) do the needed sorting work while
processing image requests.  Yours will need to do the same thing.

>The
>problem probably is the same with any other of the IMG.. commands, but
>I'd be happy if I could at least get IMGPROBE to workl.

I worked my way through the various IMG* commands when getting my
servers to work.  

Thank you so much for your help,

I think I understand your problem, but please let me restate it so you
can verify that I understand:

o you have your own ADIR and AGET servers for MSG images, and those servers
  are relying on day, time, and band specifications in the invocation
  command line in order to construct the name of a file that contains
  the data being requested

o IMGPROBE when run directly from the command line specifying the
  keywords DATaset= and BANd= does not specify the information
  needed/expected by your servers

If I have this right, the "simple" way of making things work is to
display the image using IMGDISP in a frame, and then run IMGPROBE
without specifying either of the DATaset= or BANd= keywords.  IMGPROBE
will then pickup the day, time, band, etc. from the frame directory and
include that in the request to the server.  This should be all of the
information that is needed by your servers to construct the file name
for the data to read.  The downside to this approach, of course, is
that you will sample at the resolution of the displayed image.

The other thing that I would recommend -- and you are not going to like
this one -- is to rewrite your servers to so that they have knowledge
of the name space for the files that compose the dataset.  They would
then be able to sort the images into temporal order and extract
information out of elements by dataset position number.  This requires
the server to do a log more work, but it makes it much more flexible.
The good news is that the NEXRAD Level III servers I wrote are part of
core McIDAS, so they can be used as an example of how to write the new
server.  I actually have three different servers that work for image
data that is not in McIDAS AREA format.  All three are based on the
exact same routines contained in the NEXRAD Level III servers.

I can provide advice with a rewrite if you choose to pursue that route.
Of course, the work would be better done in Darmstadt or someplace
warm like the Canary Isands ;-)

>and many regards from the now extremely cold Germany,

While it is cold in Germany, it has been extremely warm and dry here
in Colorado.  Three of us went skiing yesterday, and it was so warm
I was removing my gloves while riding the chair lifts even at
12,000 ft (3650 m)!  I am hoping that the temperature will drop and
it will start snowing!!

Tom

>From address@hidden Thu Jan  9 09:34:10 2003
>Subject: Re: 20030109: McIDAS ADDE server for MSG imagery
Tom,

Thank you for the advice.  Of course I tried the IMGPROBE, and that
just does not work (and yes, I think you see my problem):  IMGPROBE
(just the naked command like that), used on an IMGDISP displayed frame
of MSG, comes back with

"You must specify a DAY"    (which probably comes from my server???)
"ival for mcadir =-1000"
"failed to set up directory server"

Would you have an example of a simple image server plus a corresponding
dataset that I could pick up from your ftp site which work with
IMGPROBE (without sorting)? I suspect there is something wrong with my
server ...

Again, thank you,

Marianne

>From address@hidden Fri Jan 10 07:39:17 2003
>Subject: success ....

Hi Tom,

It's me again, and I found my mistake: I did not set the first area
directory element within my servers to -1 to correctly populate the
frame directory  -  and of course it took me about 4 hrs to go through
all of IMGPROBE to find this out. But I am so happy now that this
works. I am completely full of myself ....

Have a wonderful weekend,

Marianne

>From address@hidden Mon Jan 13 01:35:17 2003

Tom,

The day is pretty far where I create new servers that are even more
elaborate than yours and you are on this beach lounge ...  Mcidas still
amazes me - the truth is I still have no clue why all this works

We have horrible snow here today (unusual for Darmstadt), and it is
supposed to turn into ice rain later this afternoon. To be on the safe
side I took the train.

Thank you for your help,

Marianne