McIDAS-V displays missing values in GRIB2 files as the minimum value; they should not be displayed

Robb Kambic


	Posted on: 20100222.1103 MST 
Hiya Jess,

The missing value was set to the reference value instead of the Standard Grib2 
missing value of NaN. The reference value was 1 which caused your problem of no 
distinction between data and the missing value. A fix was made to use a missing 
value of NaN so the problem will go away. The fix will be available in the next 
netCDF-java snapshot release, at this time I don't know when it will be 
available. Thanks for your patience.

RObb...
  [Edit Ticket Post]

E-mail: address@hidden  IP: 128.117.140.185
Don Murray


	Posted on: 20100222.0757 MST 
Hi Jess-

I'm going to transfer this over to the grib decoder support staff. You should 
be hearing from him shortly. Thanks for the sample files.

Don

> A few of our users came across a problem with missing values in GRIB2 files. 
> The found that McIDAS-V (and I checked IDV) displays missing values as the 
> minimum values instead of not displaying them.
>
> If you'd like to see the two files:
> ftp://ftp.ssec.wisc.edu/pub/incoming/ci_inst.sampleim_only.gb2
> ftp://ftp.ssec.wisc.edu/pub/incoming/ctc_inst.sampleim_only.gb2
>
> There is plenty of additional information below, please let us know if you have 
> any questions,
>
> Thanks,
>
> ~Jess
>
> ----==== Inquiry ====----
> 723
>
> ----==== Summary ====----
> McIDAS-V displays missing values in GRIB2 files as the minimum value; they 
> should not be displayed at all.
>
> ----==== Request ====----
> 2010-02-18 - Becky Schaffer
> Hello Becky,
>
> I am not sure I understand the distinction. You might have to look closer at the contour maps to find the error, but it 
> is evident there as well. In the CI GRIB2 file, if we have a value of "1" surrounded by missing data, and 
> there is no value higher than "1" bounded by missing data, then there is no contour drawn because the missing 
> data is also interpreted as "1" and thus there is no gradient on which to draw contours. We see no 
> "1" contours, which we should. A similar problem exists with the CTC GRIB2 file but more difficult to 
> identify because the minimum value is the most anomalous cooling rate. I hope this explanation is clear, but if you 
> would like to come down and visit the CAVE, I (or Kaba) can add a visual explanation.
>
> Regards,
>
> Jordan Gerth
>
> 2010-02-17 - Jessica Staude
> Tried in IDV 2.8u1. ctc_inst.sampleim_only.gb2 displayed as a color shaded plan 
> view resulted in the min value being displayed as -33.5, which is defined as a 
> missing value.
>
> 2010-02-17 - Becky Schaffer
> Thanks, Jordan. That makes a lot more sense. In Joleen's original email, she 
> said that the missing data was being displayed as the minimum value, which I 
> interpreted as meaning that the minimum value was always being displayed on the 
> screen wherever there was missing data. From what I've seen, contours and 
> color-filled contours seem to display correctly, whereas the probes and the 
> color-shaded plan view seem to show the problem. I'll write up an inquiry to 
> look into the probe and the color-shaded plan view.
>
> - Becky
>
>
>
> 2010-02-17 - Becky Schaffer
> Jordan Gerth wrote:
> Hello,
>
> At least in the CAVE, the center mouse button sampling feature reveals all
> missing (NaN) values (black areas in your included map) as the lowest value in
> the data range (-33.5, for example). This also impacts some of the shading
> options I believe. It is difficult to discern whether the included image is
> correct or incorrect.
>
> I am using wgrib2 to produce the GRIB2 message and determine the value of the
> data points. The McIDAS-X statistics appear to be representative of what output
> from wgrib2 indicates, though, so I think it is solely a display issue with V.
>
> If you would like me to demonstrate in the CAVE, I would be happy to.
>
> Jordan
>
> 2010-02-17 - Becky Schaffer
> What data values do you expect to see with these files? I tried them in
> McIDAS-X (version 2009.2), and the results seem to be the same as McIDAS-V
> (version beta5). The CI file has data ranging from 1 to 3, and the CTC file
> has data ranging from -33.5 to -4.1. Since -X and -V read grib2 files using
> different methods, seeing the same results seems like it's a good sign.
>
> ...
>
> 2010-02-13 - Barry Roth
> Note: The two test GRIB2 messages that Joleen provided in the request below are
> available at sweetpea:/home/barryr/helpdesk/2010_feb12_joleen:
> -rwxr-x--- 1 barryr mug 92787 Feb 12 21:36 ci_inst.sampleim_only.gb2*
> -rwxr-x--- 1 barryr mug 92810 Feb 12 21:36 ctc_inst.sampleim_only.gb2*
>
> 2010-02-13 - Barry Roth
> Joleen wrote 12feb10 2:36pm:
> ----------------------------
> A NCEP-standard GRIB2 message generated in-house was loaded with some data
> values missing. McIDAS-V displayed these as the minimum value of the values
> containing data. This is incorrect. Missing values should be displayed as
> having no value at all.
>
> Please see ftp://ftp.ssec.wisc.edu/pub/ssec/jordang/gb2 for test GRIB2 messages
> which contain this problem.
>
> Please keep Wayne Feltz, Jordan Gerth, Kaba Bah, Tim Schmit, Kris Bedka, Justin
> Sieglaff, and Lee Cronce in the loop about the status of this support request.
> This request was issued from the CAVE.
>
>
> ################################################################################
>
> http://dcdbs.ssec.wisc.edu/inquiry-v/index.php?inquiry=723
  [Edit Ticket Post]

E-mail: address@hidden  IP: 128.117.140.183
MUG Help Desk


        Posted on: 20100219.1358 MST

Hello!

A few of our users came across a problem with missing values in GRIB2 files. 
The found that McIDAS-V (and I checked IDV) displays missing values as the 
minimum values instead of not displaying them.

If you'd like to see the two files:
ftp://ftp.ssec.wisc.edu/pub/incoming/ci_inst.sampleim_only.gb2
ftp://ftp.ssec.wisc.edu/pub/incoming/ctc_inst.sampleim_only.gb2

There is plenty of additional information below, please let us know if you have 
any questions,

Thanks,

~Jess

----==== Inquiry ====----
723

----==== Summary ====----
McIDAS-V displays missing values in GRIB2 files as the minimum value; they 
should not be displayed at all.

----==== Request ====----
2010-02-18 - Becky Schaffer
Hello Becky,

I am not sure I understand the distinction. You might have to look closer at the contour maps to find the error, but it 
is evident there as well. In the CI GRIB2 file, if we have a value of "1" surrounded by missing data, and 
there is no value higher than "1" bounded by missing data, then there is no contour drawn because the missing 
data is also interpreted as "1" and thus there is no gradient on which to draw contours. We see no 
"1" contours, which we should. A similar problem exists with the CTC GRIB2 file but more difficult to 
identify because the minimum value is the most anomalous cooling rate. I hope this explanation is clear, but if you 
would like to come down and visit the CAVE, I (or Kaba) can add a visual explanation.

Regards,

Jordan Gerth

2010-02-17 - Jessica Staude
Tried in IDV 2.8u1. ctc_inst.sampleim_only.gb2 displayed as a color shaded plan 
view resulted in the min value being displayed as -33.5, which is defined as a 
missing value.

2010-02-17 - Becky Schaffer
Thanks, Jordan. That makes a lot more sense. In Joleen's original email, she 
said that the missing data was being displayed as the minimum value, which I 
interpreted as meaning that the minimum value was always being displayed on the 
screen wherever there was missing data. From what I've seen, contours and 
color-filled contours seem to display correctly, whereas the probes and the 
color-shaded plan view seem to show the problem. I'll write up an inquiry to 
look into the probe and the color-shaded plan view.

- Becky



2010-02-17 - Becky Schaffer
Jordan Gerth wrote:
Hello,

At least in the CAVE, the center mouse button sampling feature reveals all
missing (NaN) values (black areas in your included map) as the lowest value in
the data range (-33.5, for example). This also impacts some of the shading
options I believe. It is difficult to discern whether the included image is
correct or incorrect.

I am using wgrib2 to produce the GRIB2 message and determine the value of the
data points. The McIDAS-X statistics appear to be representative of what output
from wgrib2 indicates, though, so I think it is solely a display issue with V.

If you would like me to demonstrate in the CAVE, I would be happy to.

Jordan

2010-02-17 - Becky Schaffer
What data values do you expect to see with these files? I tried them in
McIDAS-X (version 2009.2), and the results seem to be the same as McIDAS-V
(version beta5). The CI file has data ranging from 1 to 3, and the CTC file
has data ranging from -33.5 to -4.1. Since -X and -V read grib2 files using
different methods, seeing the same results seems like it's a good sign.

...

2010-02-13 - Barry Roth
Note: The two test GRIB2 messages that Joleen provided in the request below are
available at sweetpea:/home/barryr/helpdesk/2010_feb12_joleen:
-rwxr-x--- 1 barryr mug 92787 Feb 12 21:36 ci_inst.sampleim_only.gb2*
-rwxr-x--- 1 barryr mug 92810 Feb 12 21:36 ctc_inst.sampleim_only.gb2*

2010-02-13 - Barry Roth
Joleen wrote 12feb10 2:36pm:
----------------------------
A NCEP-standard GRIB2 message generated in-house was loaded with some data
values missing. McIDAS-V displayed these as the minimum value of the values
containing data. This is incorrect. Missing values should be displayed as
having no value at all.

Please see ftp://ftp.ssec.wisc.edu/pub/ssec/jordang/gb2 for test GRIB2 messages
which contain this problem.

Please keep Wayne Feltz, Jordan Gerth, Kaba Bah, Tim Schmit, Kris Bedka, Justin
Sieglaff, and Lee Cronce in the loop about the status of this support request.
This request was issued from the CAVE.


################################################################################

http://dcdbs.ssec.wisc.edu/inquiry-v/index.php?inquiry=723

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================