[Met_help] [rt.rap.ucar.edu #95781] History for Forecast objects unidentified

John Halley Gotway via RT met_help at ucar.edu
Tue Jun 30 14:27:10 MDT 2020


----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

All,

We are trying to compare two 2D netCDF files and it seems like mode doesn't identify any forecast cluster objects.

Both files contain the same 2d variable.

[cid:ae48cceb-f6f1-4ce3-b858-de10196cf579]

Any recommendations?

Thank You,


Evan Jaramillo
Software Developer

Integration Innovation, Inc. (i3)
689 Discovery Drive, Floor 2
Huntsville, AL 35806
(o) 256-562-0669
evan.jaramillo at i3-corps.com
evan.r.jaramillo.ctr at mail.mil

[cid:e4013310-3931-45e5-9795-9d3522fb53b7]

Confidentiality Notice - The information contained in this e-mail and any attachments to it may be legally privileged and include confidential information.  If you are not the intended recipient, be aware that any disclosure, distribution or copying of this e-mail or its attachments is prohibited.  If you have received this e-mail in error, please notify the sender immediately of that fact by return e-mail and permanently delete the e-mail and any attachments to it.


----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: Forecast objects unidentified
From: John Halley Gotway
Time: Tue Jun 30 12:32:38 2020

Evan,

Judging by clues in the single image you sent, I have a guess as to
what
might be going on. I notice that the forecast field contains blobs of
data
values surrounded by areas of white whereas the observation field
contains
blobs of data values surrounded by areas of gray.

I'm guessing that you're running MODE using the default color table
and
white means a value of 0 while dark grey means missing data values. So
while both the forecast and observation fields contain the "same
variable"
those data fields are defined very differently. Surrounding values of
0
create very different objects than missing data values.

The MODE logic consists of multiple steps:
(1) Apply a convolution filter to smooth the forecast and observation
data
(controlled by conv_radius).
(2) Threshold the smoothed data to create objects (controlled by
conv_thresh).
(3) Applying matching/merging logic to compare those objects.

In step (1), the surrounding 0's in the forecast field are included in
the
smoothing step and are dragging the smoothed values down below the
conv_thresh threshold. In the observation field, the gray missing data
values are just being ignored and are therefore not dragging down the
smoothed values. So you get 0 forecast objects and many observation
objects.

There's several ways you could resolve this, and the choice is up to
you.

(1) Disable smoothing entirely by setting "conv_radius = 0".
(2) Replace all forecast values of 0 with bad data values by putting
the
following in the forecast dictionary:
   censor_thresh = [ ==0 ]; censor_val = [ -9999 ];
   When you think about it, that has the same impact as option (1).
(3) Replace all observation values of bad data with 0 by putting the
following in the observation dictionary:
   censor_thresh = [ ==-9999 ]; censor_val = [ 0 ];

Of these, I'd recommend option 3. However you'll find that you still
get 0
objects, but now in both fields. From there though, you can adjust the
amount of smoothing (conv_radius) and object definition threshold
(conv_thresh) to create objects that match the scale of the features
you're
trying to study.

Hope that helps clarify.

Thanks,
John

On Tue, Jun 30, 2020 at 9:22 AM Evan Jaramillo via RT
<met_help at ucar.edu>
wrote:

>
> Tue Jun 30 09:21:39 2020: Request 95781 was acted upon.
> Transaction: Ticket created by evan.jaramillo at i3-corps.com
>        Queue: met_help
>      Subject: Forecast objects unidentified
>        Owner: Nobody
>   Requestors: evan.jaramillo at i3-corps.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95781 >
>
>
> All,
>
> We are trying to compare two 2D netCDF files and it seems like mode
> doesn't identify any forecast cluster objects.
>
> Both files contain the same 2d variable.
>
> [cid:ae48cceb-f6f1-4ce3-b858-de10196cf579]
>
> Any recommendations?
>
> Thank You,
>
>
> Evan Jaramillo
> Software Developer
>
> Integration Innovation, Inc. (i3)
> 689 Discovery Drive, Floor 2
> Huntsville, AL 35806
> (o) 256-562-0669
> evan.jaramillo at i3-corps.com
> evan.r.jaramillo.ctr at mail.mil
>
> [cid:e4013310-3931-45e5-9795-9d3522fb53b7]
>
> Confidentiality Notice - The information contained in this e-mail
and any
> attachments to it may be legally privileged and include confidential
> information.  If you are not the intended recipient, be aware that
any
> disclosure, distribution or copying of this e-mail or its
attachments is
> prohibited.  If you have received this e-mail in error, please
notify the
> sender immediately of that fact by return e-mail and permanently
delete the
> e-mail and any attachments to it.
>
>

------------------------------------------------
Subject: Forecast objects unidentified
From: Evan Jaramillo
Time: Tue Jun 30 12:55:37 2020

John,

Your recommendations worked!

Thank you for your help!


Evan Jaramillo
Software Developer

Integration Innovation, Inc. (i3)
689 Discovery Drive, Floor 2
Huntsville, AL 35806
(o) 256-562-0669
evan.jaramillo at i3-corps.com
evan.r.jaramillo.ctr at mail.mil

[cid:9c6e2514-9409-467c-9e53-04c73d2948f2]

Confidentiality Notice - The information contained in this e-mail and
any attachments to it may be legally privileged and include
confidential information.  If you are not the intended recipient, be
aware that any disclosure, distribution or copying of this e-mail or
its attachments is prohibited.  If you have received this e-mail in
error, please notify the sender immediately of that fact by return e-
mail and permanently delete the e-mail and any attachments to it.

________________________________
From: John Halley Gotway via RT <met_help at ucar.edu>
Sent: Tuesday, June 30, 2020 1:32 PM
To: Evan Jaramillo <evan.jaramillo at i3-corps.com>
Cc: Mariana Scott <mariana.scott at i3-corps.com>
Subject: Re: [rt.rap.ucar.edu #95781] Forecast objects unidentified

WARNING: The sender of this email could not be validated and may not
match the person in the "From" field.

CAUTION: This email originated from outside of the organization. Do
not click links or open attachments unless you recognize the sender
and know the content is safe.


Evan,

Judging by clues in the single image you sent, I have a guess as to
what
might be going on. I notice that the forecast field contains blobs of
data
values surrounded by areas of white whereas the observation field
contains
blobs of data values surrounded by areas of gray.

I'm guessing that you're running MODE using the default color table
and
white means a value of 0 while dark grey means missing data values. So
while both the forecast and observation fields contain the "same
variable"
those data fields are defined very differently. Surrounding values of
0
create very different objects than missing data values.

The MODE logic consists of multiple steps:
(1) Apply a convolution filter to smooth the forecast and observation
data
(controlled by conv_radius).
(2) Threshold the smoothed data to create objects (controlled by
conv_thresh).
(3) Applying matching/merging logic to compare those objects.

In step (1), the surrounding 0's in the forecast field are included in
the
smoothing step and are dragging the smoothed values down below the
conv_thresh threshold. In the observation field, the gray missing data
values are just being ignored and are therefore not dragging down the
smoothed values. So you get 0 forecast objects and many observation
objects.

There's several ways you could resolve this, and the choice is up to
you.

(1) Disable smoothing entirely by setting "conv_radius = 0".
(2) Replace all forecast values of 0 with bad data values by putting
the
following in the forecast dictionary:
   censor_thresh = [ ==0 ]; censor_val = [ -9999 ];
   When you think about it, that has the same impact as option (1).
(3) Replace all observation values of bad data with 0 by putting the
following in the observation dictionary:
   censor_thresh = [ ==-9999 ]; censor_val = [ 0 ];

Of these, I'd recommend option 3. However you'll find that you still
get 0
objects, but now in both fields. From there though, you can adjust the
amount of smoothing (conv_radius) and object definition threshold
(conv_thresh) to create objects that match the scale of the features
you're
trying to study.

Hope that helps clarify.

Thanks,
John

On Tue, Jun 30, 2020 at 9:22 AM Evan Jaramillo via RT
<met_help at ucar.edu>
wrote:

>
> Tue Jun 30 09:21:39 2020: Request 95781 was acted upon.
> Transaction: Ticket created by evan.jaramillo at i3-corps.com
>        Queue: met_help
>      Subject: Forecast objects unidentified
>        Owner: Nobody
>   Requestors: evan.jaramillo at i3-corps.com
>       Status: new
>  Ticket <URL:
https://usg02.safelinks.protection.office365.us/?url=https%3A%2F%2Frt.rap.ucar.edu%2Frt%2FTicket%2FDisplay.html%3Fid%3D95781&data=02%7C01%7C%7C3bc40e64f21741a80c1708d81d23f8d7%7C25c8b235e83442a5bae3f4362eaff8bf%7C0%7C0%7C637291387620313578&sdata=Gm6WdWZVPLAN0fidAibeABOem%2FdrnG6fp4mJb7rPTkA%3D&reserved=0
>
>
>
> All,
>
> We are trying to compare two 2D netCDF files and it seems like mode
> doesn't identify any forecast cluster objects.
>
> Both files contain the same 2d variable.
>
> [cid:ae48cceb-f6f1-4ce3-b858-de10196cf579]
>
> Any recommendations?
>
> Thank You,
>
>
> Evan Jaramillo
> Software Developer
>
> Integration Innovation, Inc. (i3)
> 689 Discovery Drive, Floor 2
> Huntsville, AL 35806
> (o) 256-562-0669
> evan.jaramillo at i3-corps.com
> evan.r.jaramillo.ctr at mail.mil
>
> [cid:e4013310-3931-45e5-9795-9d3522fb53b7]
>
> Confidentiality Notice - The information contained in this e-mail
and any
> attachments to it may be legally privileged and include confidential
> information.  If you are not the intended recipient, be aware that
any
> disclosure, distribution or copying of this e-mail or its
attachments is
> prohibited.  If you have received this e-mail in error, please
notify the
> sender immediately of that fact by return e-mail and permanently
delete the
> e-mail and any attachments to it.
>
>


------------------------------------------------


More information about the Met_help mailing list