[Met_help] [rt.rap.ucar.edu #47360] History for Question on Interpolation

RAL HelpDesk {for John Halley Gotway} met_help at ucar.edu
Fri Jun 10 15:58:33 MDT 2011


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



MET Team,

A quick question on verification at certain levels.  As you know, sensors
at point sites are not always placed at 10-meters (wind) and 2-meters
(temperature) - they are often times placed above and below these points.
Given that the user provides observation information (site elevation AGL
and observation elevation AGL) before running ASCII2NC, does the MET
package do the interpolation (for the observations) to 10-m (wind) and 2-m
(temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast vs. obs?

Thanks.

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

Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: Paul Oldenburg
Time: Tue Jun 07 12:39:56 2011

James,

If your observations are of type ADPSFC or SFCSHP, no interpolation is
done, since these are defined as "surface"
observations.  If, however, your observations are not of that type,
MET will perform interpolation of model data to
whatever the elevation of the observation is.  For more information,
see section 4.2.1 of the MET User's Guide.

If you want interpolation to be performed, I think you just need to
set the observation type to something besides ADPSFC
or SFCSHP.  If you are using ascii2nc, for example, you can specify a
Message_Type in the first column of your input.

Paul


On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>
> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
> Transaction: Ticket created by jpcipria at us.ibm.com
>        Queue: met_help
>      Subject: Question on Interpolation
>        Owner: Nobody
>   Requestors: jpcipria at us.ibm.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
>
>
> MET Team,
>
> A quick question on verification at certain levels.  As you know,
sensors
> at point sites are not always placed at 10-meters (wind) and 2-
meters
> (temperature) - they are often times placed above and below these
points.
> Given that the user provides observation information (site elevation
AGL
> and observation elevation AGL) before running ASCII2NC, does the MET
> package do the interpolation (for the observations) to 10-m (wind)
and 2-m
> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast vs.
obs?
>
> Thanks.


------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Tue Jun 07 13:44:54 2011


Paul,

Thanks for the timely reply.  A bit of clarification:

(1) When using ASCII2NC, I'm aware that there must be 10 columns in
each
text file - two of the columns are station elevation (msl) and
observation
(msl).  Let's use temperature as an example here and say the station
was
located 60 meters above sea level...if the observation was truly a 2-
meter
temperature, then the difference between station elevation (60 meters)
and
observation elevation would be 2-meters, thus making the observation
elevation 62 meters...correct?  If the difference was > or < 2 meters,
then
vertical interpolation would be done (forecasts to observation values)
for
2-meter temperature (assuming we choose a message type like "ADPUPA"
rather
than "ADPSFC" or "SFCSHP").  Same would hold true for any other
variable
(say 10-meter wind speed)....correct?

Thanks,



From:	"RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/07/2011 02:40 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

If your observations are of type ADPSFC or SFCSHP, no interpolation is
done, since these are defined as "surface"
observations.  If, however, your observations are not of that type,
MET
will perform interpolation of model data to
whatever the elevation of the observation is.  For more information,
see
section 4.2.1 of the MET User's Guide.

If you want interpolation to be performed, I think you just need to
set the
observation type to something besides ADPSFC
or SFCSHP.  If you are using ascii2nc, for example, you can specify a
Message_Type in the first column of your input.

Paul


On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>
> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
> Transaction: Ticket created by jpcipria at us.ibm.com
>        Queue: met_help
>      Subject: Question on Interpolation
>        Owner: Nobody
>   Requestors: jpcipria at us.ibm.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
>
>
> MET Team,
>
> A quick question on verification at certain levels.  As you know,
sensors
> at point sites are not always placed at 10-meters (wind) and 2-
meters
> (temperature) - they are often times placed above and below these
points.
> Given that the user provides observation information (site elevation
AGL
> and observation elevation AGL) before running ASCII2NC, does the MET
> package do the interpolation (for the observations) to 10-m (wind)
and
2-m
> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast vs.
obs?
>
> Thanks.



------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: Paul Oldenburg
Time: Tue Jun 07 14:05:55 2011

James,

A surface observation for temperature is 2m above the surface, which
has an absolute elevation of 2m + sea level
elevation of the observation station location.  This is obviously
terrain dependent.  If you are using the GRIB output
of WPP as your model data, for example, and it contains surface
temperature, then it will have a 2m TMP forecast value.
 This will not require interpolation to match a "surface" (i.e. ADPSFC
or SFCSHP) observation in point_stat.

On the other hand, if you do not have a "surface" observation, then
the model data will be interpolated using the two
adjacent model levels to the elevation of the observation.  This logic
holds for all forecast fields, regardless of
their defined "surface" elevation.

Unfortunately, the individual most qualified to discuss this matter
with you is on vacation until Thursday.  I will talk
about this issue with him when he returns.  I hope I have provided
some clarification.

Paul


On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> Paul,
>
> Thanks for the timely reply.  A bit of clarification:
>
> (1) When using ASCII2NC, I'm aware that there must be 10 columns in
each
> text file - two of the columns are station elevation (msl) and
observation
> (msl).  Let's use temperature as an example here and say the station
was
> located 60 meters above sea level...if the observation was truly a
2-meter
> temperature, then the difference between station elevation (60
meters) and
> observation elevation would be 2-meters, thus making the observation
> elevation 62 meters...correct?  If the difference was > or < 2
meters, then
> vertical interpolation would be done (forecasts to observation
values) for
> 2-meter temperature (assuming we choose a message type like "ADPUPA"
rather
> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
variable
> (say 10-meter wind speed)....correct?
>
> Thanks,
>
>
>
> From:	"RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
> To:	James P Cipriani/Watson/IBM at IBMUS
> Date:	06/07/2011 02:40 PM
> Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> If your observations are of type ADPSFC or SFCSHP, no interpolation
is
> done, since these are defined as "surface"
> observations.  If, however, your observations are not of that type,
MET
> will perform interpolation of model data to
> whatever the elevation of the observation is.  For more information,
see
> section 4.2.1 of the MET User's Guide.
>
> If you want interpolation to be performed, I think you just need to
set the
> observation type to something besides ADPSFC
> or SFCSHP.  If you are using ascii2nc, for example, you can specify
a
> Message_Type in the first column of your input.
>
> Paul
>
>
> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>> Transaction: Ticket created by jpcipria at us.ibm.com
>>        Queue: met_help
>>      Subject: Question on Interpolation
>>        Owner: Nobody
>>   Requestors: jpcipria at us.ibm.com
>>       Status: new
>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>>
>>
>> MET Team,
>>
>> A quick question on verification at certain levels.  As you know,
sensors
>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>> (temperature) - they are often times placed above and below these
points.
>> Given that the user provides observation information (site
elevation AGL
>> and observation elevation AGL) before running ASCII2NC, does the
MET
>> package do the interpolation (for the observations) to 10-m (wind)
and
> 2-m
>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast vs.
obs?
>>
>> Thanks.
>
>
>


------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: Paul Oldenburg
Time: Thu Jun 09 12:45:43 2011

James,

We have discussed your situation, and the answer to your problem
depends on your model data.  Is your model data in GRIB
format?  If so, can you please send us the wgrib output from one of
your model files.  If you have observations in a
particular range of elevations or levels, say 8-12m, and you have
model levels at 10m, 20m and 30m, then the following
fcst_field will definitely match observations and perform vertical
interpolation for obs at 10m and higher:

fcst_field[] = [ "WIND/Z10-20" ];

We are not sure what the behavior will be for obs points below 10m in
this case.  They may be matched without
interpolation to the bottom model layer at 10m, but maybe not.
Remember, it is important that your observation type is
not a "surface" type.  One good type to use might be ADPUPA, which
will trigger vertical interpolation between model
layers in MET.  We have not encountered this type of situation before.
Please let us know what you find.

Paul


On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
> James,
>
> A surface observation for temperature is 2m above the surface, which
has an absolute elevation of 2m + sea level
> elevation of the observation station location.  This is obviously
terrain dependent.  If you are using the GRIB output
> of WPP as your model data, for example, and it contains surface
temperature, then it will have a 2m TMP forecast value.
>  This will not require interpolation to match a "surface" (i.e.
ADPSFC or SFCSHP) observation in point_stat.
>
> On the other hand, if you do not have a "surface" observation, then
the model data will be interpolated using the two
> adjacent model levels to the elevation of the observation.  This
logic holds for all forecast fields, regardless of
> their defined "surface" elevation.
>
> Unfortunately, the individual most qualified to discuss this matter
with you is on vacation until Thursday.  I will talk
> about this issue with him when he returns.  I hope I have provided
some clarification.
>
> Paul
>
>
> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> Paul,
>>
>> Thanks for the timely reply.  A bit of clarification:
>>
>> (1) When using ASCII2NC, I'm aware that there must be 10 columns in
each
>> text file - two of the columns are station elevation (msl) and
observation
>> (msl).  Let's use temperature as an example here and say the
station was
>> located 60 meters above sea level...if the observation was truly a
2-meter
>> temperature, then the difference between station elevation (60
meters) and
>> observation elevation would be 2-meters, thus making the
observation
>> elevation 62 meters...correct?  If the difference was > or < 2
meters, then
>> vertical interpolation would be done (forecasts to observation
values) for
>> 2-meter temperature (assuming we choose a message type like
"ADPUPA" rather
>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
variable
>> (say 10-meter wind speed)....correct?
>>
>> Thanks,
>>
>>
>>
>> From:	"RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
>> To:	James P Cipriani/Watson/IBM at IBMUS
>> Date:	06/07/2011 02:40 PM
>> Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>>
>>
>>
>> James,
>>
>> If your observations are of type ADPSFC or SFCSHP, no interpolation
is
>> done, since these are defined as "surface"
>> observations.  If, however, your observations are not of that type,
MET
>> will perform interpolation of model data to
>> whatever the elevation of the observation is.  For more
information, see
>> section 4.2.1 of the MET User's Guide.
>>
>> If you want interpolation to be performed, I think you just need to
set the
>> observation type to something besides ADPSFC
>> or SFCSHP.  If you are using ascii2nc, for example, you can specify
a
>> Message_Type in the first column of your input.
>>
>> Paul
>>
>>
>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>        Queue: met_help
>>>      Subject: Question on Interpolation
>>>        Owner: Nobody
>>>   Requestors: jpcipria at us.ibm.com
>>>       Status: new
>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>>
>>>
>>> MET Team,
>>>
>>> A quick question on verification at certain levels.  As you know,
sensors
>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>> (temperature) - they are often times placed above and below these
points.
>>> Given that the user provides observation information (site
elevation AGL
>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>> package do the interpolation (for the observations) to 10-m (wind)
and
>> 2-m
>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs. obs?
>>>
>>> Thanks.
>>
>>
>>
>


------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Thu Jun 09 13:34:31 2011


MET team,

I appreciate your thoughts in this matter.  A few comments/questions:

(1) I have indeed changed my message type to be "ADPUPA" going
forward, per
your previous request (thank you, by the way).
(2) The raw model data is not in GRIB1 so I run it through the WRF
Post
Processor (WPPV3) - soon to be the Unified Post Processor.  However,
the
only wind output that I specify from WPPV3 is U10 and V10, which MET
uses
to calculate total wind speed on the fly - (u_10^2 + v_10^2)^(1/2).
Thus,
when specifying that I want U10 and V10, I am only outputting the
speeds at
10 meters.  So I am not outputting winds at 20 meters, 30 meters, etc.
Should I be altering my WPPV3 control files to specify U and V winds
at
additional levels?
(3) I can definitely tell you that I have stations with sensor
elevations
(both temperature and wind speed) are < 10 meters and some with sensor
elevations > 20 meters.

Any additional information and/or thoughts you have would be greatly
appreciated.

Thanks.



From:	"RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/09/2011 02:46 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

We have discussed your situation, and the answer to your problem
depends on
your model data.  Is your model data in GRIB
format?  If so, can you please send us the wgrib output from one of
your
model files.  If you have observations in a
particular range of elevations or levels, say 8-12m, and you have
model
levels at 10m, 20m and 30m, then the following
fcst_field will definitely match observations and perform vertical
interpolation for obs at 10m and higher:

fcst_field[] = [ "WIND/Z10-20" ];

We are not sure what the behavior will be for obs points below 10m in
this
case.  They may be matched without
interpolation to the bottom model layer at 10m, but maybe not.
Remember,
it is important that your observation type is
not a "surface" type.  One good type to use might be ADPUPA, which
will
trigger vertical interpolation between model
layers in MET.  We have not encountered this type of situation before.
Please let us know what you find.

Paul


On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
> James,
>
> A surface observation for temperature is 2m above the surface, which
has
an absolute elevation of 2m + sea level
> elevation of the observation station location.  This is obviously
terrain
dependent.  If you are using the GRIB output
> of WPP as your model data, for example, and it contains surface
temperature, then it will have a 2m TMP forecast value.
>  This will not require interpolation to match a "surface" (i.e.
ADPSFC or
SFCSHP) observation in point_stat.
>
> On the other hand, if you do not have a "surface" observation, then
the
model data will be interpolated using the two
> adjacent model levels to the elevation of the observation.  This
logic
holds for all forecast fields, regardless of
> their defined "surface" elevation.
>
> Unfortunately, the individual most qualified to discuss this matter
with
you is on vacation until Thursday.  I will talk
> about this issue with him when he returns.  I hope I have provided
some
clarification.
>
> Paul
>
>
> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> Paul,
>>
>> Thanks for the timely reply.  A bit of clarification:
>>
>> (1) When using ASCII2NC, I'm aware that there must be 10 columns in
each
>> text file - two of the columns are station elevation (msl) and
observation
>> (msl).  Let's use temperature as an example here and say the
station was
>> located 60 meters above sea level...if the observation was truly a
2-meter
>> temperature, then the difference between station elevation (60
meters)
and
>> observation elevation would be 2-meters, thus making the
observation
>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
then
>> vertical interpolation would be done (forecasts to observation
values)
for
>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
rather
>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
variable
>> (say 10-meter wind speed)....correct?
>>
>> Thanks,
>>
>>
>>
>> From:		 "RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
>> To:		 James P Cipriani/Watson/IBM at IBMUS
>> Date:		 06/07/2011 02:40 PM
>> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>>
>>
>>
>> James,
>>
>> If your observations are of type ADPSFC or SFCSHP, no interpolation
is
>> done, since these are defined as "surface"
>> observations.  If, however, your observations are not of that type,
MET
>> will perform interpolation of model data to
>> whatever the elevation of the observation is.  For more
information, see
>> section 4.2.1 of the MET User's Guide.
>>
>> If you want interpolation to be performed, I think you just need to
set
the
>> observation type to something besides ADPSFC
>> or SFCSHP.  If you are using ascii2nc, for example, you can specify
a
>> Message_Type in the first column of your input.
>>
>> Paul
>>
>>
>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>        Queue: met_help
>>>      Subject: Question on Interpolation
>>>        Owner: Nobody
>>>   Requestors: jpcipria at us.ibm.com
>>>       Status: new
>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>>
>>>
>>> MET Team,
>>>
>>> A quick question on verification at certain levels.  As you know,
sensors
>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>> (temperature) - they are often times placed above and below these
points.
>>> Given that the user provides observation information (site
elevation
AGL
>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>> package do the interpolation (for the observations) to 10-m (wind)
and
>> 2-m
>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
obs?
>>>
>>> Thanks.
>>
>>
>>
>



------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: John Halley Gotway
Time: Fri Jun 10 10:14:21 2011

James,

This is John Halley Gotway.  I've been working with Paul on the issues
you're having.

It sounds like you'd like to verify wind speed and temperature at
various heights above ground level.  And you have point observations
for those heights.  We've helped some other wind energy folks in
the past do exactly what you're requesting.  As you list in (2), the
most important thing is that you configure WPP to output U, V, and
temperature at additional AGL heights.  So the question is, how
do you configure WPP to do this?  I actually worked on some of the WPP
code in the past to add in this sort of functionality, but I don't
know what, if any of it, was released.

I've sent an email to the WPP support person here asking for detail
about this.  I'll let you know what I hear back.

Supposing you are able to get output at other AGL heights (like 10m,
20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
fcst_field[] = [ "WIND/Z0-50" ];
message_type[] = [ "ADPUPA" ];

Suppose you have an observation of wind speed occurring at 15m.
Point-Stat will look at the model output above/below that - so at 10m
and 20m.  It'll horizontally interpolate to the observation
location at 10m and 20m, and then vertically interpolate those two
values to 15m, the height of the observation.

John

On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> MET team,
>
> I appreciate your thoughts in this matter.  A few
comments/questions:
>
> (1) I have indeed changed my message type to be "ADPUPA" going
forward, per
> your previous request (thank you, by the way).
> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
> Processor (WPPV3) - soon to be the Unified Post Processor.  However,
the
> only wind output that I specify from WPPV3 is U10 and V10, which MET
uses
> to calculate total wind speed on the fly - (u_10^2 + v_10^2)^(1/2).
Thus,
> when specifying that I want U10 and V10, I am only outputting the
speeds at
> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
> Should I be altering my WPPV3 control files to specify U and V winds
at
> additional levels?
> (3) I can definitely tell you that I have stations with sensor
elevations
> (both temperature and wind speed) are < 10 meters and some with
sensor
> elevations > 20 meters.
>
> Any additional information and/or thoughts you have would be greatly
> appreciated.
>
> Thanks.
>
>
>
> From:	"RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
> To:	James P Cipriani/Watson/IBM at IBMUS
> Date:	06/09/2011 02:46 PM
> Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> We have discussed your situation, and the answer to your problem
depends on
> your model data.  Is your model data in GRIB
> format?  If so, can you please send us the wgrib output from one of
your
> model files.  If you have observations in a
> particular range of elevations or levels, say 8-12m, and you have
model
> levels at 10m, 20m and 30m, then the following
> fcst_field will definitely match observations and perform vertical
> interpolation for obs at 10m and higher:
>
> fcst_field[] = [ "WIND/Z10-20" ];
>
> We are not sure what the behavior will be for obs points below 10m
in this
> case.  They may be matched without
> interpolation to the bottom model layer at 10m, but maybe not.
Remember,
> it is important that your observation type is
> not a "surface" type.  One good type to use might be ADPUPA, which
will
> trigger vertical interpolation between model
> layers in MET.  We have not encountered this type of situation
before.
> Please let us know what you find.
>
> Paul
>
>
> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>> James,
>>
>> A surface observation for temperature is 2m above the surface,
which has
> an absolute elevation of 2m + sea level
>> elevation of the observation station location.  This is obviously
terrain
> dependent.  If you are using the GRIB output
>> of WPP as your model data, for example, and it contains surface
> temperature, then it will have a 2m TMP forecast value.
>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC or
> SFCSHP) observation in point_stat.
>>
>> On the other hand, if you do not have a "surface" observation, then
the
> model data will be interpolated using the two
>> adjacent model levels to the elevation of the observation.  This
logic
> holds for all forecast fields, regardless of
>> their defined "surface" elevation.
>>
>> Unfortunately, the individual most qualified to discuss this matter
with
> you is on vacation until Thursday.  I will talk
>> about this issue with him when he returns.  I hope I have provided
some
> clarification.
>>
>> Paul
>>
>>
>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>> Paul,
>>>
>>> Thanks for the timely reply.  A bit of clarification:
>>>
>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in each
>>> text file - two of the columns are station elevation (msl) and
> observation
>>> (msl).  Let's use temperature as an example here and say the
station was
>>> located 60 meters above sea level...if the observation was truly a
> 2-meter
>>> temperature, then the difference between station elevation (60
meters)
> and
>>> observation elevation would be 2-meters, thus making the
observation
>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
> then
>>> vertical interpolation would be done (forecasts to observation
values)
> for
>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
> rather
>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
variable
>>> (say 10-meter wind speed)....correct?
>>>
>>> Thanks,
>>>
>>>
>>>
>>> From:		 "RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
>>> To:		 James P Cipriani/Watson/IBM at IBMUS
>>> Date:		 06/07/2011 02:40 PM
>>> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>>>
>>>
>>>
>>> James,
>>>
>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation is
>>> done, since these are defined as "surface"
>>> observations.  If, however, your observations are not of that
type, MET
>>> will perform interpolation of model data to
>>> whatever the elevation of the observation is.  For more
information, see
>>> section 4.2.1 of the MET User's Guide.
>>>
>>> If you want interpolation to be performed, I think you just need
to set
> the
>>> observation type to something besides ADPSFC
>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>> Message_Type in the first column of your input.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>        Queue: met_help
>>>>      Subject: Question on Interpolation
>>>>        Owner: Nobody
>>>>   Requestors: jpcipria at us.ibm.com
>>>>       Status: new
>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>
>>>>
>>>>
>>>>
>>>> MET Team,
>>>>
>>>> A quick question on verification at certain levels.  As you know,
> sensors
>>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>>> (temperature) - they are often times placed above and below these
> points.
>>>> Given that the user provides observation information (site
elevation
> AGL
>>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>>> package do the interpolation (for the observations) to 10-m
(wind) and
>>> 2-m
>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
> obs?
>>>>
>>>> Thanks.
>>>
>>>
>>>
>>
>
>
>

------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Fri Jun 10 11:38:02 2011


John,

Thanks for the reply.

I do believe that outputting fields such as temperature, dew point,
wind
speed at AGL heights (in meters) is not possible with WPPV3 (at least
not
with the default code compile).  As far as I know, the only options
are to
output these variables on pressure levels (right now, there are 47
levels
possible, from 2 to 1013 mb) or on model levels (hybrid levels).

That was exactly the reason for my last question (about outputting U
and V
at additional 'levels').  So you have responded exactly how I
expected.

Assuming we can get these values at specific heights (m), your
settings are
exactly what I was thinking as well.  Using the MET package has been
very
useful and easy to adapt.

Please let me know what you hear from the WPP group...

Thanks.



From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/10/2011 12:15 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

This is John Halley Gotway.  I've been working with Paul on the issues
you're having.

It sounds like you'd like to verify wind speed and temperature at
various
heights above ground level.  And you have point observations for those
heights.  We've helped some other wind energy folks in
the past do exactly what you're requesting.  As you list in (2), the
most
important thing is that you configure WPP to output U, V, and
temperature
at additional AGL heights.  So the question is, how
do you configure WPP to do this?  I actually worked on some of the WPP
code
in the past to add in this sort of functionality, but I don't know
what, if
any of it, was released.

I've sent an email to the WPP support person here asking for detail
about
this.  I'll let you know what I hear back.

Supposing you are able to get output at other AGL heights (like 10m,
20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
fcst_field[] = [ "WIND/Z0-50" ];
message_type[] = [ "ADPUPA" ];

Suppose you have an observation of wind speed occurring at 15m.
Point-Stat
will look at the model output above/below that - so at 10m and 20m.
It'll
horizontally interpolate to the observation
location at 10m and 20m, and then vertically interpolate those two
values
to 15m, the height of the observation.

John

On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> MET team,
>
> I appreciate your thoughts in this matter.  A few
comments/questions:
>
> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
per
> your previous request (thank you, by the way).
> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
> Processor (WPPV3) - soon to be the Unified Post Processor.  However,
the
> only wind output that I specify from WPPV3 is U10 and V10, which MET
uses
> to calculate total wind speed on the fly - (u_10^2 + v_10^2)^(1/2).
Thus,
> when specifying that I want U10 and V10, I am only outputting the
speeds
at
> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
> Should I be altering my WPPV3 control files to specify U and V winds
at
> additional levels?
> (3) I can definitely tell you that I have stations with sensor
elevations
> (both temperature and wind speed) are < 10 meters and some with
sensor
> elevations > 20 meters.
>
> Any additional information and/or thoughts you have would be greatly
> appreciated.
>
> Thanks.
>
>
>
> From:		 "RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
> To:		 James P Cipriani/Watson/IBM at IBMUS
> Date:		 06/09/2011 02:46 PM
> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> We have discussed your situation, and the answer to your problem
depends
on
> your model data.  Is your model data in GRIB
> format?  If so, can you please send us the wgrib output from one of
your
> model files.  If you have observations in a
> particular range of elevations or levels, say 8-12m, and you have
model
> levels at 10m, 20m and 30m, then the following
> fcst_field will definitely match observations and perform vertical
> interpolation for obs at 10m and higher:
>
> fcst_field[] = [ "WIND/Z10-20" ];
>
> We are not sure what the behavior will be for obs points below 10m
in
this
> case.  They may be matched without
> interpolation to the bottom model layer at 10m, but maybe not.
Remember,
> it is important that your observation type is
> not a "surface" type.  One good type to use might be ADPUPA, which
will
> trigger vertical interpolation between model
> layers in MET.  We have not encountered this type of situation
before.
> Please let us know what you find.
>
> Paul
>
>
> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>> James,
>>
>> A surface observation for temperature is 2m above the surface,
which has
> an absolute elevation of 2m + sea level
>> elevation of the observation station location.  This is obviously
terrain
> dependent.  If you are using the GRIB output
>> of WPP as your model data, for example, and it contains surface
> temperature, then it will have a 2m TMP forecast value.
>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
or
> SFCSHP) observation in point_stat.
>>
>> On the other hand, if you do not have a "surface" observation, then
the
> model data will be interpolated using the two
>> adjacent model levels to the elevation of the observation.  This
logic
> holds for all forecast fields, regardless of
>> their defined "surface" elevation.
>>
>> Unfortunately, the individual most qualified to discuss this matter
with
> you is on vacation until Thursday.  I will talk
>> about this issue with him when he returns.  I hope I have provided
some
> clarification.
>>
>> Paul
>>
>>
>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>> Paul,
>>>
>>> Thanks for the timely reply.  A bit of clarification:
>>>
>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in
each
>>> text file - two of the columns are station elevation (msl) and
> observation
>>> (msl).  Let's use temperature as an example here and say the
station
was
>>> located 60 meters above sea level...if the observation was truly a
> 2-meter
>>> temperature, then the difference between station elevation (60
meters)
> and
>>> observation elevation would be 2-meters, thus making the
observation
>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
> then
>>> vertical interpolation would be done (forecasts to observation
values)
> for
>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
> rather
>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
variable
>>> (say 10-meter wind speed)....correct?
>>>
>>> Thanks,
>>>
>>>
>>>
>>> From:		 		  "RAL HelpDesk {for Paul Oldenburg}"
<met_help at ucar.edu>
>>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>>> Date:		 		  06/07/2011 02:40 PM
>>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question
on Interpolation
>>>
>>>
>>>
>>> James,
>>>
>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation is
>>> done, since these are defined as "surface"
>>> observations.  If, however, your observations are not of that
type, MET
>>> will perform interpolation of model data to
>>> whatever the elevation of the observation is.  For more
information,
see
>>> section 4.2.1 of the MET User's Guide.
>>>
>>> If you want interpolation to be performed, I think you just need
to set
> the
>>> observation type to something besides ADPSFC
>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>> Message_Type in the first column of your input.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>        Queue: met_help
>>>>      Subject: Question on Interpolation
>>>>        Owner: Nobody
>>>>   Requestors: jpcipria at us.ibm.com
>>>>       Status: new
>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>
>>>>
>>>>
>>>>
>>>>
>>>> MET Team,
>>>>
>>>> A quick question on verification at certain levels.  As you know,
> sensors
>>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>>> (temperature) - they are often times placed above and below these
> points.
>>>> Given that the user provides observation information (site
elevation
> AGL
>>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>>> package do the interpolation (for the observations) to 10-m
(wind) and
>>> 2-m
>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
> obs?
>>>>
>>>> Thanks.
>>>
>>>
>>>
>>
>
>
>


------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: John Halley Gotway
Time: Fri Jun 10 14:32:52 2011

James,

OK, that capability is not available in the currently released version
of WPPv3.2.  It's still an open question of whether or not it's
available in the initial release of UPP - the unified post
processor - which replaces UPP.  The WRF person I'm working with will
check next week.

I had made some changes to WPPv3.2 to allow for the dumping of winds
and temperature at various AGL levels.  And I'm happy to send you that
code.  However, I do so with a warning.  The changes are
basically an adaptation of the code that was used to dump out TMP, U,
and V at flight data levels.  It isn't clear whether or not it is
appropriate to apply that same algorithm to levels much closer
to the ground.  From a software standpoint, it gives you reasonable
data, but I can't tell you how correct it is from a scientific
standpoint.  So use it at your own risk.

The output of these records is controlled by the namelist settings for
(wrf_cntrl.parm):
   "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"

And these settings give you AGL output at 20m through 200m every 10
meters, and then 200m through 1000m every 50 meters.

You can retrieve the modified version of WPP here:
   ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz

Hope that helps.

John

On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> John,
>
> Thanks for the reply.
>
> I do believe that outputting fields such as temperature, dew point,
wind
> speed at AGL heights (in meters) is not possible with WPPV3 (at
least not
> with the default code compile).  As far as I know, the only options
are to
> output these variables on pressure levels (right now, there are 47
levels
> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>
> That was exactly the reason for my last question (about outputting U
and V
> at additional 'levels').  So you have responded exactly how I
expected.
>
> Assuming we can get these values at specific heights (m), your
settings are
> exactly what I was thinking as well.  Using the MET package has been
very
> useful and easy to adapt.
>
> Please let me know what you hear from the WPP group...
>
> Thanks.
>
>
>
> From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
> To:	James P Cipriani/Watson/IBM at IBMUS
> Date:	06/10/2011 12:15 PM
> Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> This is John Halley Gotway.  I've been working with Paul on the
issues
> you're having.
>
> It sounds like you'd like to verify wind speed and temperature at
various
> heights above ground level.  And you have point observations for
those
> heights.  We've helped some other wind energy folks in
> the past do exactly what you're requesting.  As you list in (2), the
most
> important thing is that you configure WPP to output U, V, and
temperature
> at additional AGL heights.  So the question is, how
> do you configure WPP to do this?  I actually worked on some of the
WPP code
> in the past to add in this sort of functionality, but I don't know
what, if
> any of it, was released.
>
> I've sent an email to the WPP support person here asking for detail
about
> this.  I'll let you know what I hear back.
>
> Supposing you are able to get output at other AGL heights (like 10m,
> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
> fcst_field[] = [ "WIND/Z0-50" ];
> message_type[] = [ "ADPUPA" ];
>
> Suppose you have an observation of wind speed occurring at 15m.
Point-Stat
> will look at the model output above/below that - so at 10m and 20m.
It'll
> horizontally interpolate to the observation
> location at 10m and 20m, and then vertically interpolate those two
values
> to 15m, the height of the observation.
>
> John
>
> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> MET team,
>>
>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>
>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
> per
>> your previous request (thank you, by the way).
>> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
>> Processor (WPPV3) - soon to be the Unified Post Processor.
However, the
>> only wind output that I specify from WPPV3 is U10 and V10, which
MET uses
>> to calculate total wind speed on the fly - (u_10^2 + v_10^2)^(1/2).
> Thus,
>> when specifying that I want U10 and V10, I am only outputting the
speeds
> at
>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>> Should I be altering my WPPV3 control files to specify U and V
winds at
>> additional levels?
>> (3) I can definitely tell you that I have stations with sensor
elevations
>> (both temperature and wind speed) are < 10 meters and some with
sensor
>> elevations > 20 meters.
>>
>> Any additional information and/or thoughts you have would be
greatly
>> appreciated.
>>
>> Thanks.
>>
>>
>>
>> From:		 "RAL HelpDesk {for Paul Oldenburg}" <met_help at ucar.edu>
>> To:		 James P Cipriani/Watson/IBM at IBMUS
>> Date:		 06/09/2011 02:46 PM
>> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>>
>>
>>
>> James,
>>
>> We have discussed your situation, and the answer to your problem
depends
> on
>> your model data.  Is your model data in GRIB
>> format?  If so, can you please send us the wgrib output from one of
your
>> model files.  If you have observations in a
>> particular range of elevations or levels, say 8-12m, and you have
model
>> levels at 10m, 20m and 30m, then the following
>> fcst_field will definitely match observations and perform vertical
>> interpolation for obs at 10m and higher:
>>
>> fcst_field[] = [ "WIND/Z10-20" ];
>>
>> We are not sure what the behavior will be for obs points below 10m
in
> this
>> case.  They may be matched without
>> interpolation to the bottom model layer at 10m, but maybe not.
Remember,
>> it is important that your observation type is
>> not a "surface" type.  One good type to use might be ADPUPA, which
will
>> trigger vertical interpolation between model
>> layers in MET.  We have not encountered this type of situation
before.
>> Please let us know what you find.
>>
>> Paul
>>
>>
>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>> James,
>>>
>>> A surface observation for temperature is 2m above the surface,
which has
>> an absolute elevation of 2m + sea level
>>> elevation of the observation station location.  This is obviously
> terrain
>> dependent.  If you are using the GRIB output
>>> of WPP as your model data, for example, and it contains surface
>> temperature, then it will have a 2m TMP forecast value.
>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
> or
>> SFCSHP) observation in point_stat.
>>>
>>> On the other hand, if you do not have a "surface" observation,
then the
>> model data will be interpolated using the two
>>> adjacent model levels to the elevation of the observation.  This
logic
>> holds for all forecast fields, regardless of
>>> their defined "surface" elevation.
>>>
>>> Unfortunately, the individual most qualified to discuss this
matter with
>> you is on vacation until Thursday.  I will talk
>>> about this issue with him when he returns.  I hope I have provided
some
>> clarification.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>
>>>>
>>>> Paul,
>>>>
>>>> Thanks for the timely reply.  A bit of clarification:
>>>>
>>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in
> each
>>>> text file - two of the columns are station elevation (msl) and
>> observation
>>>> (msl).  Let's use temperature as an example here and say the
station
> was
>>>> located 60 meters above sea level...if the observation was truly
a
>> 2-meter
>>>> temperature, then the difference between station elevation (60
meters)
>> and
>>>> observation elevation would be 2-meters, thus making the
observation
>>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
>> then
>>>> vertical interpolation would be done (forecasts to observation
values)
>> for
>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>> rather
>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
> variable
>>>> (say 10-meter wind speed)....correct?
>>>>
>>>> Thanks,
>>>>
>>>>
>>>>
>>>> From:		 		  "RAL HelpDesk {for Paul Oldenburg}"
> <met_help at ucar.edu>
>>>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>>>> Date:		 		  06/07/2011 02:40 PM
>>>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question
> on Interpolation
>>>>
>>>>
>>>>
>>>> James,
>>>>
>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation is
>>>> done, since these are defined as "surface"
>>>> observations.  If, however, your observations are not of that
type, MET
>>>> will perform interpolation of model data to
>>>> whatever the elevation of the observation is.  For more
information,
> see
>>>> section 4.2.1 of the MET User's Guide.
>>>>
>>>> If you want interpolation to be performed, I think you just need
to set
>> the
>>>> observation type to something besides ADPSFC
>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>>> Message_Type in the first column of your input.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>
>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>        Queue: met_help
>>>>>      Subject: Question on Interpolation
>>>>>        Owner: Nobody
>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>       Status: new
>>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> MET Team,
>>>>>
>>>>> A quick question on verification at certain levels.  As you
know,
>> sensors
>>>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>>>> (temperature) - they are often times placed above and below
these
>> points.
>>>>> Given that the user provides observation information (site
elevation
>> AGL
>>>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>>>> package do the interpolation (for the observations) to 10-m
(wind) and
>>>> 2-m
>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
>> obs?
>>>>>
>>>>> Thanks.
>>>>
>>>>
>>>>
>>>
>>
>>
>>
>
>

------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Fri Jun 10 15:10:52 2011


John,

Thanks so much.  This will be a big help.  The algorithm for Flight
levels
should still be applicable for these low levels.  I'll keep you
posted.

What is the specific piece of code (in sorc/wrfpost) to which you
applied
these changes?

As far as the UPP is concerned, I read through the first version of
the
documentation and the following fields have been added:

(1) Fields on sigma surfaces
(2) Fields on PV surfaces
(3) TEMP at 10 meters

No fields have been added for output to specific AGL's.

Let me recompile and run some tests and I'll be sure to keep you
informed.
This will be useful for others as well.

Thanks.




From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/10/2011 04:33 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

OK, that capability is not available in the currently released version
of
WPPv3.2.  It's still an open question of whether or not it's available
in
the initial release of UPP - the unified post
processor - which replaces UPP.  The WRF person I'm working with will
check
next week.

I had made some changes to WPPv3.2 to allow for the dumping of winds
and
temperature at various AGL levels.  And I'm happy to send you that
code.
However, I do so with a warning.  The changes are
basically an adaptation of the code that was used to dump out TMP, U,
and V
at flight data levels.  It isn't clear whether or not it is
appropriate to
apply that same algorithm to levels much closer
to the ground.  From a software standpoint, it gives you reasonable
data,
but I can't tell you how correct it is from a scientific standpoint.
So
use it at your own risk.

The output of these records is controlled by the namelist settings for
(wrf_cntrl.parm):
   "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"

And these settings give you AGL output at 20m through 200m every 10
meters,
and then 200m through 1000m every 50 meters.

You can retrieve the modified version of WPP here:
   ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz

Hope that helps.

John

On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> John,
>
> Thanks for the reply.
>
> I do believe that outputting fields such as temperature, dew point,
wind
> speed at AGL heights (in meters) is not possible with WPPV3 (at
least not
> with the default code compile).  As far as I know, the only options
are
to
> output these variables on pressure levels (right now, there are 47
levels
> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>
> That was exactly the reason for my last question (about outputting U
and
V
> at additional 'levels').  So you have responded exactly how I
expected.
>
> Assuming we can get these values at specific heights (m), your
settings
are
> exactly what I was thinking as well.  Using the MET package has been
very
> useful and easy to adapt.
>
> Please let me know what you hear from the WPP group...
>
> Thanks.
>
>
>
> From:		 "RAL HelpDesk {for John Halley Gotway}"
<met_help at ucar.edu>
> To:		 James P Cipriani/Watson/IBM at IBMUS
> Date:		 06/10/2011 12:15 PM
> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> This is John Halley Gotway.  I've been working with Paul on the
issues
> you're having.
>
> It sounds like you'd like to verify wind speed and temperature at
various
> heights above ground level.  And you have point observations for
those
> heights.  We've helped some other wind energy folks in
> the past do exactly what you're requesting.  As you list in (2), the
most
> important thing is that you configure WPP to output U, V, and
temperature
> at additional AGL heights.  So the question is, how
> do you configure WPP to do this?  I actually worked on some of the
WPP
code
> in the past to add in this sort of functionality, but I don't know
what,
if
> any of it, was released.
>
> I've sent an email to the WPP support person here asking for detail
about
> this.  I'll let you know what I hear back.
>
> Supposing you are able to get output at other AGL heights (like 10m,
> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
> fcst_field[] = [ "WIND/Z0-50" ];
> message_type[] = [ "ADPUPA" ];
>
> Suppose you have an observation of wind speed occurring at 15m.
Point-Stat
> will look at the model output above/below that - so at 10m and 20m.
It'll
> horizontally interpolate to the observation
> location at 10m and 20m, and then vertically interpolate those two
values
> to 15m, the height of the observation.
>
> John
>
> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> MET team,
>>
>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>
>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
> per
>> your previous request (thank you, by the way).
>> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
>> Processor (WPPV3) - soon to be the Unified Post Processor.
However, the
>> only wind output that I specify from WPPV3 is U10 and V10, which
MET
uses
>> to calculate total wind speed on the fly - (u_10^2 + v_10^2)^(1/2).
> Thus,
>> when specifying that I want U10 and V10, I am only outputting the
speeds
> at
>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>> Should I be altering my WPPV3 control files to specify U and V
winds at
>> additional levels?
>> (3) I can definitely tell you that I have stations with sensor
elevations
>> (both temperature and wind speed) are < 10 meters and some with
sensor
>> elevations > 20 meters.
>>
>> Any additional information and/or thoughts you have would be
greatly
>> appreciated.
>>
>> Thanks.
>>
>>
>>
>> From:		 		  "RAL HelpDesk {for Paul Oldenburg}"
<met_help at ucar.edu>
>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>> Date:		 		  06/09/2011 02:46 PM
>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question on
Interpolation
>>
>>
>>
>> James,
>>
>> We have discussed your situation, and the answer to your problem
depends
> on
>> your model data.  Is your model data in GRIB
>> format?  If so, can you please send us the wgrib output from one of
your
>> model files.  If you have observations in a
>> particular range of elevations or levels, say 8-12m, and you have
model
>> levels at 10m, 20m and 30m, then the following
>> fcst_field will definitely match observations and perform vertical
>> interpolation for obs at 10m and higher:
>>
>> fcst_field[] = [ "WIND/Z10-20" ];
>>
>> We are not sure what the behavior will be for obs points below 10m
in
> this
>> case.  They may be matched without
>> interpolation to the bottom model layer at 10m, but maybe not.
Remember,
>> it is important that your observation type is
>> not a "surface" type.  One good type to use might be ADPUPA, which
will
>> trigger vertical interpolation between model
>> layers in MET.  We have not encountered this type of situation
before.
>> Please let us know what you find.
>>
>> Paul
>>
>>
>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>> James,
>>>
>>> A surface observation for temperature is 2m above the surface,
which
has
>> an absolute elevation of 2m + sea level
>>> elevation of the observation station location.  This is obviously
> terrain
>> dependent.  If you are using the GRIB output
>>> of WPP as your model data, for example, and it contains surface
>> temperature, then it will have a 2m TMP forecast value.
>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
> or
>> SFCSHP) observation in point_stat.
>>>
>>> On the other hand, if you do not have a "surface" observation,
then the
>> model data will be interpolated using the two
>>> adjacent model levels to the elevation of the observation.  This
logic
>> holds for all forecast fields, regardless of
>>> their defined "surface" elevation.
>>>
>>> Unfortunately, the individual most qualified to discuss this
matter
with
>> you is on vacation until Thursday.  I will talk
>>> about this issue with him when he returns.  I hope I have provided
some
>> clarification.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>
>>>>
>>>> Paul,
>>>>
>>>> Thanks for the timely reply.  A bit of clarification:
>>>>
>>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in
> each
>>>> text file - two of the columns are station elevation (msl) and
>> observation
>>>> (msl).  Let's use temperature as an example here and say the
station
> was
>>>> located 60 meters above sea level...if the observation was truly
a
>> 2-meter
>>>> temperature, then the difference between station elevation (60
meters)
>> and
>>>> observation elevation would be 2-meters, thus making the
observation
>>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
>> then
>>>> vertical interpolation would be done (forecasts to observation
values)
>> for
>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>> rather
>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
> variable
>>>> (say 10-meter wind speed)....correct?
>>>>
>>>> Thanks,
>>>>
>>>>
>>>>
>>>> From:		 		  		 		   "RAL HelpDesk {for
Paul Oldenburg}"
> <met_help at ucar.edu>
>>>> To:		 		  		 		   James P
Cipriani/Watson/IBM at IBMUS
>>>> Date:		 		  		 		   06/07/2011 02:40
PM
>>>> Subject:		 		  		 		   Re:
[rt.rap.ucar.edu #47360] Question
> on Interpolation
>>>>
>>>>
>>>>
>>>> James,
>>>>
>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation is
>>>> done, since these are defined as "surface"
>>>> observations.  If, however, your observations are not of that
type,
MET
>>>> will perform interpolation of model data to
>>>> whatever the elevation of the observation is.  For more
information,
> see
>>>> section 4.2.1 of the MET User's Guide.
>>>>
>>>> If you want interpolation to be performed, I think you just need
to
set
>> the
>>>> observation type to something besides ADPSFC
>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>>> Message_Type in the first column of your input.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>
>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>        Queue: met_help
>>>>>      Subject: Question on Interpolation
>>>>>        Owner: Nobody
>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>       Status: new
>>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> MET Team,
>>>>>
>>>>> A quick question on verification at certain levels.  As you
know,
>> sensors
>>>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>>>> (temperature) - they are often times placed above and below
these
>> points.
>>>>> Given that the user provides observation information (site
elevation
>> AGL
>>>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>>>> package do the interpolation (for the observations) to 10-m
(wind)
and
>>>> 2-m
>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
>> obs?
>>>>>
>>>>> Thanks.
>>>>
>>>>
>>>>
>>>
>>
>>
>>
>
>


------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Fri Jun 10 15:12:57 2011


I see sorc/wrfpost/AGLVL.f was modified...any other modifications I
should
know about?

Thanks.



From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/10/2011 04:33 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

OK, that capability is not available in the currently released version
of
WPPv3.2.  It's still an open question of whether or not it's available
in
the initial release of UPP - the unified post
processor - which replaces UPP.  The WRF person I'm working with will
check
next week.

I had made some changes to WPPv3.2 to allow for the dumping of winds
and
temperature at various AGL levels.  And I'm happy to send you that
code.
However, I do so with a warning.  The changes are
basically an adaptation of the code that was used to dump out TMP, U,
and V
at flight data levels.  It isn't clear whether or not it is
appropriate to
apply that same algorithm to levels much closer
to the ground.  From a software standpoint, it gives you reasonable
data,
but I can't tell you how correct it is from a scientific standpoint.
So
use it at your own risk.

The output of these records is controlled by the namelist settings for
(wrf_cntrl.parm):
   "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"

And these settings give you AGL output at 20m through 200m every 10
meters,
and then 200m through 1000m every 50 meters.

You can retrieve the modified version of WPP here:
   ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz

Hope that helps.

John

On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> John,
>
> Thanks for the reply.
>
> I do believe that outputting fields such as temperature, dew point,
wind
> speed at AGL heights (in meters) is not possible with WPPV3 (at
least not
> with the default code compile).  As far as I know, the only options
are
to
> output these variables on pressure levels (right now, there are 47
levels
> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>
> That was exactly the reason for my last question (about outputting U
and
V
> at additional 'levels').  So you have responded exactly how I
expected.
>
> Assuming we can get these values at specific heights (m), your
settings
are
> exactly what I was thinking as well.  Using the MET package has been
very
> useful and easy to adapt.
>
> Please let me know what you hear from the WPP group...
>
> Thanks.
>
>
>
> From:		 "RAL HelpDesk {for John Halley Gotway}"
<met_help at ucar.edu>
> To:		 James P Cipriani/Watson/IBM at IBMUS
> Date:		 06/10/2011 12:15 PM
> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> This is John Halley Gotway.  I've been working with Paul on the
issues
> you're having.
>
> It sounds like you'd like to verify wind speed and temperature at
various
> heights above ground level.  And you have point observations for
those
> heights.  We've helped some other wind energy folks in
> the past do exactly what you're requesting.  As you list in (2), the
most
> important thing is that you configure WPP to output U, V, and
temperature
> at additional AGL heights.  So the question is, how
> do you configure WPP to do this?  I actually worked on some of the
WPP
code
> in the past to add in this sort of functionality, but I don't know
what,
if
> any of it, was released.
>
> I've sent an email to the WPP support person here asking for detail
about
> this.  I'll let you know what I hear back.
>
> Supposing you are able to get output at other AGL heights (like 10m,
> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
> fcst_field[] = [ "WIND/Z0-50" ];
> message_type[] = [ "ADPUPA" ];
>
> Suppose you have an observation of wind speed occurring at 15m.
Point-Stat
> will look at the model output above/below that - so at 10m and 20m.
It'll
> horizontally interpolate to the observation
> location at 10m and 20m, and then vertically interpolate those two
values
> to 15m, the height of the observation.
>
> John
>
> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> MET team,
>>
>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>
>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
> per
>> your previous request (thank you, by the way).
>> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
>> Processor (WPPV3) - soon to be the Unified Post Processor.
However, the
>> only wind output that I specify from WPPV3 is U10 and V10, which
MET
uses
>> to calculate total wind speed on the fly - (u_10^2 + v_10^2)^(1/2).
> Thus,
>> when specifying that I want U10 and V10, I am only outputting the
speeds
> at
>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>> Should I be altering my WPPV3 control files to specify U and V
winds at
>> additional levels?
>> (3) I can definitely tell you that I have stations with sensor
elevations
>> (both temperature and wind speed) are < 10 meters and some with
sensor
>> elevations > 20 meters.
>>
>> Any additional information and/or thoughts you have would be
greatly
>> appreciated.
>>
>> Thanks.
>>
>>
>>
>> From:		 		  "RAL HelpDesk {for Paul Oldenburg}"
<met_help at ucar.edu>
>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>> Date:		 		  06/09/2011 02:46 PM
>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question on
Interpolation
>>
>>
>>
>> James,
>>
>> We have discussed your situation, and the answer to your problem
depends
> on
>> your model data.  Is your model data in GRIB
>> format?  If so, can you please send us the wgrib output from one of
your
>> model files.  If you have observations in a
>> particular range of elevations or levels, say 8-12m, and you have
model
>> levels at 10m, 20m and 30m, then the following
>> fcst_field will definitely match observations and perform vertical
>> interpolation for obs at 10m and higher:
>>
>> fcst_field[] = [ "WIND/Z10-20" ];
>>
>> We are not sure what the behavior will be for obs points below 10m
in
> this
>> case.  They may be matched without
>> interpolation to the bottom model layer at 10m, but maybe not.
Remember,
>> it is important that your observation type is
>> not a "surface" type.  One good type to use might be ADPUPA, which
will
>> trigger vertical interpolation between model
>> layers in MET.  We have not encountered this type of situation
before.
>> Please let us know what you find.
>>
>> Paul
>>
>>
>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>> James,
>>>
>>> A surface observation for temperature is 2m above the surface,
which
has
>> an absolute elevation of 2m + sea level
>>> elevation of the observation station location.  This is obviously
> terrain
>> dependent.  If you are using the GRIB output
>>> of WPP as your model data, for example, and it contains surface
>> temperature, then it will have a 2m TMP forecast value.
>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
> or
>> SFCSHP) observation in point_stat.
>>>
>>> On the other hand, if you do not have a "surface" observation,
then the
>> model data will be interpolated using the two
>>> adjacent model levels to the elevation of the observation.  This
logic
>> holds for all forecast fields, regardless of
>>> their defined "surface" elevation.
>>>
>>> Unfortunately, the individual most qualified to discuss this
matter
with
>> you is on vacation until Thursday.  I will talk
>>> about this issue with him when he returns.  I hope I have provided
some
>> clarification.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>
>>>>
>>>> Paul,
>>>>
>>>> Thanks for the timely reply.  A bit of clarification:
>>>>
>>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in
> each
>>>> text file - two of the columns are station elevation (msl) and
>> observation
>>>> (msl).  Let's use temperature as an example here and say the
station
> was
>>>> located 60 meters above sea level...if the observation was truly
a
>> 2-meter
>>>> temperature, then the difference between station elevation (60
meters)
>> and
>>>> observation elevation would be 2-meters, thus making the
observation
>>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
>> then
>>>> vertical interpolation would be done (forecasts to observation
values)
>> for
>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>> rather
>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
> variable
>>>> (say 10-meter wind speed)....correct?
>>>>
>>>> Thanks,
>>>>
>>>>
>>>>
>>>> From:		 		  		 		   "RAL HelpDesk {for
Paul Oldenburg}"
> <met_help at ucar.edu>
>>>> To:		 		  		 		   James P
Cipriani/Watson/IBM at IBMUS
>>>> Date:		 		  		 		   06/07/2011 02:40
PM
>>>> Subject:		 		  		 		   Re:
[rt.rap.ucar.edu #47360] Question
> on Interpolation
>>>>
>>>>
>>>>
>>>> James,
>>>>
>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation is
>>>> done, since these are defined as "surface"
>>>> observations.  If, however, your observations are not of that
type,
MET
>>>> will perform interpolation of model data to
>>>> whatever the elevation of the observation is.  For more
information,
> see
>>>> section 4.2.1 of the MET User's Guide.
>>>>
>>>> If you want interpolation to be performed, I think you just need
to
set
>> the
>>>> observation type to something besides ADPSFC
>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>>> Message_Type in the first column of your input.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>
>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>        Queue: met_help
>>>>>      Subject: Question on Interpolation
>>>>>        Owner: Nobody
>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>       Status: new
>>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> MET Team,
>>>>>
>>>>> A quick question on verification at certain levels.  As you
know,
>> sensors
>>>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>>>> (temperature) - they are often times placed above and below
these
>> points.
>>>>> Given that the user provides observation information (site
elevation
>> AGL
>>>>> and observation elevation AGL) before running ASCII2NC, does the
MET
>>>>> package do the interpolation (for the observations) to 10-m
(wind)
and
>>>> 2-m
>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
>> obs?
>>>>>
>>>>> Thanks.
>>>>
>>>>
>>>>
>>>
>>
>>
>>
>
>


------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: John Halley Gotway
Time: Fri Jun 10 15:21:30 2011

James,

I see differences in the following files:
  v3.2/WPPV3/sorc/wrfpost/AGLVL.f # NEW
  v3.2/WPPV3/sorc/wrfpost/MISCLN.f
  v3.2/WPPV3/sorc/wrfpost/POSTDATA.f
  v3.2/WPPV3/sorc/wrfpost/makefile
  v3.2/WPPV3/parm/wrf_cntrl.parm

Not sure if that's all of the differences or not.

John

On 06/10/2011 03:12 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> I see sorc/wrfpost/AGLVL.f was modified...any other modifications I
should
> know about?
>
> Thanks.
>
>
>
> From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
> To:	James P Cipriani/Watson/IBM at IBMUS
> Date:	06/10/2011 04:33 PM
> Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> OK, that capability is not available in the currently released
version of
> WPPv3.2.  It's still an open question of whether or not it's
available in
> the initial release of UPP - the unified post
> processor - which replaces UPP.  The WRF person I'm working with
will check
> next week.
>
> I had made some changes to WPPv3.2 to allow for the dumping of winds
and
> temperature at various AGL levels.  And I'm happy to send you that
code.
> However, I do so with a warning.  The changes are
> basically an adaptation of the code that was used to dump out TMP,
U, and V
> at flight data levels.  It isn't clear whether or not it is
appropriate to
> apply that same algorithm to levels much closer
> to the ground.  From a software standpoint, it gives you reasonable
data,
> but I can't tell you how correct it is from a scientific standpoint.
So
> use it at your own risk.
>
> The output of these records is controlled by the namelist settings
for
> (wrf_cntrl.parm):
>    "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"
>
> And these settings give you AGL output at 20m through 200m every 10
meters,
> and then 200m through 1000m every 50 meters.
>
> You can retrieve the modified version of WPP here:
>
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz
>
> Hope that helps.
>
> John
>
> On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> John,
>>
>> Thanks for the reply.
>>
>> I do believe that outputting fields such as temperature, dew point,
wind
>> speed at AGL heights (in meters) is not possible with WPPV3 (at
least not
>> with the default code compile).  As far as I know, the only options
are
> to
>> output these variables on pressure levels (right now, there are 47
levels
>> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>>
>> That was exactly the reason for my last question (about outputting
U and
> V
>> at additional 'levels').  So you have responded exactly how I
expected.
>>
>> Assuming we can get these values at specific heights (m), your
settings
> are
>> exactly what I was thinking as well.  Using the MET package has
been very
>> useful and easy to adapt.
>>
>> Please let me know what you hear from the WPP group...
>>
>> Thanks.
>>
>>
>>
>> From:		 "RAL HelpDesk {for John Halley Gotway}"
> <met_help at ucar.edu>
>> To:		 James P Cipriani/Watson/IBM at IBMUS
>> Date:		 06/10/2011 12:15 PM
>> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>>
>>
>>
>> James,
>>
>> This is John Halley Gotway.  I've been working with Paul on the
issues
>> you're having.
>>
>> It sounds like you'd like to verify wind speed and temperature at
various
>> heights above ground level.  And you have point observations for
those
>> heights.  We've helped some other wind energy folks in
>> the past do exactly what you're requesting.  As you list in (2),
the most
>> important thing is that you configure WPP to output U, V, and
temperature
>> at additional AGL heights.  So the question is, how
>> do you configure WPP to do this?  I actually worked on some of the
WPP
> code
>> in the past to add in this sort of functionality, but I don't know
what,
> if
>> any of it, was released.
>>
>> I've sent an email to the WPP support person here asking for detail
about
>> this.  I'll let you know what I hear back.
>>
>> Supposing you are able to get output at other AGL heights (like
10m,
>> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
>> fcst_field[] = [ "WIND/Z0-50" ];
>> message_type[] = [ "ADPUPA" ];
>>
>> Suppose you have an observation of wind speed occurring at 15m.
> Point-Stat
>> will look at the model output above/below that - so at 10m and 20m.
> It'll
>> horizontally interpolate to the observation
>> location at 10m and 20m, and then vertically interpolate those two
values
>> to 15m, the height of the observation.
>>
>> John
>>
>> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>> MET team,
>>>
>>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>>
>>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
>> per
>>> your previous request (thank you, by the way).
>>> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
>>> Processor (WPPV3) - soon to be the Unified Post Processor.
However, the
>>> only wind output that I specify from WPPV3 is U10 and V10, which
MET
> uses
>>> to calculate total wind speed on the fly - (u_10^2 +
v_10^2)^(1/2).
>> Thus,
>>> when specifying that I want U10 and V10, I am only outputting the
speeds
>> at
>>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>>> Should I be altering my WPPV3 control files to specify U and V
winds at
>>> additional levels?
>>> (3) I can definitely tell you that I have stations with sensor
> elevations
>>> (both temperature and wind speed) are < 10 meters and some with
sensor
>>> elevations > 20 meters.
>>>
>>> Any additional information and/or thoughts you have would be
greatly
>>> appreciated.
>>>
>>> Thanks.
>>>
>>>
>>>
>>> From:		 		  "RAL HelpDesk {for Paul Oldenburg}"
> <met_help at ucar.edu>
>>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>>> Date:		 		  06/09/2011 02:46 PM
>>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question on
> Interpolation
>>>
>>>
>>>
>>> James,
>>>
>>> We have discussed your situation, and the answer to your problem
depends
>> on
>>> your model data.  Is your model data in GRIB
>>> format?  If so, can you please send us the wgrib output from one
of your
>>> model files.  If you have observations in a
>>> particular range of elevations or levels, say 8-12m, and you have
model
>>> levels at 10m, 20m and 30m, then the following
>>> fcst_field will definitely match observations and perform vertical
>>> interpolation for obs at 10m and higher:
>>>
>>> fcst_field[] = [ "WIND/Z10-20" ];
>>>
>>> We are not sure what the behavior will be for obs points below 10m
in
>> this
>>> case.  They may be matched without
>>> interpolation to the bottom model layer at 10m, but maybe not.
> Remember,
>>> it is important that your observation type is
>>> not a "surface" type.  One good type to use might be ADPUPA, which
will
>>> trigger vertical interpolation between model
>>> layers in MET.  We have not encountered this type of situation
before.
>>> Please let us know what you find.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>>> James,
>>>>
>>>> A surface observation for temperature is 2m above the surface,
which
> has
>>> an absolute elevation of 2m + sea level
>>>> elevation of the observation station location.  This is obviously
>> terrain
>>> dependent.  If you are using the GRIB output
>>>> of WPP as your model data, for example, and it contains surface
>>> temperature, then it will have a 2m TMP forecast value.
>>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
>> or
>>> SFCSHP) observation in point_stat.
>>>>
>>>> On the other hand, if you do not have a "surface" observation,
then the
>>> model data will be interpolated using the two
>>>> adjacent model levels to the elevation of the observation.  This
logic
>>> holds for all forecast fields, regardless of
>>>> their defined "surface" elevation.
>>>>
>>>> Unfortunately, the individual most qualified to discuss this
matter
> with
>>> you is on vacation until Thursday.  I will talk
>>>> about this issue with him when he returns.  I hope I have
provided some
>>> clarification.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>
>>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>>
>>>>>
>>>>> Paul,
>>>>>
>>>>> Thanks for the timely reply.  A bit of clarification:
>>>>>
>>>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in
>> each
>>>>> text file - two of the columns are station elevation (msl) and
>>> observation
>>>>> (msl).  Let's use temperature as an example here and say the
station
>> was
>>>>> located 60 meters above sea level...if the observation was truly
a
>>> 2-meter
>>>>> temperature, then the difference between station elevation (60
meters)
>>> and
>>>>> observation elevation would be 2-meters, thus making the
observation
>>>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
>>> then
>>>>> vertical interpolation would be done (forecasts to observation
values)
>>> for
>>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>>> rather
>>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
>> variable
>>>>> (say 10-meter wind speed)....correct?
>>>>>
>>>>> Thanks,
>>>>>
>>>>>
>>>>>
>>>>> From:		 		  		 		   "RAL HelpDesk {for
> Paul Oldenburg}"
>> <met_help at ucar.edu>
>>>>> To:		 		  		 		   James P
> Cipriani/Watson/IBM at IBMUS
>>>>> Date:		 		  		 		   06/07/2011 02:40
> PM
>>>>> Subject:		 		  		 		   Re:
> [rt.rap.ucar.edu #47360] Question
>> on Interpolation
>>>>>
>>>>>
>>>>>
>>>>> James,
>>>>>
>>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation is
>>>>> done, since these are defined as "surface"
>>>>> observations.  If, however, your observations are not of that
type,
> MET
>>>>> will perform interpolation of model data to
>>>>> whatever the elevation of the observation is.  For more
information,
>> see
>>>>> section 4.2.1 of the MET User's Guide.
>>>>>
>>>>> If you want interpolation to be performed, I think you just need
to
> set
>>> the
>>>>> observation type to something besides ADPSFC
>>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>>>> Message_Type in the first column of your input.
>>>>>
>>>>> Paul
>>>>>
>>>>>
>>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>>
>>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>>        Queue: met_help
>>>>>>      Subject: Question on Interpolation
>>>>>>        Owner: Nobody
>>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>>       Status: new
>>>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> MET Team,
>>>>>>
>>>>>> A quick question on verification at certain levels.  As you
know,
>>> sensors
>>>>>> at point sites are not always placed at 10-meters (wind) and 2-
meters
>>>>>> (temperature) - they are often times placed above and below
these
>>> points.
>>>>>> Given that the user provides observation information (site
elevation
>>> AGL
>>>>>> and observation elevation AGL) before running ASCII2NC, does
the MET
>>>>>> package do the interpolation (for the observations) to 10-m
(wind)
> and
>>>>> 2-m
>>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
>>> obs?
>>>>>>
>>>>>> Thanks.
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>>
>>
>>
>
>

------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Fri Jun 10 15:39:05 2011


John,

Thanks for the reply.

Is it possible to get dew point at AGL heights included in the update
of
WPPV3.2...?  As of now, we have U, V, and T.

Thanks.



From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/10/2011 05:23 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

I see differences in the following files:
  v3.2/WPPV3/sorc/wrfpost/AGLVL.f # NEW
  v3.2/WPPV3/sorc/wrfpost/MISCLN.f
  v3.2/WPPV3/sorc/wrfpost/POSTDATA.f
  v3.2/WPPV3/sorc/wrfpost/makefile
  v3.2/WPPV3/parm/wrf_cntrl.parm

Not sure if that's all of the differences or not.

John

On 06/10/2011 03:12 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> I see sorc/wrfpost/AGLVL.f was modified...any other modifications I
should
> know about?
>
> Thanks.
>
>
>
> From:		 "RAL HelpDesk {for John Halley Gotway}"
<met_help at ucar.edu>
> To:		 James P Cipriani/Watson/IBM at IBMUS
> Date:		 06/10/2011 04:33 PM
> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> OK, that capability is not available in the currently released
version of
> WPPv3.2.  It's still an open question of whether or not it's
available in
> the initial release of UPP - the unified post
> processor - which replaces UPP.  The WRF person I'm working with
will
check
> next week.
>
> I had made some changes to WPPv3.2 to allow for the dumping of winds
and
> temperature at various AGL levels.  And I'm happy to send you that
code.
> However, I do so with a warning.  The changes are
> basically an adaptation of the code that was used to dump out TMP,
U, and
V
> at flight data levels.  It isn't clear whether or not it is
appropriate
to
> apply that same algorithm to levels much closer
> to the ground.  From a software standpoint, it gives you reasonable
data,
> but I can't tell you how correct it is from a scientific standpoint.
So
> use it at your own risk.
>
> The output of these records is controlled by the namelist settings
for
> (wrf_cntrl.parm):
>    "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"
>
> And these settings give you AGL output at 20m through 200m every 10
meters,
> and then 200m through 1000m every 50 meters.
>
> You can retrieve the modified version of WPP here:
>
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz
>
> Hope that helps.
>
> John
>
> On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> John,
>>
>> Thanks for the reply.
>>
>> I do believe that outputting fields such as temperature, dew point,
wind
>> speed at AGL heights (in meters) is not possible with WPPV3 (at
least
not
>> with the default code compile).  As far as I know, the only options
are
> to
>> output these variables on pressure levels (right now, there are 47
levels
>> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>>
>> That was exactly the reason for my last question (about outputting
U and
> V
>> at additional 'levels').  So you have responded exactly how I
expected.
>>
>> Assuming we can get these values at specific heights (m), your
settings
> are
>> exactly what I was thinking as well.  Using the MET package has
been
very
>> useful and easy to adapt.
>>
>> Please let me know what you hear from the WPP group...
>>
>> Thanks.
>>
>>
>>
>> From:		 		  "RAL HelpDesk {for John Halley Gotway}"
> <met_help at ucar.edu>
>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>> Date:		 		  06/10/2011 12:15 PM
>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question on
Interpolation
>>
>>
>>
>> James,
>>
>> This is John Halley Gotway.  I've been working with Paul on the
issues
>> you're having.
>>
>> It sounds like you'd like to verify wind speed and temperature at
various
>> heights above ground level.  And you have point observations for
those
>> heights.  We've helped some other wind energy folks in
>> the past do exactly what you're requesting.  As you list in (2),
the
most
>> important thing is that you configure WPP to output U, V, and
temperature
>> at additional AGL heights.  So the question is, how
>> do you configure WPP to do this?  I actually worked on some of the
WPP
> code
>> in the past to add in this sort of functionality, but I don't know
what,
> if
>> any of it, was released.
>>
>> I've sent an email to the WPP support person here asking for detail
about
>> this.  I'll let you know what I hear back.
>>
>> Supposing you are able to get output at other AGL heights (like
10m,
>> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
>> fcst_field[] = [ "WIND/Z0-50" ];
>> message_type[] = [ "ADPUPA" ];
>>
>> Suppose you have an observation of wind speed occurring at 15m.
> Point-Stat
>> will look at the model output above/below that - so at 10m and 20m.
> It'll
>> horizontally interpolate to the observation
>> location at 10m and 20m, and then vertically interpolate those two
values
>> to 15m, the height of the observation.
>>
>> John
>>
>> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>> MET team,
>>>
>>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>>
>>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
>> per
>>> your previous request (thank you, by the way).
>>> (2) The raw model data is not in GRIB1 so I run it through the WRF
Post
>>> Processor (WPPV3) - soon to be the Unified Post Processor.
However,
the
>>> only wind output that I specify from WPPV3 is U10 and V10, which
MET
> uses
>>> to calculate total wind speed on the fly - (u_10^2 +
v_10^2)^(1/2).
>> Thus,
>>> when specifying that I want U10 and V10, I am only outputting the
speeds
>> at
>>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>>> Should I be altering my WPPV3 control files to specify U and V
winds at
>>> additional levels?
>>> (3) I can definitely tell you that I have stations with sensor
> elevations
>>> (both temperature and wind speed) are < 10 meters and some with
sensor
>>> elevations > 20 meters.
>>>
>>> Any additional information and/or thoughts you have would be
greatly
>>> appreciated.
>>>
>>> Thanks.
>>>
>>>
>>>
>>> From:		 		  		 		   "RAL HelpDesk {for
Paul Oldenburg}"
> <met_help at ucar.edu>
>>> To:		 		  		 		   James P
Cipriani/Watson/IBM at IBMUS
>>> Date:		 		  		 		   06/09/2011 02:46
PM
>>> Subject:		 		  		 		   Re:
[rt.rap.ucar.edu #47360] Question on
> Interpolation
>>>
>>>
>>>
>>> James,
>>>
>>> We have discussed your situation, and the answer to your problem
depends
>> on
>>> your model data.  Is your model data in GRIB
>>> format?  If so, can you please send us the wgrib output from one
of
your
>>> model files.  If you have observations in a
>>> particular range of elevations or levels, say 8-12m, and you have
model
>>> levels at 10m, 20m and 30m, then the following
>>> fcst_field will definitely match observations and perform vertical
>>> interpolation for obs at 10m and higher:
>>>
>>> fcst_field[] = [ "WIND/Z10-20" ];
>>>
>>> We are not sure what the behavior will be for obs points below 10m
in
>> this
>>> case.  They may be matched without
>>> interpolation to the bottom model layer at 10m, but maybe not.
> Remember,
>>> it is important that your observation type is
>>> not a "surface" type.  One good type to use might be ADPUPA, which
will
>>> trigger vertical interpolation between model
>>> layers in MET.  We have not encountered this type of situation
before.
>>> Please let us know what you find.
>>>
>>> Paul
>>>
>>>
>>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>>> James,
>>>>
>>>> A surface observation for temperature is 2m above the surface,
which
> has
>>> an absolute elevation of 2m + sea level
>>>> elevation of the observation station location.  This is obviously
>> terrain
>>> dependent.  If you are using the GRIB output
>>>> of WPP as your model data, for example, and it contains surface
>>> temperature, then it will have a 2m TMP forecast value.
>>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
>> or
>>> SFCSHP) observation in point_stat.
>>>>
>>>> On the other hand, if you do not have a "surface" observation,
then
the
>>> model data will be interpolated using the two
>>>> adjacent model levels to the elevation of the observation.  This
logic
>>> holds for all forecast fields, regardless of
>>>> their defined "surface" elevation.
>>>>
>>>> Unfortunately, the individual most qualified to discuss this
matter
> with
>>> you is on vacation until Thursday.  I will talk
>>>> about this issue with him when he returns.  I hope I have
provided
some
>>> clarification.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>
>>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>>
>>>>>
>>>>> Paul,
>>>>>
>>>>> Thanks for the timely reply.  A bit of clarification:
>>>>>
>>>>> (1) When using ASCII2NC, I'm aware that there must be 10 columns
in
>> each
>>>>> text file - two of the columns are station elevation (msl) and
>>> observation
>>>>> (msl).  Let's use temperature as an example here and say the
station
>> was
>>>>> located 60 meters above sea level...if the observation was truly
a
>>> 2-meter
>>>>> temperature, then the difference between station elevation (60
meters)
>>> and
>>>>> observation elevation would be 2-meters, thus making the
observation
>>>>> elevation 62 meters...correct?  If the difference was > or < 2
meters,
>>> then
>>>>> vertical interpolation would be done (forecasts to observation
values)
>>> for
>>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>>> rather
>>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
>> variable
>>>>> (say 10-meter wind speed)....correct?
>>>>>
>>>>> Thanks,
>>>>>
>>>>>
>>>>>
>>>>> From:
  		 		    "RAL HelpDesk {for
> Paul Oldenburg}"
>> <met_help at ucar.edu>
>>>>> To:
  		 		    James P
> Cipriani/Watson/IBM at IBMUS
>>>>> Date:
  		 		    06/07/2011 02:40
> PM
>>>>> Subject:
	  		 		    Re:
> [rt.rap.ucar.edu #47360] Question
>> on Interpolation
>>>>>
>>>>>
>>>>>
>>>>> James,
>>>>>
>>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation
is
>>>>> done, since these are defined as "surface"
>>>>> observations.  If, however, your observations are not of that
type,
> MET
>>>>> will perform interpolation of model data to
>>>>> whatever the elevation of the observation is.  For more
information,
>> see
>>>>> section 4.2.1 of the MET User's Guide.
>>>>>
>>>>> If you want interpolation to be performed, I think you just need
to
> set
>>> the
>>>>> observation type to something besides ADPSFC
>>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>>>> Message_Type in the first column of your input.
>>>>>
>>>>> Paul
>>>>>
>>>>>
>>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>>
>>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>>        Queue: met_help
>>>>>>      Subject: Question on Interpolation
>>>>>>        Owner: Nobody
>>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>>       Status: new
>>>>>>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> MET Team,
>>>>>>
>>>>>> A quick question on verification at certain levels.  As you
know,
>>> sensors
>>>>>> at point sites are not always placed at 10-meters (wind) and
2-meters
>>>>>> (temperature) - they are often times placed above and below
these
>>> points.
>>>>>> Given that the user provides observation information (site
elevation
>>> AGL
>>>>>> and observation elevation AGL) before running ASCII2NC, does
the MET
>>>>>> package do the interpolation (for the observations) to 10-m
(wind)
> and
>>>>> 2-m
>>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10 forecast
vs.
>>> obs?
>>>>>>
>>>>>> Thanks.
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>>
>>
>>
>
>


------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: John Halley Gotway
Time: Fri Jun 10 15:51:46 2011

James,

Sorry, I don't have time or funding to work on extending this.  I'm
sending it to you as-is, but you're more than welcome to extend to the
functionality.  This mimics what is done for flight data
levels which only includes output for "TEMP/U/V AT FD HEIGHTS" in
wrf_cntrl.parm.  So I added corresponding entries to wrf_cntrl.parm
for "TEMP/U/V AT AGL".  If other fields were dumped for flight
data levels, I would have mimicked them in the AGL output.

Since we're in the process of replacing WPP with UPP, this
functionality really needs to be added to UPP.  Like I said, I haven't
heard back regarding the status of AGL level output for UPP.  But I've
already told them that if it's not yet in there, it should be added.

Thanks,
John

On 06/10/2011 03:39 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> John,
>
> Thanks for the reply.
>
> Is it possible to get dew point at AGL heights included in the
update of
> WPPV3.2...?  As of now, we have U, V, and T.
>
> Thanks.
>
>
>
> From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
> To:	James P Cipriani/Watson/IBM at IBMUS
> Date:	06/10/2011 05:23 PM
> Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> I see differences in the following files:
>   v3.2/WPPV3/sorc/wrfpost/AGLVL.f # NEW
>   v3.2/WPPV3/sorc/wrfpost/MISCLN.f
>   v3.2/WPPV3/sorc/wrfpost/POSTDATA.f
>   v3.2/WPPV3/sorc/wrfpost/makefile
>   v3.2/WPPV3/parm/wrf_cntrl.parm
>
> Not sure if that's all of the differences or not.
>
> John
>
> On 06/10/2011 03:12 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> I see sorc/wrfpost/AGLVL.f was modified...any other modifications I
> should
>> know about?
>>
>> Thanks.
>>
>>
>>
>> From:		 "RAL HelpDesk {for John Halley Gotway}"
> <met_help at ucar.edu>
>> To:		 James P Cipriani/Watson/IBM at IBMUS
>> Date:		 06/10/2011 04:33 PM
>> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>>
>>
>>
>> James,
>>
>> OK, that capability is not available in the currently released
version of
>> WPPv3.2.  It's still an open question of whether or not it's
available in
>> the initial release of UPP - the unified post
>> processor - which replaces UPP.  The WRF person I'm working with
will
> check
>> next week.
>>
>> I had made some changes to WPPv3.2 to allow for the dumping of
winds and
>> temperature at various AGL levels.  And I'm happy to send you that
code.
>> However, I do so with a warning.  The changes are
>> basically an adaptation of the code that was used to dump out TMP,
U, and
> V
>> at flight data levels.  It isn't clear whether or not it is
appropriate
> to
>> apply that same algorithm to levels much closer
>> to the ground.  From a software standpoint, it gives you reasonable
data,
>> but I can't tell you how correct it is from a scientific
standpoint.  So
>> use it at your own risk.
>>
>> The output of these records is controlled by the namelist settings
for
>> (wrf_cntrl.parm):
>>    "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"
>>
>> And these settings give you AGL output at 20m through 200m every 10
> meters,
>> and then 200m through 1000m every 50 meters.
>>
>> You can retrieve the modified version of WPP here:
>>
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz
>>
>> Hope that helps.
>>
>> John
>>
>> On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>> John,
>>>
>>> Thanks for the reply.
>>>
>>> I do believe that outputting fields such as temperature, dew
point, wind
>>> speed at AGL heights (in meters) is not possible with WPPV3 (at
least
> not
>>> with the default code compile).  As far as I know, the only
options are
>> to
>>> output these variables on pressure levels (right now, there are 47
> levels
>>> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>>>
>>> That was exactly the reason for my last question (about outputting
U and
>> V
>>> at additional 'levels').  So you have responded exactly how I
expected.
>>>
>>> Assuming we can get these values at specific heights (m), your
settings
>> are
>>> exactly what I was thinking as well.  Using the MET package has
been
> very
>>> useful and easy to adapt.
>>>
>>> Please let me know what you hear from the WPP group...
>>>
>>> Thanks.
>>>
>>>
>>>
>>> From:		 		  "RAL HelpDesk {for John Halley Gotway}"
>> <met_help at ucar.edu>
>>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>>> Date:		 		  06/10/2011 12:15 PM
>>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question on
> Interpolation
>>>
>>>
>>>
>>> James,
>>>
>>> This is John Halley Gotway.  I've been working with Paul on the
issues
>>> you're having.
>>>
>>> It sounds like you'd like to verify wind speed and temperature at
> various
>>> heights above ground level.  And you have point observations for
those
>>> heights.  We've helped some other wind energy folks in
>>> the past do exactly what you're requesting.  As you list in (2),
the
> most
>>> important thing is that you configure WPP to output U, V, and
> temperature
>>> at additional AGL heights.  So the question is, how
>>> do you configure WPP to do this?  I actually worked on some of the
WPP
>> code
>>> in the past to add in this sort of functionality, but I don't know
what,
>> if
>>> any of it, was released.
>>>
>>> I've sent an email to the WPP support person here asking for
detail
> about
>>> this.  I'll let you know what I hear back.
>>>
>>> Supposing you are able to get output at other AGL heights (like
10m,
>>> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
>>> fcst_field[] = [ "WIND/Z0-50" ];
>>> message_type[] = [ "ADPUPA" ];
>>>
>>> Suppose you have an observation of wind speed occurring at 15m.
>> Point-Stat
>>> will look at the model output above/below that - so at 10m and
20m.
>> It'll
>>> horizontally interpolate to the observation
>>> location at 10m and 20m, and then vertically interpolate those two
> values
>>> to 15m, the height of the observation.
>>>
>>> John
>>>
>>> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>
>>>>
>>>> MET team,
>>>>
>>>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>>>
>>>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
>>> per
>>>> your previous request (thank you, by the way).
>>>> (2) The raw model data is not in GRIB1 so I run it through the
WRF Post
>>>> Processor (WPPV3) - soon to be the Unified Post Processor.
However,
> the
>>>> only wind output that I specify from WPPV3 is U10 and V10, which
MET
>> uses
>>>> to calculate total wind speed on the fly - (u_10^2 +
v_10^2)^(1/2).
>>> Thus,
>>>> when specifying that I want U10 and V10, I am only outputting the
> speeds
>>> at
>>>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>>>> Should I be altering my WPPV3 control files to specify U and V
winds at
>>>> additional levels?
>>>> (3) I can definitely tell you that I have stations with sensor
>> elevations
>>>> (both temperature and wind speed) are < 10 meters and some with
sensor
>>>> elevations > 20 meters.
>>>>
>>>> Any additional information and/or thoughts you have would be
greatly
>>>> appreciated.
>>>>
>>>> Thanks.
>>>>
>>>>
>>>>
>>>> From:		 		  		 		   "RAL HelpDesk {for
> Paul Oldenburg}"
>> <met_help at ucar.edu>
>>>> To:		 		  		 		   James P
> Cipriani/Watson/IBM at IBMUS
>>>> Date:		 		  		 		   06/09/2011 02:46
> PM
>>>> Subject:		 		  		 		   Re:
> [rt.rap.ucar.edu #47360] Question on
>> Interpolation
>>>>
>>>>
>>>>
>>>> James,
>>>>
>>>> We have discussed your situation, and the answer to your problem
> depends
>>> on
>>>> your model data.  Is your model data in GRIB
>>>> format?  If so, can you please send us the wgrib output from one
of
> your
>>>> model files.  If you have observations in a
>>>> particular range of elevations or levels, say 8-12m, and you have
model
>>>> levels at 10m, 20m and 30m, then the following
>>>> fcst_field will definitely match observations and perform
vertical
>>>> interpolation for obs at 10m and higher:
>>>>
>>>> fcst_field[] = [ "WIND/Z10-20" ];
>>>>
>>>> We are not sure what the behavior will be for obs points below
10m in
>>> this
>>>> case.  They may be matched without
>>>> interpolation to the bottom model layer at 10m, but maybe not.
>> Remember,
>>>> it is important that your observation type is
>>>> not a "surface" type.  One good type to use might be ADPUPA,
which will
>>>> trigger vertical interpolation between model
>>>> layers in MET.  We have not encountered this type of situation
before.
>>>> Please let us know what you find.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>>>> James,
>>>>>
>>>>> A surface observation for temperature is 2m above the surface,
which
>> has
>>>> an absolute elevation of 2m + sea level
>>>>> elevation of the observation station location.  This is
obviously
>>> terrain
>>>> dependent.  If you are using the GRIB output
>>>>> of WPP as your model data, for example, and it contains surface
>>>> temperature, then it will have a 2m TMP forecast value.
>>>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
>>> or
>>>> SFCSHP) observation in point_stat.
>>>>>
>>>>> On the other hand, if you do not have a "surface" observation,
then
> the
>>>> model data will be interpolated using the two
>>>>> adjacent model levels to the elevation of the observation.  This
logic
>>>> holds for all forecast fields, regardless of
>>>>> their defined "surface" elevation.
>>>>>
>>>>> Unfortunately, the individual most qualified to discuss this
matter
>> with
>>>> you is on vacation until Thursday.  I will talk
>>>>> about this issue with him when he returns.  I hope I have
provided
> some
>>>> clarification.
>>>>>
>>>>> Paul
>>>>>
>>>>>
>>>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>>
>>>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>>>
>>>>>>
>>>>>> Paul,
>>>>>>
>>>>>> Thanks for the timely reply.  A bit of clarification:
>>>>>>
>>>>>> (1) When using ASCII2NC, I'm aware that there must be 10
columns in
>>> each
>>>>>> text file - two of the columns are station elevation (msl) and
>>>> observation
>>>>>> (msl).  Let's use temperature as an example here and say the
station
>>> was
>>>>>> located 60 meters above sea level...if the observation was
truly a
>>>> 2-meter
>>>>>> temperature, then the difference between station elevation (60
> meters)
>>>> and
>>>>>> observation elevation would be 2-meters, thus making the
observation
>>>>>> elevation 62 meters...correct?  If the difference was > or < 2
> meters,
>>>> then
>>>>>> vertical interpolation would be done (forecasts to observation
> values)
>>>> for
>>>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>>>> rather
>>>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
>>> variable
>>>>>> (say 10-meter wind speed)....correct?
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>>
>>>>>>
>>>>>> From:
>   		 		    "RAL HelpDesk {for
>> Paul Oldenburg}"
>>> <met_help at ucar.edu>
>>>>>> To:
>   		 		    James P
>> Cipriani/Watson/IBM at IBMUS
>>>>>> Date:
>   		 		    06/07/2011 02:40
>> PM
>>>>>> Subject:
> 	  		 		    Re:
>> [rt.rap.ucar.edu #47360] Question
>>> on Interpolation
>>>>>>
>>>>>>
>>>>>>
>>>>>> James,
>>>>>>
>>>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation
> is
>>>>>> done, since these are defined as "surface"
>>>>>> observations.  If, however, your observations are not of that
type,
>> MET
>>>>>> will perform interpolation of model data to
>>>>>> whatever the elevation of the observation is.  For more
information,
>>> see
>>>>>> section 4.2.1 of the MET User's Guide.
>>>>>>
>>>>>> If you want interpolation to be performed, I think you just
need to
>> set
>>>> the
>>>>>> observation type to something besides ADPSFC
>>>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify a
>>>>>> Message_Type in the first column of your input.
>>>>>>
>>>>>> Paul
>>>>>>
>>>>>>
>>>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>>>
>>>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>>>        Queue: met_help
>>>>>>>      Subject: Question on Interpolation
>>>>>>>        Owner: Nobody
>>>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>>>       Status: new
>>>>>>>  Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> MET Team,
>>>>>>>
>>>>>>> A quick question on verification at certain levels.  As you
know,
>>>> sensors
>>>>>>> at point sites are not always placed at 10-meters (wind) and
> 2-meters
>>>>>>> (temperature) - they are often times placed above and below
these
>>>> points.
>>>>>>> Given that the user provides observation information (site
elevation
>>>> AGL
>>>>>>> and observation elevation AGL) before running ASCII2NC, does
the MET
>>>>>>> package do the interpolation (for the observations) to 10-m
(wind)
>> and
>>>>>> 2-m
>>>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10
forecast vs.
>>>> obs?
>>>>>>>
>>>>>>> Thanks.
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>

------------------------------------------------
Subject: Question on Interpolation 
From: James P Cipriani
Time: Fri Jun 10 15:56:59 2011


John,

Sure..I am in the process of doing this already.  I just wondering if
you
had any additional suggestions.

I appreciate your help to this point, and I will update as soon as I
run
some tests with WPPV3/MET.

This capability should definitely be included in the UPPv1 (it is not
as of
the beta version).

Thanks again for everything.



From:	"RAL HelpDesk {for John Halley Gotway}" <met_help at ucar.edu>
To:	James P Cipriani/Watson/IBM at IBMUS
Date:	06/10/2011 05:51 PM
Subject:	Re: [rt.rap.ucar.edu #47360] Question on Interpolation



James,

Sorry, I don't have time or funding to work on extending this.  I'm
sending
it to you as-is, but you're more than welcome to extend to the
functionality.  This mimics what is done for flight data
levels which only includes output for "TEMP/U/V AT FD HEIGHTS" in
wrf_cntrl.parm.  So I added corresponding entries to wrf_cntrl.parm
for
"TEMP/U/V AT AGL".  If other fields were dumped for flight
data levels, I would have mimicked them in the AGL output.

Since we're in the process of replacing WPP with UPP, this
functionality
really needs to be added to UPP.  Like I said, I haven't heard back
regarding the status of AGL level output for UPP.  But I've
already told them that if it's not yet in there, it should be added.

Thanks,
John

On 06/10/2011 03:39 PM, RAL HelpDesk {for James P Cipriani} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>
>
> John,
>
> Thanks for the reply.
>
> Is it possible to get dew point at AGL heights included in the
update of
> WPPV3.2...?  As of now, we have U, V, and T.
>
> Thanks.
>
>
>
> From:		 "RAL HelpDesk {for John Halley Gotway}"
<met_help at ucar.edu>
> To:		 James P Cipriani/Watson/IBM at IBMUS
> Date:		 06/10/2011 05:23 PM
> Subject:		 Re: [rt.rap.ucar.edu #47360] Question on Interpolation
>
>
>
> James,
>
> I see differences in the following files:
>   v3.2/WPPV3/sorc/wrfpost/AGLVL.f # NEW
>   v3.2/WPPV3/sorc/wrfpost/MISCLN.f
>   v3.2/WPPV3/sorc/wrfpost/POSTDATA.f
>   v3.2/WPPV3/sorc/wrfpost/makefile
>   v3.2/WPPV3/parm/wrf_cntrl.parm
>
> Not sure if that's all of the differences or not.
>
> John
>
> On 06/10/2011 03:12 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>
>>
>> I see sorc/wrfpost/AGLVL.f was modified...any other modifications I
> should
>> know about?
>>
>> Thanks.
>>
>>
>>
>> From:		 		  "RAL HelpDesk {for John Halley Gotway}"
> <met_help at ucar.edu>
>> To:		 		  James P Cipriani/Watson/IBM at IBMUS
>> Date:		 		  06/10/2011 04:33 PM
>> Subject:		 		  Re: [rt.rap.ucar.edu #47360] Question on
Interpolation
>>
>>
>>
>> James,
>>
>> OK, that capability is not available in the currently released
version
of
>> WPPv3.2.  It's still an open question of whether or not it's
available
in
>> the initial release of UPP - the unified post
>> processor - which replaces UPP.  The WRF person I'm working with
will
> check
>> next week.
>>
>> I had made some changes to WPPv3.2 to allow for the dumping of
winds and
>> temperature at various AGL levels.  And I'm happy to send you that
code.
>> However, I do so with a warning.  The changes are
>> basically an adaptation of the code that was used to dump out TMP,
U,
and
> V
>> at flight data levels.  It isn't clear whether or not it is
appropriate
> to
>> apply that same algorithm to levels much closer
>> to the ground.  From a software standpoint, it gives you reasonable
data,
>> but I can't tell you how correct it is from a scientific
standpoint.  So
>> use it at your own risk.
>>
>> The output of these records is controlled by the namelist settings
for
>> (wrf_cntrl.parm):
>>    "TEMP AT AGL", "U WIND AT AGL", and "V WIND AT AGL"
>>
>> And these settings give you AGL output at 20m through 200m every 10
> meters,
>> and then 200m through 1000m every 50 meters.
>>
>> You can retrieve the modified version of WPP here:
>>
ftp://ftp.rap.ucar.edu/incoming/irap/met_help/WPPV3_AGL_WINDS.tar.gz
>>
>> Hope that helps.
>>
>> John
>>
>> On 06/10/2011 11:38 AM, RAL HelpDesk {for James P Cipriani} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>
>>>
>>> John,
>>>
>>> Thanks for the reply.
>>>
>>> I do believe that outputting fields such as temperature, dew
point,
wind
>>> speed at AGL heights (in meters) is not possible with WPPV3 (at
least
> not
>>> with the default code compile).  As far as I know, the only
options are
>> to
>>> output these variables on pressure levels (right now, there are 47
> levels
>>> possible, from 2 to 1013 mb) or on model levels (hybrid levels).
>>>
>>> That was exactly the reason for my last question (about outputting
U
and
>> V
>>> at additional 'levels').  So you have responded exactly how I
expected.
>>>
>>> Assuming we can get these values at specific heights (m), your
settings
>> are
>>> exactly what I was thinking as well.  Using the MET package has
been
> very
>>> useful and easy to adapt.
>>>
>>> Please let me know what you hear from the WPP group...
>>>
>>> Thanks.
>>>
>>>
>>>
>>> From:		 		  		 		   "RAL HelpDesk {for
John Halley Gotway}"
>> <met_help at ucar.edu>
>>> To:		 		  		 		   James P
Cipriani/Watson/IBM at IBMUS
>>> Date:		 		  		 		   06/10/2011 12:15
PM
>>> Subject:		 		  		 		   Re:
[rt.rap.ucar.edu #47360] Question on
> Interpolation
>>>
>>>
>>>
>>> James,
>>>
>>> This is John Halley Gotway.  I've been working with Paul on the
issues
>>> you're having.
>>>
>>> It sounds like you'd like to verify wind speed and temperature at
> various
>>> heights above ground level.  And you have point observations for
those
>>> heights.  We've helped some other wind energy folks in
>>> the past do exactly what you're requesting.  As you list in (2),
the
> most
>>> important thing is that you configure WPP to output U, V, and
> temperature
>>> at additional AGL heights.  So the question is, how
>>> do you configure WPP to do this?  I actually worked on some of the
WPP
>> code
>>> in the past to add in this sort of functionality, but I don't know
what,
>> if
>>> any of it, was released.
>>>
>>> I've sent an email to the WPP support person here asking for
detail
> about
>>> this.  I'll let you know what I hear back.
>>>
>>> Supposing you are able to get output at other AGL heights (like
10m,
>>> 20m, ..., 50m), I'd suggest using the following settings in Point-
Stat:
>>> fcst_field[] = [ "WIND/Z0-50" ];
>>> message_type[] = [ "ADPUPA" ];
>>>
>>> Suppose you have an observation of wind speed occurring at 15m.
>> Point-Stat
>>> will look at the model output above/below that - so at 10m and
20m.
>> It'll
>>> horizontally interpolate to the observation
>>> location at 10m and 20m, and then vertically interpolate those two
> values
>>> to 15m, the height of the observation.
>>>
>>> John
>>>
>>> On 06/09/2011 01:34 PM, RAL HelpDesk {for James P Cipriani} wrote:
>>>>
>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>
>>>>
>>>> MET team,
>>>>
>>>> I appreciate your thoughts in this matter.  A few
comments/questions:
>>>>
>>>> (1) I have indeed changed my message type to be "ADPUPA" going
forward,
>>> per
>>>> your previous request (thank you, by the way).
>>>> (2) The raw model data is not in GRIB1 so I run it through the
WRF
Post
>>>> Processor (WPPV3) - soon to be the Unified Post Processor.
However,
> the
>>>> only wind output that I specify from WPPV3 is U10 and V10, which
MET
>> uses
>>>> to calculate total wind speed on the fly - (u_10^2 +
v_10^2)^(1/2).
>>> Thus,
>>>> when specifying that I want U10 and V10, I am only outputting the
> speeds
>>> at
>>>> 10 meters.  So I am not outputting winds at 20 meters, 30 meters,
etc.
>>>> Should I be altering my WPPV3 control files to specify U and V
winds
at
>>>> additional levels?
>>>> (3) I can definitely tell you that I have stations with sensor
>> elevations
>>>> (both temperature and wind speed) are < 10 meters and some with
sensor
>>>> elevations > 20 meters.
>>>>
>>>> Any additional information and/or thoughts you have would be
greatly
>>>> appreciated.
>>>>
>>>> Thanks.
>>>>
>>>>
>>>>
>>>> From:
  		 		    "RAL HelpDesk {for
> Paul Oldenburg}"
>> <met_help at ucar.edu>
>>>> To:
  		 		    James P
> Cipriani/Watson/IBM at IBMUS
>>>> Date:
  		 		    06/09/2011 02:46
> PM
>>>> Subject:
	  		 		    Re:
> [rt.rap.ucar.edu #47360] Question on
>> Interpolation
>>>>
>>>>
>>>>
>>>> James,
>>>>
>>>> We have discussed your situation, and the answer to your problem
> depends
>>> on
>>>> your model data.  Is your model data in GRIB
>>>> format?  If so, can you please send us the wgrib output from one
of
> your
>>>> model files.  If you have observations in a
>>>> particular range of elevations or levels, say 8-12m, and you have
model
>>>> levels at 10m, 20m and 30m, then the following
>>>> fcst_field will definitely match observations and perform
vertical
>>>> interpolation for obs at 10m and higher:
>>>>
>>>> fcst_field[] = [ "WIND/Z10-20" ];
>>>>
>>>> We are not sure what the behavior will be for obs points below
10m in
>>> this
>>>> case.  They may be matched without
>>>> interpolation to the bottom model layer at 10m, but maybe not.
>> Remember,
>>>> it is important that your observation type is
>>>> not a "surface" type.  One good type to use might be ADPUPA,
which
will
>>>> trigger vertical interpolation between model
>>>> layers in MET.  We have not encountered this type of situation
before.
>>>> Please let us know what you find.
>>>>
>>>> Paul
>>>>
>>>>
>>>> On 06/07/2011 02:05 PM, Paul Oldenburg wrote:
>>>>> James,
>>>>>
>>>>> A surface observation for temperature is 2m above the surface,
which
>> has
>>>> an absolute elevation of 2m + sea level
>>>>> elevation of the observation station location.  This is
obviously
>>> terrain
>>>> dependent.  If you are using the GRIB output
>>>>> of WPP as your model data, for example, and it contains surface
>>>> temperature, then it will have a 2m TMP forecast value.
>>>>>  This will not require interpolation to match a "surface" (i.e.
ADPSFC
>>> or
>>>> SFCSHP) observation in point_stat.
>>>>>
>>>>> On the other hand, if you do not have a "surface" observation,
then
> the
>>>> model data will be interpolated using the two
>>>>> adjacent model levels to the elevation of the observation.  This
logic
>>>> holds for all forecast fields, regardless of
>>>>> their defined "surface" elevation.
>>>>>
>>>>> Unfortunately, the individual most qualified to discuss this
matter
>> with
>>>> you is on vacation until Thursday.  I will talk
>>>>> about this issue with him when he returns.  I hope I have
provided
> some
>>>> clarification.
>>>>>
>>>>> Paul
>>>>>
>>>>>
>>>>> On 06/07/2011 01:44 PM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>>
>>>>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360 >
>>>>>>
>>>>>>
>>>>>> Paul,
>>>>>>
>>>>>> Thanks for the timely reply.  A bit of clarification:
>>>>>>
>>>>>> (1) When using ASCII2NC, I'm aware that there must be 10
columns in
>>> each
>>>>>> text file - two of the columns are station elevation (msl) and
>>>> observation
>>>>>> (msl).  Let's use temperature as an example here and say the
station
>>> was
>>>>>> located 60 meters above sea level...if the observation was
truly a
>>>> 2-meter
>>>>>> temperature, then the difference between station elevation (60
> meters)
>>>> and
>>>>>> observation elevation would be 2-meters, thus making the
observation
>>>>>> elevation 62 meters...correct?  If the difference was > or < 2
> meters,
>>>> then
>>>>>> vertical interpolation would be done (forecasts to observation
> values)
>>>> for
>>>>>> 2-meter temperature (assuming we choose a message type like
"ADPUPA"
>>>> rather
>>>>>> than "ADPSFC" or "SFCSHP").  Same would hold true for any other
>>> variable
>>>>>> (say 10-meter wind speed)....correct?
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>>
>>>>>>
>>>>>> From:
>   		 		  		 		     "RAL HelpDesk {for
>> Paul Oldenburg}"
>>> <met_help at ucar.edu>
>>>>>> To:
>   		 		  		 		     James P
>> Cipriani/Watson/IBM at IBMUS
>>>>>> Date:
>   		 		  		 		     06/07/2011 02:40
>> PM
>>>>>> Subject:
> 		   		 		  		 		     Re:
>> [rt.rap.ucar.edu #47360] Question
>>> on Interpolation
>>>>>>
>>>>>>
>>>>>>
>>>>>> James,
>>>>>>
>>>>>> If your observations are of type ADPSFC or SFCSHP, no
interpolation
> is
>>>>>> done, since these are defined as "surface"
>>>>>> observations.  If, however, your observations are not of that
type,
>> MET
>>>>>> will perform interpolation of model data to
>>>>>> whatever the elevation of the observation is.  For more
information,
>>> see
>>>>>> section 4.2.1 of the MET User's Guide.
>>>>>>
>>>>>> If you want interpolation to be performed, I think you just
need to
>> set
>>>> the
>>>>>> observation type to something besides ADPSFC
>>>>>> or SFCSHP.  If you are using ascii2nc, for example, you can
specify
a
>>>>>> Message_Type in the first column of your input.
>>>>>>
>>>>>> Paul
>>>>>>
>>>>>>
>>>>>> On 06/07/2011 08:56 AM, RAL HelpDesk {for James P Cipriani}
wrote:
>>>>>>>
>>>>>>> Tue Jun 07 08:56:31 2011: Request 47360 was acted upon.
>>>>>>> Transaction: Ticket created by jpcipria at us.ibm.com
>>>>>>>        Queue: met_help
>>>>>>>      Subject: Question on Interpolation
>>>>>>>        Owner: Nobody
>>>>>>>   Requestors: jpcipria at us.ibm.com
>>>>>>>       Status: new
>>>>>>>  Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47360
>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> MET Team,
>>>>>>>
>>>>>>> A quick question on verification at certain levels.  As you
know,
>>>> sensors
>>>>>>> at point sites are not always placed at 10-meters (wind) and
> 2-meters
>>>>>>> (temperature) - they are often times placed above and below
these
>>>> points.
>>>>>>> Given that the user provides observation information (site
elevation
>>>> AGL
>>>>>>> and observation elevation AGL) before running ASCII2NC, does
the
MET
>>>>>>> package do the interpolation (for the observations) to 10-m
(wind)
>> and
>>>>>> 2-m
>>>>>>> (temp) when running Point-Stat on TEMP/Z2 and WIND/Z10
forecast vs.
>>>> obs?
>>>>>>>
>>>>>>> Thanks.
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>


------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47360] Question on Interpolation
From: John Halley Gotway
Time: Fri Jun 10 15:58:17 2011

James,

OK, sounds good.  I'll go ahead and resolve this ticket.  Just let us
know if any more questions arise in your use of MET.

Thanks,
John

On 06/10/2011 03:57 PM, RAL HelpDesk {for James P Cipriani} wrote:
> Thanks again for everything.

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


More information about the Met_help mailing list