[wrfems] Impact of the NAM changes to the local modeling efforts
rozumal at ucar.edu
Thu Aug 4 13:16:23 MDT 2011
> A few questions relating to NAM upgrade impacts on local modeling
> efforts. First thanks for all the previous information.
> Now the questions...
> 1. For the time being you still plan on keeping the WRF-EMS
> infrastructure the same? In other words even though the WRF-NMM core
> will likely no longer be supported by NCEP, for the time being the NMM
> inside EMS will stay pretty much the same (example NMM runs in the EMS
> will still need the center point for all domains to be the same and
> will remain on an E-grid). What is the future of the WRF-NMM within
> the EMS and in the research community? (just looking for your thoughts
> here as this question goes beyond just what you are involved in)
Semi-official word from Geoff DiMego is that NCEP will attempt to
provide continued support for the WRF-NMM through
the Development Testbed Center (DTC). I think the operative word is
"attempt" as I don't really anticipate any improvements
to the code except for bug fixes and some minor enhancements. NCEP
really doesn't have the resources to support
> This goes along with the whole NEMS-NMM-B transition as well. Have you
> looked more into incorporating the NMM-B as a separate package within
> the EMS system?
Well, yes I have. I've had preliminary talks with Geoff and the DTC
about the NMM-B and expressed a desire to
integrate the code into the WRF EMS. I've already received parts of the
pre-processor but have not had the time
to begin the project. Hopefully I'll start this Fall.
> It seems NCEP and partners also plan on incorporating the ARW core
> into NEMS as well based on some of the documents I have read so I am
> unsure what the future of the WRF is in general especially once DTC
> starts to support the NEMS.
I suspect NCAR will continue to build and support the WRF infrastructure
while NCEP will go their own way
> 2. Change from USGS to MODIS land use dataset - Is the MODIS dataset
> currently available in EMS? I thought the only thing that could be
> changed with the "geo" terrestrial files was the resolution of the
> USGS dataset, but I am not sure.
Yes, simply check the MODIS box on the DW GUI or pass the --modis flag
> 3. Does the NMM core in WRF-EMS already use use a first layer
> correction for the shelter fields? What does the ARW core do?
Short answer is yes, or at least my verysion of the 1st layer
correction. With the ARW, it depends upon the PBL scheme used
as some provide 2m temps while others do not. Foe the most part the
shelter fields are derived from the 1st layer fields.
> 4. I am trying to better understand what release of the WRF-EMS all
> the little changes will be completed for. The next beta release? or
> the next official release? Speaking of which what is the current
> official release V3.1? V3.2 is still being beta tested I think.
The next release will be the official release of version 3.2.1.
Hopefully in October.
I've been working on a new Modsnd release, which I've completed but not
announced, and a new website, which is time consuming.
If anyone wants the new Modsnd, with SREF support, just email me.
> When you make some of the post-processing changes will they be applied
> to just output from the NMM core? For example with the new virtual
> CAPE calculation (which is going to be in place in all NCEP models)
> just be in the NMM core or in the ARW core as well? Same question for
> the reflectivity processing and other post-processing changes.
I will take any new code from the updated unified post-processor (UPP)
and integrated into the EMS version of the WPP.
> 5. One more thing is the availability of the NAM nest grids for use in
> initializing model runs. Where do u currently go to grab the grids so
> that offices can make the tiles from your server? It seems like a full
> set of variables will be available from the NWS ftp server and
> eventually NOMADS for the NAM nests, but u seemed to think that not
> enough information would be present.
My understanding is that you will only get the near-surface fields plus
a few others. Not enough to initialize a simulation.
> Thanks for taking the time to respond,
> Brian Miretzky
> ERH SSD
> On 6/7/2011 11:36 AM, Robert Rozumalski wrote:
>> Good morning All,
>> I have been asked to comment on the impact of the NAM changes to the
>> local modeling efforts
>> within the WFOs. As many of you already/should know, NCEP will have
>> a major change to the
>> NAM, the evaluation period for which should take effect next week.
>> From what I can tell, none
>> of the changes will directly impact current local modeling activities
>> with the WRF EMS; however,
>> I will have to make some changes to the next official release (V3.3)
>> to incorporate some of
>> the enhancements.
>> Below is the original technical announcement for the NAM update. I
>> have embedded comments
>> regarding the changes and how they relate to local modeling with the
>> EMS. Feel free to contact
>> me with questions or corrections.
>> So read my comments below - Bob
>> Subject:Changes to the North American Mesoscale (NAM) Analysisand
>> Forecast System: Effective July 12, 2011
>> Effective Tuesday, July 12, 2011, beginning with the 1200 Coordinated
>> Universal Time (UTC) run,
>> the National Centers for Environmental Prediction (NCEP) will modify
>> the North American Mesoscale
>> (NAM) Analysis and Forecast System.
>> These changes include:
>> -Introducing a new modeling framework
>> -Installing a major upgrade to the prediction model
>> -Modifying the data analysis and assimilation system
>> -Adding to existing NAM products
>> -Adding new higher-resolution nests within the NAM including
>> a very high-resolution but small domain to support fire weather
>> and Incident Meteorologist (IMET) interests.
>> Details on the various changes are provided below, along with a
>> notice about possible changes to product generation time.
>> Model Upgrades:
>> NCEP will introduce the use of the new NOAA Environmental Modeling
>> System (NEMS), which is based on the tenets put forth by the Earth
>> System Modeling Framework to which NOAA has subscribed. Eventually
>> all of NCEP's major modeling will be performed within NEMS. This NAM
>> upgrade represents the first implementation of NEMS and a major step
>> in the evolution of NCEP's modeling suite.
>> I will ask Bill Bua to comment on the NEMS since he is more familiar
>> with the framework
>> The prediction model used in the NAM run will go from being the
>> strictly regional Nonhydrostatic Mesoscale Model (NMM) to a new
>> extended capability version now known as Nonhydrostatic Multiscale
>> Model on B-grid (NMMB), which can be run either regionally or
>> globally, and with or without embedded nests.This NEMS-NMMB will also
>> serve as the prediction model running in the NAM Data Assimilation
>> System (NDAS). Model changes/ enhancements in the NEMS-NMMB will
>> a) Native horizontal grid is an Arakawa staggered B-grid (rotated 45
>> deg from the Arakawa E-grid in the WRF-NMM NAM)
>> This change alone is huge for model development as the "B" grid is a
>> more traditional model grid that is not rotated like the "E".
>> This should facilitate the integration of new model physics and
>> dynamics schemes into the "NMM-B".
>> Also, communication across grid patches (and thus CPUs) should be
>> faster resulting in greater model efficiency.
>> The change to the B-grid will eliminate the major navigation problem
>> with WRF-NMM core nested domains.
>> If you run a nested NMM core simulation, inner domains can not be
>> correctly projected unless the center point
>> of the nest is co-located with that of the parent, which means all
>> domains must have the same center point.
>> HOWEVER - It is my understanding that NCEP has NO plans on
>> integrating the NMM-B grid into WRF. I will contact
>> Geoff DiMego today regarding this issue but I don't think these plans
>> have changed.
>> If I can get the NMM-B as a separate package from NCEP then I will
>> attempt to integrate it into the EMS (but not the WRF).
>> This may take time and be a bit awkward but it can be done.
>> The switch to the B grid should not have an operational impact within
>> the WFOs through the introduction of
>> model biases but more will be known after the evaluation period.
>> b) New more conservative Eulerian advection scheme for passive
>> tracers like water vapor condensate fields
>> The preliminary results of which have shown a reduction in forecast
>> c) Generalized vertical coordinate
>> No more hybrid Sigma/Pressure coordinate, which is in the current NMM.
>> d) Modified vertical level distribution with more layers in the
>> stratosphere (14 layers above 200 mb instead of 7 in the current
>> operational NAM)
>> In theory, this should improve the forecast of all fields near the
>> tropopause as well as reduce the amount
>> of reflection off the model top.
>> e) Boundary condition treatment changed to specify 5 boundary rows
>> instead of one.
>> This should reduce the amount of noise along the lateral boundaries
>> in the NMM. The 5 rows is more
>> in-line with conventional treatments such as that in the WRF-ARW.
>> f) Microphysics changes to produce higher peak reflectivities above
>> 45 dBZ with higher peak surface rainfall rates in the high-resolution
>> nests and more realistic grid-scale cloud fractions from cold, high
>> These fields are part of the NAM post processing and will be part of
>> the wrfpost for the next EMS release. I will
>> have get the changes from NCEP to add to the code.
>> g) Change from USGS to MODIS_IGBP land-use definitions
>> Oddly, this may have the biggest immediate impact on local modeling
>> activities. WRF EMS users may
>> need to re-localize their domains initialized with the operational
>> NAM to use the MODIS land-use fields.
>> More investigation is needed on my part
>> h) Enhanced diffusion for specific humidity and cloud water
>> i) Run 5 high-resolution nested domains inside the 12km NAM every
>> cycle.These nests will run with greatly reduced convective
>> triggering, which improved quantitative precipitation forecast (QPF)
>> bias compared to using explicit convection.
>> I want to see the results of the evaluation from the 4km nests. The
>> modified BMJ scheme used in these simulations will
>> be less sensitive and do less work than the traditional BMJ sheme,
>> thus allowing the microphysics scheme to generate
>> more precipitation. The use of a CU scheme in high resolution runs
>> has shown to improve the timing of convection but
>> results in a less "realistic looking" precipitation field.
>> NDAS Changes:
>> a) Initial first guess at T-12hr will reflect relocation of tropical
>> b) Will use 1/12 degree high resolution real-time sea surface
>> temperature (RTG_SST_HR) analysis instead of the 1/2th degree
>> real-time SST analysis.
>> c) Will update 2 m temperature and moisture and 10 m wind with
>> portion of first layer correction
>> Important for me as I occasionally get questions about the "quality"
>> of the 00-hour shelter fields coming out of the
>> NMM core in the EMS.
>> Changes to the NAM post-processing code:
>> a) The height and wind speed at the maximum wind level will be
>> computed by assuming the wind speed varies quadratically in height in
>> the neighborhood of the maximum wind level with the search being
>> capped at 100mb. The previous algorithm defines maximum wind level as
>> the level with the maximum wind speed among all the model levels.
>> Will be added to the next EMS release
>> b) The static Tropopause level will be obtained by finding the lowest
>> level that has a temperature lapse rate of less than 2 K/km over a
>> 2km depth above it.If no such level is found below 50 mb, the
>> Tropopause is set to 50 mb.The previous algorithm defines the
>> Tropopause by finding the lowest level that has mean temperature
>> lapse rate of 2 K/Km over three model layers.
>> Will be added to the next EMS release
>> c) All Convective Available Potential Energy (CAPE), Convective
>> Inhibition (CIN), and Lifted Index (LI) variables in the NAM output
>> will be computed using virtual temperature instead of sensible
>> temperature.See the NWS Public Information Statement issued January
>> 12, 2011 for more details on this change.
>> Will be added to the next EMS release
>> Users should be aware that this NAM upgrade will impact all
>> downstream models and systems that use the NAM or NDAS as input.
>> Output Product Changes:
>> New NAM Nest runs, including placeable Fire Weather Nest:
>> Output grids from the NAM nest runs will be available on the NCEP ftp
>> server immediately and on the NCEP NOMADS server in the future. The
>> fields contained in the NAM nest output grids are listed at:
>> Changes to parent NAM gridded output on the NCEP ftp server and the
>> NWS ftp server:
>> a) In all NAM grids that contain roughness length, the precision was
>> changed to provide better decimal scaling because many land-use types
>> have roughness length much less than 1.
>> I will make the necessary changes to the WRF code if appropriate
>> b) Surface slope type was dropped from all NAM output grids as it is
>> no longer used in the model land-surface physics
>> c) Time-averaged surface momentum flux record is being removed from
>> all NAM GRIB files as it is not defined and thus had been set to zero
>> in the current operational NAM
>> I'm sure you are all upset as these fields will be turned on in the EMS
>> d) New output fields:
>> - Clear Air Turbulence (Ellrod Index), every 25 mb from
>> 150-525 mb, and Inflight Icing every 25 mb from 300-600 mb,
>> every 50 mb from 650-950 mb, and 725 mb; added to grid
>> #221 (32 km grid over North America)
>> CAT IS currently available in the EMS
>> - 80 km above ground level pressure, temperature, wind, and
>> specific humidity; added to grid #221 (32 km grid over
>> North America), grid #218 (12 km grid over CONUS, pressure
>> file only) and grid #242 (11.25 km grid over Alaska,
>> pressure file only).
>> Also currently available in the EMS
>> - Hourly max and min surface fields (10-meter wind, 2-meter
>> temperature, 2-meter relative humidity);added to grid
>> #221 (32 km grid over North America), grid #218 (12 km
>> gridover CONUS, both versions) and grid #242 (11.25 km
>> grid over Alaska).
>> Also currently available in the EMS agonf with various max/min
>> reflectivity fields.
>> - Haines Index: Added to grid #218 (12 km grid over CONUS,
>> both versions) and grid #242 (11.25 km grid over Alaska,
>> both versions
>> Will be available in next release
>> - Ventilation Rate: Added to grid #218 (12 km grid over
>> CONUS) and grid #242 (11.25 km grid over Alaska).
>> - Rain and snow on lowest model level: added to grid #218
>> surface file "awip12"
>> Will be available in next release
>> - Thunder parameter (called lightning in GRIB): added to
>> both grid #218 files
>> Will be available in next release - already tested
>> e) All NAM output files on the native horizontal staggered
>> E-grid of the current operational WRF-NMM will be on the staggered
>> B-grid of the NEMS-NMMB. The current native grid files with names
>> "egrd3d," "edgawp," and "egrdsf" will be removed from the NCEP FTP
>> server and replaced with files with names "bgrd3d", "bgdawp," and
>> "bgrdsf," respectively. Users who process current NAM native GRIB
>> files with e-grid staggering will need to modify their software to
>> process b-grid staggering.
>> NO impact as EMS users do not use the "native" model grids
>> f)To provide the critical variables used at Weather Forecast Offices
>> to generate their National Digital Forecast Database (NDFD) forecast
>> grids, NCEP will begin to populate the current NAM Downscaled
>> Numerical Guidance (DNG --a small set of 2-dimensional fields) in the
>> 0--60 hr range from the new high-resolution NAM nested fields instead
>> of the parent NAM 12 km fields. The NAM DNG grids are currently
>> distributed to CONUS, Alaska, Hawaii and Puerto Rico over the AWIPS
>> SBN/NOAAPORT.In response to requests from the NWS Regions, NCEP also
>> plans on producing a CONUS 2.5km NAM-DNG for the 0-60 hr in addition
>> to the current CONUS 5 km version.A separate TIN will announce the
>> addition of those 2.5 km grids to NOAAPORT.
>> I doubt that the full suite of fields necessary to initialize a local
>> run will be available from the nested runs, simply because
>> the file size would be prohibitively large.
>> Product Delivery Time Change:
>> Changes may require EMS users to reduce the DELAY setting in the nam
>> gribinfo.conf files to DELAY = 2
>> and adjust the start time accordingly.
>> NCEP anticipates that output delivery times after the NAM upgrade
>> will differ slightly throughout the run compared to the current
>> operational NAM products.The precise amounts still need to come out
>> of the final pre-implementation testing by NCEP Central Operations,
>> but the following is the worst case scenario. When the CONUS Fire
>> Weather nest is run during the first 36 hours of the forecast,
>> product delivery will lag the current delivery by ~13 seconds each
>> forecast hour such that
>> 36 hour guidance will be 468 seconds or 7.8 minutes later than at
>> present. Following the 36 hour point of the forecast, NWS will
>> recover delivery time at roughly 10 seconds each forecast hour such
>> that delivery of the 84 hr guidance will be at the same time as it is
>> delivered today, which is 10 minutes earlier than its target.As an
>> example, for 0000UTC run, the following delivery time differences
>> would be expected, with current time in parentheses:
>> 12hr PRODUCTS at 01:52:52 (01:50:16)
>> 24hr PRODUCTS at 02:05:26 (02:00:14)
>> 36hr PRODUCTS at 02:17:49 (02:10:01)
>> 48hr PRODUCTS at 02:25:47 (02:19:59)
>> 60hr PRODUCTS at 02:34:04 (02:30:16)
>> 72hr PRODUCTS at 02:40:23 (02:38:35)
>> 84hr PRODUCTS at 02:50:23 (02:50:35)
>> The target delivery for 84 hr guidance is 3:00:00.
>> When the fire weather nest is over Alaska, product delivery will lag
>> the current delivery by ~5 seconds each forecast hour such that 36
>> hour guidance will be 180 seconds or 3 minutes later than at present.
>> For a detailed description of this NAM and NDAS upgrade, including
>> verification statistics, please see:
>> For more general information about the NAM and NDAS, please see:
>> A consistent parallel feed of data will become available on the NCEP
>> server once the model is running in parallel on the NCEP Central
>> Computing System bylate May.The parallel data will be available via
>> the following URLs:
>> NCEP has tried to anticipate all filename and product content changes
>> associated with this implementation, but if we discover additional
>> changes during the course of the testing, we will send an amended
>> version of this TIN with that informationas soon as possible.
>> NCEP urges all users to ensure their decoders can handle changes in
>> content order, changes in the scaling factor component within the
>> product definition section (PDS) of the GRIB files, and volume
>> changes. These elements may change with future NCEP model
>> implementations. NCEP will make every attempt to alert users to these
>> changes before implementation.
>> Robert A. Rozumalski, PhD
>> NWS National SOO Science and Training Resource Coordinator
>> COMET/UCAR PO Box 3000 Phone: 303.497.8356
>> Boulder, CO 80307-3000
>> wrfems mailing list
>> wrfems at comet.ucar.edu
Robert A. Rozumalski, PhD
NWS National SOO Science and Training Resource Coordinator
COMET/UCAR PO Box 3000 Phone: 303.497.8356
Boulder, CO 80307-3000
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the wrfems