NOUS41 KWBC 181438



Technical Implementation Notice 16-26

National Weather Service Headquarters Washington DC

238 PM EDT Mon Jul 18 2016


To:       Subscribers:

          -NOAA Weather Wire Service

          -Emergency Managers Weather Information Network


          Other NWS Partners, Users and Employees


From:     Tim McClung

          Portfolio Manager

          Office of Science and Technology Integration


Subject:  Upgrade to the Rapid Refresh (RAP) and the

          High-Resolution Rapid Refresh (HRRR)

          Analysis and Forecast System

          Effective August 23, 2016


Effective on or about Tuesday, August 23, 2016, beginning with

the 1200 Coordinated Universal Time (UTC) run, the National

Centers for Environmental Prediction (NCEP) will implement

Version 3 of the Rapid Refresh (RAP) and Version 2 of the 

High-Resolution Rapid Refresh (HRRR) systems.


A major change to the RAP will be an expanded computational

domain which will now include Hawaii. This expansion will

facilitate future NCEP plans for ensemble systems and in time

improve the initialization of Short Range Ensemble Forecast

(SREF) members that use the RAP for initial conditions.


Analysis Changes:


Both the RAP and HRRR will use an updated version of the

Gridpoint Statistical Interpolation (GSI) analysis code.

Refinements were made to the GSI to improve the assimilation of

surface observations, soil moisture adjustment, and three-

dimensional cloud and precipitation hydrometeors. In addition,

the HRRR will start using the ensemble/hybrid data assimilation;

it is already used in the RAP, but the weighting of the

ensemble-based component in the RAP will increase from 0.50 to

0.75. In addition, while the RAP already cycles land-surface

states, this cycling is being introduced into the HRRR. In HRRR

Version 1, all runs are independent.


Other analysis changes include:


-Assimilating radial wind  and mesonet data

-Applying PBL-based pseudo-innovations for 2-meter temperatures

(already used for 2-meter dew points)

-Changing the cloud-hydrometeor assimilation to avoid METAR-

based cloud building when satellite data shows clear skies at

all times of day (currently used just in daytime)

-Introducing direct use of 2-meter temperature and dew point

model diagnostics in the GSI.


Specific to the HRRR, the application of radar reflectivity data

in the GSI to direct specification of 3-dimensional hydrometeors

is increased to apply to a broader range of weather conditions,

including warm-season events with reflectivity up to 28 dBZ.


Changes to Model:


- The RAP and HRRR will both begin using WRF version 3.6.1; both

will continue to use the ARW core.

- The MYNN planetary boundary layer scheme is being updated to

include the effects of subgrid-scale clouds. The mixing length

formulation in the boundary layer scheme and thermal roughness

in the surface layer are being changed. 

- The 9-level RUC land-surface model is being updated to add a

mosaic approach for fractional snow cover, improve the fluxes

from snow cover, and modify the wilting point for cropland use.     

- Major updates are being made to the Thompson microphysics

scheme, including making it aerosol-aware with use of an 

ice-friendly and water-friendly aerosol field.

- Shortwave and longwave radiation have been changed to use the

RRTMG (RRTM global) scheme that includes the effects of aerosols

and boundary layer subgrid-scale clouds. 

- The WRF-ARW diagnostics for 2-meter temperature and dew point

are being improved.  

- The convective scheme in the RAP is changed from the Grell 3-D

scheme to the scale-aware Grell-Freitas scheme. The HRRR, at 

3 km horizontal resolution, explicitly resolves convection and

does not use a convective scheme.


Many of these changes to the data assimilation, land-surface

model, boundary layer scheme, microphysics, radiation, and (in

the RAP only) convective scheme are designed to mitigate the

low-level warm, dry bias in the RAP and HRRR, most notable

during afternoons in the warm season. Significant reduction of

these biases has been evident in extensive testing.


Output Changes:


The HRRR directory structure will be migrated out of

nonoperational on the NCEP ftp and http servers. Data will also

be available on both the primary and secondary servers.






The forecast lengths of both the RAP and the HRRR will be

extended on the NCEP servers:


The RAP hourly output will be extended from 18 to 21 hours for

all cycles. The HRRR hourly output will be extended from 15 to

18 hours for all cycles.


Output files will be written directly to GRIB2 instead of first

being written out to GRIB1 and converted to GRIB2. Users may see

some differences with the encoding and are encouraged to upgrade

their software to handle this. Please see the links below for

upgraded code:


The following GRIB1 RAP output is being discontinued on NCEP and

NWS servers. The 40km Regional RAP will be discontinued here

only. There will be no impacts to NOAAPORT.


The GRIB2 equivalent is online at:






Where CC is cycle, FF is forecast hour (00-21), YYYYMMDD is

year, month and day


- The following file names will change to remove “tm00”:


rap.tCCz.wrfprsHH.tm00.grib2 -> rap.tCCz.wrfprsfHH.grib2


rap.tCCz.wrfnatHH.tm00.grib2 -> rap.tCCz.wrfnatfHH.grib2


Where CC is the cycle (00-23), and HH is the forecast hour (00-



In conjunction with the expansion of the RAP domain, output on a

new forecast grid will be added to the NCEP ftp/http servers,

opendap and grib filter. RAP output on grid 243, which covers

Hawaii and the eastern Pacific Ocean, will be available with









Where CC is the cycle (00-23) and FF is the forecast hour (00-



The 3-hr pressure tendency field, available in several RAP

output grids, will no longer be available. Users will need to

compute this parameter on their own.


The value that represents the resolution and component flags of

the grid definition template is changed in all RAP output from 

8 to 56. This indicates the resolved u and v components of

vector quantities are relative to the defined grid in the

direction of increasing x and y (or i and j) coordinates,

respectively, and the i and j direction increments (representing

dx and dy of the model), are stored within the grid definition

templates for all RAP output grids.


The labeling of the following RAP parameters are being modified,

mainly due to different conventions in the way direct grib2

output is post-processed. These changes apply to any and all RAP

output grids on which the parameter is found:


     Composite reflectivity (Discipline 0, Category 16,

     Number 196):  The fixed surface type is changed from

     200 (Entire Atmosphere (considered as a single layer))

     to 10 (Entire Atmosphere).


     U and V components of storm motion (Discipline 0,

     Category 2, Numbers 194 and 195):  The current

     labeling of the first specified height level above

     ground as 6000 and the second level as 0 is changed so

     that 0 is the first level and 6000 is the second.


     U and V components of vertical wind shear (Discipline

     0, Category 2, Numbers 15 and 16):  The current

     labeling of the first specified height level above

     ground as 6000 and the second level as 0 is changed so

     that 0 is the first level and 6000 is the second.


     Total Cloud Cover (Discipline 0, Category 6, Number

     1):  The fixed surface type is changed from 200

     (Entire Atmosphere (considered as a single layer))

     to 10 (Entire Atmosphere).


     Soil Moisture availability (Discipline 0, Category 0,

     Number 194):  The fixed surface type is changed from 1

     (surface) to 106 (depth below land surface), with the

     depth specified at 0 meters. 


     Relative humidity computed with respect to

     precipitable water (Discipline 0, Category 1, Number

     242):  The fixed surface type is changed from 200

     (Entire Atmosphere (considered as a single layer)) to

     10 (Entire Atmosphere).


     For all soil moisture (Discipline 2, Category 0,

     Number 192) and soil temperatures (Discipline 2,

     Category 0, Number 2), the fixed surface type is still

     106 (depth below land surface), but the depth is now

     reported as a range between 2 levels (with the chosen

     level listed as both levels) instead as a single

     level.  As an example, soil moisture at 40 cm is

     currently listed as soil moisture at a depth of 0.40

     meters; it will now be listed as soil moisture between

     a depth of 0.4 meters and 0.4 meters.


Changes to Labeling of Parameters in the HRRR


     For the lightning flash rate (Discipline 0, Category

     17, Number 192), the fixed surface type is changed

     from 1 (Surface) to 10 (Entire Atmosphere).


     The sea level pressure field (Discipline 0, Category

     3, Number 1) is replaced by an alternate sea level

     pressure field (Discipline 0, Category 3, Number 198)

     to be consistent with the RAP.


     The best lifted index field (Discipline 0, Category 7,

     Number 11) is changed to Discipline 0, Category 7,

     Number 193, making it consistent with the RAP.


     The number of concentration of ice particles is

     changed from Discipline 0, Category 1, Number 207 to

     Discipline 0, Category 6, Number 29.


     The number of concentration of rain is changed from

     Discipline 0, Category 255, Number 255 to Discipline

     0, Category 1, Number 100.


RAP Precipitation Fields


The RAP currently generates convective and non-convective

precipitation totals. Total precipitation will be added to the

output so users no longer need to sum the convective and non-

convective amounts. To become consistent with other NCEP models

that generate only total and convective precipitation, the run

total and 1-hour accumulation non-convective precipitation

fields will be eliminated in the 2017 RAP Version 4 upgrade.

Users should start preparing for this change but all three

precipitation parameters will be provided in RAP Version 3 to

assist with the transition.


For the precipitation fields, RAP files currently contain run

total and 1-hr accumulations for all forecast hours. In

addition, forecast hours 5, 8, 11, 14 and 17 contain an

additional 2-hour accumulation, while forecast hours 6, 9, 12,

15 and 18 contain an additional 3-hour accumulation.  The 2-hour

accumulations will be discontinued in all output files on NCEP

and NWS servers but will be retained in files sent to AWIPS. The

3-hour accumulations at hours 6, 9, 12, 15, 18 (and 21) will be



Similar changes are made to the snow water equivalent field. The

RAP output grids right now contain various combinations of

instantaneous, run total and 1, 2 or 3-hour accumulation of snow

water equivalent. The output will be unified so that all grids

at all output times have instantaneous, run total, and 1-hr

accumulation of snow water equivalent. The files for all

forecast hours divisible by 3 will contain 3-hour accumulations

of snow water equivalent. These snow water equivalent

accumulations are being changed to contain contributions only

from snow hydrometeors.


Two new winter precipitation fields are being added: 

accumulated snow depth that includes variable density

hydrometeor accumulation, and snowmelt affects (Discipline 0,

Category 1, Number 29) and accumulated graupel  (Discipline 0, 

Category 1, Number 227), both available in 1-hour and run total

amounts. Note that the parameter listing is for frozen rain

because no parameter currently exists for graupel.


Changes to RAP Cloud Base Height / Ceiling


The RAP output currently contains a parameter called cloud base

height (fixed surface type 2) which actually represents the

cloud ceiling. In this upgrade, cloud base height will actually

represent the true cloud base height, and the old computation is

now properly labeled as cloud ceiling (fixed surface type 215).  

This makes the RAP consistent with the current operational HRRR.


Added and Removed RAP and HRRR Parameters:


A full list of parameters added to and removed from all RAP and

HRRR output grids can be found at:


Changes to NOAAPORT:


With this upgrade, the volume of data transmitted over NOAAPORT

will increase due to GRIB packing changes:


-RAP 13km increase 80MB/cycle

-RAP 40km increase 6MB/cycle


In the future, NCEP will add the following products to NOAAPORT.

Please see the following documents for the details of these

future additions:


RAP Puerto Rico:


HRRR sub-hourly output:


RAP 13km extended hourly forecast hours:


HRRR extended hourly forecast hours:


Product Delivery Times:


HRRR analysis products are available approximately 6 minutes

later in Version 2, but the delay is diminished as the forecast

progresses such that the products arrive 2 minutes earlier by

forecast hour 15. RAP analysis products are available 3 minutes

later in Version 3, but the delay is diminished as the forecast

progresses such that the products arrive at the same time by

forecast hour 18.


Parallel Data Location:


A consistent parallel feed of data is available on the NCEP

server via the following URLs:


For more general information about the RAP and HRRR, please see:


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.


For questions regarding this implementation, please contact:


      Geoff Manikin

      NCEP/EMC/Mesoscale Modeling Branch

      College Park, MD



      Curtis Alexander

      ESRL / Global Systems Division

      Boulder, Colorado




For questions regarding the data flow aspects of these datasets,

please contact:


     Carissa Klemmer

     NCEP/NCO Dataflow Team Lead

     College Park, Maryland 20746



NWS National Technical Implementation Notices are online at: