How to Compile?

How to Run?

WRF Namelist

Known Problems and Fixes

Graphic Tools





WRF Model Version 3.6: UPDATES

WRF Model Update

The WRF model tar file has been updated to Version 3.6 on April 15, 2014. For known problems in V3.6, please click here.

The WRF Pre-Processing System (WPS) has been updated to Version 3.6.

Note: V3.6 WRF can use wrfinput and wrfbdy files from V3.5 version. However if lake physics is activated, it is recommended to rerun V3.6 WPS to obtain lake depth data.

As always, be cautious when using new options.

If you are interested in seeing how V3.6 has been tested, click here.

New in Version 3.6:


- Land surface models:

* Noah LSM: an option to use sub-tiling in the scheme is introduced. Instead of using dominant category before, a user can choose N number of categories in a grid box. This option is activated by setting sf_surface_mosaic = 1, and mosaic_cat = N. N = 3 is the default. (Thanks to Dan Li of Princeton Univ.)
* Lake mode (sf_lake_physics=1): The lake scheme was obtained from the Community Land Model version 4.5 (Oleson et al. 2013) with some modifications by Gu et al. (2013). It is a one-dimensional mass and energy balance scheme with 20-25 model layers, including up to 5 snow layers on the lake ice, 10 water layers, and 10 soil layers on the lake bottom. The lake scheme is used with actual lake points and lake depth derived from the WPS, and it also can be used with user defined lake points and lake depth in WRF (lake_min_elev and lakedepth_default).
The lake scheme is independent of a land surface scheme and therefore can be used with any land surface scheme embedded in WRF. The lake scheme developments and evaluations were included in Subin et al. (2012) and Gu et al. (2013). (Thanks to Hongping Gu of Institute of Atmospheric Physics, Chinese Academy of Sciences and Jiming Jin of Utah State Univ.)

- Microphysics option:

* HUJI spectral bin microphysics, full and fast version (Thanks to Alex Khain and Barry Lynn of Israel)
* Thompson aerosol-aware scheme. Click here for more information on this option. (Thompson and Eidhammer 2014, JAS) (Thanks to Thompson and Eidhammer of NCAR/RAL).

- Others:

* aer_opt = 2: A new way to input aerosol for RRTMG and Goddard radiation options. Either AOD at 550 nm or AOD plus Angstrom exponent, single scattering albedo and cloud asymmetry parameters can be input via constant values from namelist or 2D input via auxiliary input stream 15. Aerosol type may be set. (Thanks to Jose Ruiz-Arias of Universidad of Jaen, Spain)

* icloud: Cloud fraction option. In the past, the cloud fraction calculation is tied to a radiation option. This can sometimes lead to different cloud fraction seen short- and long-wave radiation schemes. Since v3.6, cloud fraction calculation becomes independent of the radiation options, but needs to be selected by setting icloud = 1 or 2. Option 1 is the Xu-Randall method, and 2 is the threshold method which gives either 0 or 1 cloud fraction.

* cu_rad_feedback: Sub-grid cloud fraction is diagnosed from the Kain-Fritsch cumulus scheme, and passed on to the radiation with icloud=1. In previous versions, this option works with Grell-type schemes. (Thanks to Alapaty and Herwehe of EPA)

* rdlai2d: option to use 30 sec, 12 monthly leaf-area index data. (Thanks to Mike Barlage of NCAR/RAL)
* 15 sec, 20-category MODIS landuse data. (Thanks to Patrick Broxton and Xubin Zeng of University of Arizona)

Improvements and Bug fixes:


Atmospheric Radiation:

- CAM option: bug fix for coszen for diagnostic computation of shortwave surface downward diffuse and direct irradiance. The problem appeared since v3.5.1 only.


- NSSL 2-moment scheme: Updated. For details, click here.

- Milbrandt: improved ice-snow (now less ice, more snow) and graupel-hail (now more hail) balances. Replaced box-Lagrangian sedimentation with Eulerian. More details here.


- Grell-Freitas: Updated (posted in 3.5.1 known problem page).
- NSAS deep: a method to introduce grid-size awareness. Controlled by setting nsas_dx_factor = 1 in the namelist. (Thanks to Songyou Hong of YSU and KIAPS Korea)

Planetary boundary layer and surface layer physics:

- Options to mix scalars (mostly Qnc, Qnr) and tracers in PBL. Activated by setting scalar_pblmix = 1, and tracer_pblmix = 1.
- TEMF: Robustness fix and improved surface velocity scale in surface layer scheme (Thanks to Robert Rozumalski and Wayne Angevine)
- MYNN PBL and surface layer: updated. (Thanks for Joe Olson of NOAA)

* PBL: Added extra output (TKE_PBL, DELTA, WSTAR) for coupling to GRIMS and CAMUW shallow-cu schemes.
* PBL: Added option for full cloud (water & ice) mixing.
* Surface layer: Relaxed limits on moisture flux over water that was falsely thought to be a culprit for excessive fluxes.
* Surface layer: Improved formula for the kinematic viscosity.

Surface physics:

- NoahMP: There are a number of improvements in this release

* Melting factor to 2.5
* Latent heat constant dependent on T of ground and leaf
* Min LAI/SAI = 0.05
* Canopy evaporation – sublimation dependent on phase
* Ground evaporation comes from ice when ground is frozen
* New-snow density fixed
* fixes an uninitialized variable (TGB) going into the glacier code (can be fatal);
* forces the initialization of snow height to be consistent over glacier
* limits the initialization of surface temperatures when snow on ground and input TSK > 0C to be at most 0C
* assure consistency of SNOWH and SNOW so that one does not get set to zero while the other is >0
* only uses latent heat of sublimation over ice, remove temperature dependence
* prevent snow age from getting reset to 0 each call
* when snow gone, make sure excess sublimation is taken from "soil" ice not water
* assure consistency of SNOWH and SNOW so that one does not get set to zero while the other is >0
* make snow burial limits consistent with new MPTABLE
* for 2m temperature calculation diagnostic, add a split for seaice/glacier after the NoahMP call


* assure consistency with canopy top and bottom and vegetation roughness length between vegetation types
* fix a few errors when dynamic vegetation parameters were copied from USGS to MODIS
* assure consistency in stem and leaf area index
(Thanks to Mike Barlage of RAL/NCAR)

- CLM4: update to use MODIS landuse; allow bldt option to work
- RUC LSM: improved snow model for thin snow, and capability to use input LAI data.
- SSiB: updated to add 10 m wind diagnostics.

- WRF-Hydro 2.0

* Both Noah and NoahMP are supported land models with WRF-Hydro coupling capabilities in WRF3.6.
* Stand alone Noah and NoahMP drivers coupled with WRF-Hydro hydrology components are included in WRF-Hydro 2.0.
* To improve the performance on supercomputers (such as 'yellowstone') a new domain decomposition method is applied for WRF-Hydro offline runs.
* New forcing data features and filename conventions are added to WRF-Hydro offline runs.
* A few bug fixes from WRF-Hydro 1.0 release have been made for WRF-Hydro 2.0.
NOTE: In WRF3.6 and WRF-Hydro 3.2, the FLAKE model is NOT interactive with WRF-Hydro components. Future versions will enable this interactivity.


- Wind farm drag parameterization: Bug fix to use the wind speed at the hub height instead of the mean wind within the area covered by the rotor to compute thrust and power coefficients. Update to more easily specify turbine characteristics and locations for both real and idealized cases (Thanks to Pedro Jimenez of CIEMAT, Spain)
- Fog deposition (grav_settling): bug fix (Thanks to Travis Wilson of UCLA)
- SKEBS: more user-controlled options.


- Bug fix for option obs_max_sndng_gap in obs nudging (Thanks to Brian Reen of ARL)
- Prevent negative moisture nudging, activated by setting namelist obs_scl_neg_qv_innov = 1 in &fdda (Thanks to Brian Reen of ARL. See

Moving nests:

- Moving nests can now use high resolution terrain and landuse dataset during a run. For more detail, see the moving nest page. (Thanks to Shuyi Chen, University of Miami group and John Michalakes for the contribution. Chen et al., 2007, BAMS)


- A new vertical interpolation option using cubic spline is added (Thanks to CWB of Taiwan)
- Bug fix for 2:1 nest ratio.
- Idealized seabreeze case has more levels, and levels can be specified like that in a real-data case.
- Idealized quarter ss case: more levels and more digits for moisture variable are added to give a smoother sounding.
- Idealized LES case: 5 K was removed from skin temp.
- Trajectory option ok for idealized cases. A template for initializing trajectories is added to module_initialize_quarter_ss.F.
- Bug fix for calculation of dry pressure in idealized tropical cyclone initialization (Thanks to Wenyu Zhou of Princeton).
- Add an example to initialize tracer in module_initialize_real.F.
- Bug fix for option lowest_lev_from_surface (Thanks to Brian Reen of ARL).
- Bug fix for program real using files obtained after using tc.exe.
- Improved masked interpolation
- A number of diagnostic output from AFWA (click here for its documentation. Note that more diagnostics is added in 3.6.1.)


- Bug fix for D33 term to be multiplied by Kh instead of Kv, affecting only vertical diffusion of w with diff_opt=2 (Thanks to Rich Rotunno of NCAR)
- Improve stability in complex terrain when using diff_opt = 2 and km_opt = 4.

namelist change:

- namelist option sf_sfclay_physics = 1 is now the revised MM5 surface layer scheme. The old MM5 surface layer option is now 91.
- namelists diff_opt and km_opt are now domain-dependent, so one can change them based on grid resolution
- namelist windturbines_spec is removed, and the new namelist to activate the scheme is windfarm_opt = 1. Another namelist, windfarm_ij, may be used to specify turbine locations in lat/lon or I/J.
- All SKEBS options are now under namelist record &stoch, and perturb_bdy is moved there too.


- Speed up netCDF write by doing write only once, instead of the normal write/read/write sequence (Thanks to Jim Abeles of IBM)
- CPP option adapted for newer version (-C flag removed)
- fft code updated to use double precision. Affect filtering in global run only.
- A new diagnostic driver is created to accommodate a number of diagnostics
- Binary io updated, 2003 iso and ieee tests separated.
- Improved compile dependency: now can use more processors to compile..
- Removing leading (confusing) messages in the standard output.
- Add more information about computer and compiler when compile
- Allow adaptive time steps to be less than 1 sec.


Previous Update

WRF V3.0.1
WRF V3.0.1.1


Home -- Model System -- User Support -- Doc / Pub -- Links -- Download -- WRF Real-time Forecast