WRF Model Update
The WRF model tar file has been updated to Version
3.9 on April 17, 2017. For known problems in V3.9, please click here.
The WRF Pre-Processing System (WPS) has been updated to Version 3.9.
Note: V3.9 WRF can use wrfinput and wrfbdy files from V3.6 to 3.8 versions. If you are interested in trying the new hybrid vertical coordinate option, then you must rerun program real.exe.
As always, be cautious when using new options.
If you are interested in seeing
how V3.9 has been tested, click here.
New in Version 3.9:
Hybrid sigma-pressure vertical coordinate:
* A hybrid sigma-pressure vertical coordinate is added to WRF as an option. This coordinate is terrain-following near the surface, and gradually transitions to constant pressure at higher levels. In V3.9, this is an option that requires special compile ('configure -hyb') and namelist option to activate (hybrid_opt=2). See README.hybrid_vert_coord for details.
* Physics Suites:
Introduced a new way to set physics options by using 'physics_suite' namelist. The physics options covered by the suite specification are mp_physics, cu_physics, bl_pbl_physics, sf_sfclay_physics, sf_surface_physics, ra_sw_physics and ra_lw_physics. Instead of setting these six separate options, a single suite namelist (physics_suite) sets them all. Two well-tested suites are provided in this release: "CONUS" and "TROPICAL". Any option can be overwritten by directly including the particular physics option in the namelist.
* P3 microphysics: Predicted Particle Property scheme (Morrison and Milbrandt, 2015, JAS). This scheme represents ice-phase microphysics by several physical properties in space and time. The four prognostic variables are mixing ratios of total mass, rimed ice mass, rimed ice volume, and number. The liquid phase is double moment rain and ice. The scheme is coupled to RRTMG radiation. 10 cm radar reflectivity diagnosed. Uses data file, p3_lookup_table_1.dat, provided in run/. mp_physics = 50: basic scheme, mp_physics = 51: As in 50 but adds supersaturation dependent activation and double-moment cloud water.
* Noah-MP: [Thanks to Mike Barlage (NCAR), Francisco Salamanca (Arizona State), Yizhou Zhang (IUM), Xing Liu (Purdue), Gonzalo Miguez-Macho (Santiago de Compostela)]
- Urban physics added to NoahMP.
- Making groundwater input data available via geogrid to NoahMP Converts what was an auxiliary input dataset that had to be created by the groundwater module developer (Gonzalo Miguez-Macho) to datasets created in WPS. Allows for general user with global applications to activate groundwater model.
- Added crop-related input data in existing Noah-MP module Adds four additional variables: croptype (% of corn,soybean,wheat,cotton coverage), planting (seed planting date), harvest (harvest date), and season_gdd (climatological growing degree days over crop season). croptype is used in subroutine noahmp_init to determine crop species on grid cell. The crop module was added in V3.8; these mods allow for a increased functionality: ability to run specific crops (uses can modify crop specifics in MPTABLE) and to run non-crop on non-crop grids.
* Stochastic Parameterization Schemes: (Thanks to Judith Berner of NCAR)
- Stochastically perturbed physics tendencies (SPPT) (sppt=1): A random pattern is used to perturb the accumulated physics tendencies (except those from micro-physics) of potential temperature, wind and humidity (Berner et al., 2015 MWR).
- Stochastically perturbed parameter scheme (SPP) (spp=1): A random pattern is used to perturb parameters in selected physics packages, currently the GF convection scheme, the MYNN boundary layer scheme and the RUC LSM. Parameter perturbations to a single physics package can be achieved by setting spp_conv=1, spp_pbl=1 or spp_lsm=1 (Jankov et al. 2016, MWR).
* Added new 30 arc second global, 16 soil category datasets (based on Beijing Normal University's). (Thanks to Mike Barlage of NCAR)
Improvements and Bug fixes:
RAP/HRRR Physics: (Thanks to Joe Olson, Tanya Smirnova, Georg Grell and others at GSD/NOAA)
* MYNN PBL (bl_pbl_physics=5,6) and
MYNN surface layer (sf_sfclay_physics=5):
(a) Modification to the mass-flux scheme to improve match to LES and improve HRRR. The multi-plume mass-flux scheme was using 10 plumes at all times, but now, the number is dependent on the grid spacing (only using the smaller plumes at high-resolution - sub 1.3 km) and PBL (only using the larger plumes as the PBL has been develops) structure. The subgrid clouds for the mass-flux scheme were also improved by switching to the exact form of Chaboureau & Bechtold (2005) as opposed to an approximate form. The stratus component of the subgrid clouds (Chaboureau and Bechtold 2002) was also modified to improve areal coverage and reduce the excessive liquid water content in subgrid-scale ice clouds..
(b) Added the use of the stochastic parameterization perturbation (from Judith Berner). The method has option to perturb the subgrid clouds, mixing lengths, mass-flux, and the thermal roughness lengths. This is ongoing research.
(c) Modified mixing length to improve surface wind under stable conditions (bl_mynn_mixlength = 2).
* RUC LSM (sf_surface_physics = 3):
(a) Added option for stochastic perturbation of soil moisture.
(b) Make the transpiration function now includes dependence on solar angle to reduce overestimated transpiration at sunset, moist bias in 2-m dewpoint.
(c) More accurate treatment of water dripping from the canopy in case of mixed-phase precipitation: frozen portion of drip is added to snow accumulated on the ground, and the liquid portion is added to water infiltrating into soil.
(d) Apply "mosaic" snow not only during the melting season, but also with cold temperatures when snow fraction is less than 0.75 to reduce cold/dry bias over thin snow under conditions of night-time radiative cooling.
(e) Modifications in 2-m T diagnostics: reduced threshold value of exchange coefficient when 2-m T is set equal to the first atm. level. The old threshold could cause a warm bias in 2-m T during the night hours.
(f) Added new variable - accumulated surface runoff - ACRUNOFF.
* Grell-Freitas (cu_physics=3):
This is a rework and clean-up of the GF parameterization. The original module is now broken into three pieces to be able to keep the routine for deep and shallow convection identical between different models (used in various regional and global models). The driver is wrf specific. Many unused arrays were removed in the resulting three modules. Some options were refined. Additionally, the following are added.
(a) Momentum transport (as in ECMWF).
(b) Different cloud water detrainment options (c1 constant).
(c) A Beta PDF to approximate normalized vertical mass flux.
(d) If perturbation patterns can be supplied (temporal and special correlations), 6 different stochastic options (4 for closure assumptions, one for vertical mass flux, one for momentum transport) The scale awareness has not changed, but in general this version will produce different results. (Publication is under preparation).
* NSAS scheme (cu_physics = 14): fixed a bug for non-zero cudt that makes the time step used in the schemes (deep and shallow) and for tendency calculation consistent. Small fix in indexing when height field was used. [Thanks to Sebastien Masson (LOCEAN - IPSL, Univ. Pierre et Marie Curie, France), Jihyeon Jang (NCAR), and Kun Liu (CMA, China)].
* New Tiedtke (cu_physics = 16): Making the non-equilibrium convection the default, and minor update to reduce light rain. (Thanks to Chunxi Zhang of U. of Hawaii)
* NSSL scheme: (Thanks to Ted Mansell of NSSL)
(a) The 2-moment schemes (mp_physics = 17, 18, or 22) now create number concentration tendencies from cumulus scheme mass mixing ratio rates, when active (similar to Morrison scheme).
(b) Restored older settings that allow snow aggregation starting at T > -25C (instead of T > -12, resulting in slightly higher reflectivity in stratiform regions).
(c) Adjusted Meyers number of activated nuclei by the local air density to compensate for using data at surface.
(d) Minor updates to rain-ice crystal and hail-rain collection efficiencies.
* Thompson aerosol scheme (mp_physics = 28):
There is an adjustment with the aerosol-aware scheme to account for number concentration of dust to alter from physical air temperature to the 'effective temperature' that gets used as part of the probability of freezing water drops into either cloud ice or graupel. The bug could create a negative number for something that should remain always positive. The total overall effect of the bug is quite limited due to the small temperature range of about zero to 3 degrees C and for specific range of the incoming ice nuclei variable (QNIFA).
The change is in subroutine freezeH2O, which creates a static look-up table, freezeH2O.dat, also placed in the run directory. For WRF version 3.7 or later, a user should remove existing freezeH2O.dat file in the run directory, and let the model re-create it after making code change. For 3.6 and 3.6.1, contact wrfhelp.
Planetary boundary layer and surface layer physics:
* topo_wind=1 option: Improvement of wind simulation during day-time by considering stability effect [Thanks to Raquel Lorente (formerly RAL) and Pedro Jimenez (RAL)]
* topo_wind: check to see if VAR_SSO field is available to use topo_wind option.
* Replacing rovcp with rcp (from module_model_constants.F, minor difference) for a few surface layer and LSM options: SCM, TEMF surface layer, RUC LSM, and ideal SCM surface layer options. (Thanks to the report by Tanya Smirnova)
* sf_sfclay_physics = 1, 91: Set a limit for ZNT over water under high wind conditions, the same way as in ISFTCFLX option.
* Fixed a bug to allow the revised MM5-surface layer scheme (now sf_sfclay_physics = 1) to work with seaice.
* The pressure used by these physics options are changed to hydrostatic pressure from dry hydrostatic pressure: YSU PBL, Shin-Hong PBL, and gravity wave drag.
Land Surface physics:
* NoahMP: Fixed a problem with strange, unstable soil moisture performance in special cases of high precipitation (Thanks to National Water Model team and Mike Barlage of RAL/NCAR). Fixed an uninitialized variable FH2 for case SFC_DIF=2.
* Noah: Fixed a minor bug in snow melting when the amount is small (snow cover fraction < 0.01); Increased CHS lower limit to 1.E-2 from 1.E-4 for urban areas to maintain stability and make Noah consistent with Noah-MP.
* Urban physics (1): Fixed a minor bug in the green-roof option to allow for surface-energy closure condition. The problem occurs in very dry conditions, and when time step is small. (Thanks to Jiachun Yang, Princeton Univ)
* Fixed a bug to accommodate surface current from a coupled ocean model (with YSU PBL) (Thanks to Sebastien Masson of France).
* sf_ocean_physics = 1: Allow simple mixed layer ocean model to use just the climatological mixed layer depth data by setting oml_hml0 = 0 (Need to provide the climatological data).
* Fixed a minor bug for 3-domain restart with RRTMG plus CAM ozone input (o3input = 2). This may affect restart results after V3.7 when o3input = 2 became default.
* Fixed a major bug with Ferrier-Hi-Res microphysics for ARW and its use with CAM radiation option.
* Fixed a major bug for cloud fraction calculation when using WSM3 and Kessler schemes: cloud was ignored when these MP schemes are used. Error introduced in V3.8.
* FLG: Fixed a minor bug that should check if the temperature is greater than 345 K rather than 320 K. (Thanks to Gu of UCLA)
* All physics tendencies (RTHCUTEN, RTHCUTEN, RTHRATEN, etc.) are now output without coupling to column pressure.
* HAILCAST model updated and controlled by new option hailcast_opt (Thanks to Becky Adams-Selin AER)
* Trajectory diagnostics: The original "trajectory" code in routine module_em.F only computed the trajectory spatial coordinates per trajectory, per time step. The new code, in share/module_trajectory.F, linearly interpolates in x,y, and z to get the value of up 100 variables per trajectory per time step. (Thanks to ACOM team of NCAR)
* icloud = 3: Further tuning with CONUS 10 and 3.3 km grid over a one-year (2015) period (52 simulations). (Thanks to Greg Thompson, Mei Xu and Pedro Jimenez of NCAR/RAL)
* RASM diagnostics: time averaging and diurnal averaging of many surface variables and radiative fluxes at model top. (Thanks to the RASM team at CU).
* Gravity wave drag (gwd_opt=1): If the map projection has a large rotation angle anywhere within the domain ("rotation" is directly from the sina variable in the WRF input and output, and "large" means greater than a default value of 22.5 degrees), notify the user and stop. The original GWD data has information at four directions, each separated by 45o angles. Therefore anything larger than a 22.5o projection rotation (usually only violated in parts of the coarsest domain) means that the wind components are using the wrong topographic directional information. For the determined user, the namelist variable max_rot_angle_gwd = 100 (really, anything greater than 90o) will always permit GWD to be used, regardless of the projection.
* Bug fixes for grid_sfdda = 2 (or FASDAS option). (Thanks to Tim Glotfelty and Kiran Alapaty of EPA)
* Fixed a dimension error in lightning option iccg_pr93 (iccg_method=3) (Thanks to Josh Laughner UC Berkeley)
* sst_update: if turned on, io_form_auxinput4 and auxinput4_interval are checked to warn user if they are not set.
* Fixed a minor bug with diag_print option: for small time steps, it can fail.
* Diagnostic arrays reset after DFI. (Thanks to Tanya Smirnova of NOAA)
* Adaptive time stepping: improvement for small time steps (sub-second) (Thanks to Brian Reen of ARL); Made it work with regional rotated lat/lon grid.
* Fixed a bug for restart when tmn_update is used. Variable TLAG was left out of restart file (Thanks to Heimo Truhetz of Univ of Graz, Austria)
Initialization (real) and nesting:
* Added capability to interpolate a dataset with different vertical dimension (e.g. QNWFA, QNIFA) than the primary meteorological dataset. This changes the way this data can be used for aerosol input for Thompson scheme. Click here for the updates.
* Added mixing ratio as a possible input field (previous only RH and specific humidity are ok), and this allows WRF take IC/BC from MPAS.
* real.exe: collect multiple error checks before abort; improved error message to assist debuging.
* real and wrf: checks added for mixed layer ocean and input, gwd and input, urban and input. Changing these options require rerun of real.exe.
* Fixed a bug for interpolated nest that is located over dateline. (Reported by Milan Curcic of U Miami).
The options in several namelist templates have been changed or updated.
* Speedup pd-limiter advection (default), WENO, and WENOPD by rearranging some loops in the code (Thanks to Negin Sobhani of U of Iowa and Davide Del Vento of NCAR)
* WENO advection options: added potential temperature. Also fixed a bug for backward integration in DFI.
* Fixed a minor bug in using TKE option in periodic BC in order to get bit-for-bit results when comparing MPI and serial results. (Reported by Bill Gustafson of PNNL and Pedro Jimenez of RAL).
* Fixed a minor bug in diagnostic surface sensible heat flux calculation for idealized cases (isfflx=0 or 2). The constant used should be cpm, not cp.
* Fixed a bug in diffusion calculation (diff_opt = 2, introduced in 3.8.1) for lat/lon grid.
* Changed default netCDF file support for large files. To turn large file support off, set environment variable WRFIO_NCD_NO_LARGE_FILE_SUPPORT instead.
* Corrected a bug for computing the total number of tiles. (Thaks to Samuel Elliot of CU)
* Fixed a bug for io_quilting: add missing argument. No effect.
* Tools added to compare registry and README.namelist to catch mis-stated dimensions. (Thanks initially to a script provided by Sebastien Masson of France).
* Ocean coupler: mechanism added to accommodate users who may not want to couple all variables. (Thanks to Sebastien Masson of France)