Skip to content

WRF Version 4.1
Compare
Choose a tag to compare
@kkeene44 kkeene44 released this 12 Apr 20:31
· 733 commits to master since this release
v4.1
dd5c4c3

The WRF Model has been updated to Version 4.1 on April 12, 2019.

The WRF Pre-Processing System (WPS) has been updated to Version 4.1.

Acknowledgements: We would like to thank A.-J. Deng (Penn State U.), John Henderson and Mike Iacono (AER Inc.), Patrick Hawbecker (National Renewable Energy Laboratory), HWRF team at (NCEP-EMC), Xiaoyu Xu (Inst. Of Urban Meteorology, China), Patrick Campbell, Jonathan Pleim, Limei Ran, and Robert Gilliam (US Environmental Protection Agency), Toshi Matsui and Carlos Cruz (NASA-GSFC), Jeremiah Johnson and Bart Brashers (Ramboll), Yuanbing Wang (Nanjing U. of Information Science & Technology, China), Joe Olson and Tanya Smirnova (NOAA-ESRL-GSD), Kevin Thomas (U. of Oklahoma-CAPS), Siou-Ying Jiang (Central Weather Bureau, Taiwan), Hamada S. Badr (Johns Hopkins U.), Brett Wilt of IBM, Sooya Bae and Songyou Hong (KIAPS, S. Korea.), and Anders Jensen, Michael Barlage, Greg Thompson, Mei Xu and Jamie Bresch (NCAR) for their contributions to this release.

New in Version 4.1:

Physics

Microphysics

  • Jensen Ice-Spheroids Habit Model with Aspect-ratio Evolution (ISHMAEL) microphysics (mp_physics = 55). (Jensen, A.A., J.Y. Harrington, H. Morrison, and J.A. Milbrandt, 2017: Predicting Ice Shape Evolution in a Bulk Microphysics Model. J. Atmos. Sci., 74, 2081–2104,https://doi.org/10.1175/JAS-D-16-0350.1)
    The parameterization predicts ice particle shape evolution for planar-nucleated (QICE) and columnar-nucleated (QICE2) ice. A third ice species, aggregates (QICE3), is also prognosed. Ice shape can change rapidly when ice grows by vapor deposition or riming, and thus, short microphysical timesteps are preferred. It is recommended that the option is used with 3 km grid spacing or less, and 6 sec timestep or less, but the scheme can be run at larger grid sizes and longer time steps. It’s also recommended to use positive-definite advection scheme (i.e. moist_adv_opt=2 and scalar_adv_opt = 2) with this scheme. The scheme is coupled to RRTMG radiation (ra_sw_physics=4, ra_lw_physics=4). The scheme is not officially coupled to cumulus parameterizations, but QICE is used for ice initialized as planar. The other two ice species (QICE2, QICE3) currently do not interact with cumulus parameterizations.

  • WRF Double Moment 7-Class Scheme (WDM7, mp_physics=26): This is a new double-moment microphysics with hail as a new prognostic variable. The added hail category helps to reduce the widespread light precipitation (by reducing the formation of graupel) and enhance the heavy precipitation. Both aspects are viewed as improvement over the WDM6 scheme.

  • WRF Single Moment 7-Class Scheme (WSM7, mp_physics=24): This is a new microphysics with hail as a new prognostic variable. The added hail category helps to reduce the light precipitation (by reducing the amount of graupel) in the stratiform region, and enhance the heavy precipitation in the leading edge of a simulated 3D idealized squall line as well as a real-data case. Both aspects are viewed as improvement over the WSM6 scheme.

  • Goddard 4-Ice microphysics scheme (mp_physics = 7) (Tao et al. 2016): This scheme replaces the old Goddard microphysics scheme. It has both graupel and hail. It outputs effective radii for microphysical species to interact with both the updated Goddard long- and short-wave radiation and RRTMG radiation. The 4ICE scheme was built upon improved versions of the Goddard 3ICE scheme developed for the Goddard Cumulus Ensemble model and includes dozens of new/modified functions regarding ice microphysics and unique particle-size mapping and observation-driven snow density.

Cumulus

  • Deng shallow convection option (shcu_physics=5) (Deng et al. 2003): It’s a mass-flux scheme. It considers both buoyant updraft and cloud with nearly neutral buoyancy, and predicts cloud fraction and cloud liquid content which then interact with radiation physics. Even though it is named a ‘shallow convection scheme’, it does have a deep convection component. In conditionally unstable environment, the scheme transitions to a version of Kain-Fritsch scheme. However, as the code is implemented in WRF now, this deep component is not very active, but it has been tested with a deep scheme on too with reasonable results.

Radiation

  • Goddard Radiation (ra_lw/sw_physics=5) (Matsui et al. 2018): The newest version of the Goddard radiation scheme involves three major improvements. First, a size-, shape-, and radiation-spectrum-consistent single scattering database (Yang et al., 2013) has been incorporated to represent hydrometeor-consistent optical properties (including rain, snow, graupel, and hail). Second, the molecular absorption database has been updated from HITRAN1996 to HITRAN2012, which reduces the biases for clear-sky radiation flux. Third, the radiation code has been vectorized for improved computational efficiency. This scheme is best used with Goddard 4ICE microphysics scheme.

Improvements and Bug Fixes

Physics

Land Surface Model

  • Noah-MP:

    • Updated with code modifications that are present in WRF-Hydro/Noah-MP. Non-answer changing: parameters data structure expanded, read/write statement formatting; Answer changing: remove normalized LAI, snowpack heat conservation, increase maximum SWE limit.
    • When running Noah-MP with no urban scheme (bulk method), the roughness length in urban areas used a bare soil value. This resulted in a high temperature, high wind speed, and low sensible heat flux over cities. The correction uses z0 from the Noah-MP look-up table.

  • PX LSM: The new option (pxlsm_modis_veg = 1) was added in the physics section of the namelist file. When activated, this option uses the time-varying VEGFRA and LAI from the wrflowinp_d01 file instead of the look-up values in the PX data table. Users are encouraged to use this option if they are using WPS v4.0 and later that have the higher resolution MODIS greenfrac dataset. Also, the soil calculation in the PX LSM were modified to use analytical functions from Noilhan and Mahfouf (1996) for field capacity, saturation and wilting point based on fractional soil data. Soil categories were updated in PX to 16 class consistent with the WRF system and other LSMs.

  • RUC LSM: flag_sm_adj is added to turn on/off soil moisture adjustment for RUC LSM. This flag should be turned on when RUC soil moisture is initialized from the Noah soil moisture (such as data from GFS, NAM), and turned off if it is initialized from other than Noah LSM. The default value is 0 => do not do soil moisture adjustment.

Cumulus

  • MSKF scheme: Set a hard limit for the convective time-scale to not go beyond 24 hours in the scheme. This helps with some occasional model blowups.

PBL

  • MYNN Updates: The biggest improvement is the reduction in the downward shortwave radiation bias through better cloud fraction and subgrid scale mixing ratios. Improved ensemble spread from changes to SPP in MYNN. Added a new option to mix 2nd moments and aerosols in MYNN as opposed to the scalar_pblmix option. This is activated with bl_mynn_mixscalars = 1; automatically sets scalar_pblmix = 0. Added the non-Gaussian buoyancy flux functions of Bechtold and Siebesma (1998, JAS) to improve the turbulent mixing in cloudy environments (only activated when bl_mynn_cloudpdf = 2). Added heating due to dissipation of TKE (activated by default - small impact). Important bug fix for wrf chem when transporting chemical species in the MYNN mass-flux scheme. Removed a lot of unused code, i.e., 2nd mass-flux scheme (now only bl_mynn_edmf = 1, no longer an option 2).

  • YSU, and Shin-Hong PBL schemes: fix the constant h1 in its 8th decimal place from 0.33333335 to 0.33333333. YSU now outputs exchange coefficient for momentum, exch_m.

Urban

  • BEM: Currently, the option applies air conditioning to all floors of all buildings on a model grid. Two additional urban parameter table options are added to prescribe the fraction of building (BLDAC_FRC) and fraction of floors in a building (COOLED_FRC) that have air conditioning. This reduced the air conditioning energy load and compares better with observations.

  • BEP and BEM: These multi-layer urban models pack 4D, 5D, and 6D arrays into 3D. The original code used one dimension (num_urban_layers) which depends on the largest 6D array. With no answer changes, the new code creates separate mapping for the different urban arrays so that the arrays are maximally filled for each urban array. Compared to the original code, the memory cost (relative to a non-urban run) of using UCM is reduced from 31% to 5%, BEP from 31% to 15% and BEM from 707% to 274%, equivalent to a 20%, 45% and 64% reduction in memory used, respectively. When users run BEP or BEM, new namelist values for the urban dimensions must be set (num_urban_ndm, num_urban_ng, num_urban_nwr, num_urban_ngb, num_urban_nf, num_urban_nz, num_urban_nbui) and equal the values in the BEP and BEM modules (see run/README.namelist).

Radiation

  • An error in CAM ozone time interpolation exists in WRF prior to version V4.1. The ozone interpolation is off by a month; zero value (which is wrong) and value valid at Jan 16 are used for interpolating dates from Dec 15 - Jan 16. This fix affects RRTMG radiation with o3input=2.

Microphysics

  • Thompson MP schemes (mp_physics= 8, 28): A few adjustments made for rain number concentration tendencies, two of which are bug fixes. Also a minor adjustment for consistency to minimum cloud ice size in event a balance of ice number is needed. Lastly a bug fix for partly-melted snow. The most notable changes are: (1) a larger mean rain size from cloud water to rain autoconversion that helps to produce a stronger leading edge of convective squall lines; (2) partly-melted snow was falling too rapidly while this fix produces very realistic progression of melting snow falling gradually toward rain velocity as it melts.

  • Thompson MP schemes (mp_physics = 8, 28): Most GRIB files contain mass mixing ratios for cloud/precip variables, but not all of them contain number concentrations. A good example is pressure level RAP/HRRR files whereas the native level model files contain both mass and number. Artificially initializing a model without the associated number concentration fields will cause greatly inferior assumptions at run time versus fixing the number concentration variables with something sensible during real.exe to get both initial and boundary condition 2-moment variables reasonably in sync. The solution here is creating number of cloud droplets, cloud ice, and rain for any of those that have mass but not number.

  • P3 microphysics has had several updates and a few bug-fixes since v4.0 including an important one for providing snow/ice as part of RAINNC. This brings WRF up to the P3 v3.1.11.

  • WSM6 and WDM6 schemes have been fixed to allow more freezing rain.

Cloud Fraction

  • Cloud fraction scheme (icloud=3): Compared to prior version generally produces lower cloud fraction amount at high altitude (cold tropopause level temperatures for example). Full year (every 5th day) test gives results with better shortwave radiation at ground comparison to observations.

Others

  • The time-series capability has been upgraded.
    The output option has been modified to allow output at i,j coordinate locations. One must specify i/j coordinates in the tslist and change the column headers to i/j (instead of lat/lon).
    Both real and ideal cases are capable of outputting all variables (including U/V) at the cell centers by setting the namelist option tslist_unstagger_winds = .true.
    Time series output now includes vertical velocity and pressure.

  • When nwp_diagnostics = 1, the max radar reflectivity output is not the maximum value between history output, because it is only computed once at the history output time. This change fixes it and will also turn on radar reflectivity calculation (do_radar_ref) if it isn't turned on. Note that this change has increased computational time significantly due to reflectivity being computed at every model time step.

  • In V3.8, we changed default surface_input_source option value from 1 to 3, which uses the dominant categories computed in geogrid. But we neglected that when dominant categories for land and soil are recomputed in real, the real program does checks for mismatches, making sure the lower resolution isltyp matches with landmask data. This has two effects: one is the actual isltyp, ivgtyp, and xland can be different from that from surface_input_source = 1 (a check later in real program effectively matched ivgtyp based on isltyp); and the other effect is that for certain cases, the WRF model would stop due to un-matched isltyp and ivgtyp. This change fixes this problem, and isltyp, ivgtyp, and xland are now identical to those coming from surface_input_source = 1.

  • A new soil parameter table is added (SOILPARM.TBL_Kishne_2017). The new table is based on the Kishne et al. paper. The new values reduce dry soil moisture threshold, porosity, and field capacity for some categories, and increases wilting point for all categories. In general it favors more evaporation from the soil. To test the model using this table, copy this table to the default name SOILPARM.TBL. (Kishné, Andrea Sz, Yimam, Y. T., Morgan, C. L. S., and Dornblaser, B. C., 2017: Evaluation and improvement of the default soil hydraulic parameters for the noah land surface model. Geoderma, 285, 247-259.)

  • For the initial releases of major version 4 (WRF v4.0, v4.0.1, v4.0.2, and v4.0.3) , when the option use_theta_m=1 (the default) was activated, a number of diagnostic utilities in the WRF model incorrectly used a moist potential temperature (when a dry potential temperature was required). These include: DFI, nest adjustment for T and Qv, time series (the *.TH file), the p- and z-level vertically interpolated diagnostics (impact T and Qv), and the RASM mean_output_calc and diurnalcycle_output_calc utilities. In areas of tropical moisture, the temperature errors could be in excess of 8 K (while winter-time Canadian shield errors are less than 0.5 K).

  • Since the height level interpolation routine was introduced in v3.8, it has used an incorrect formulation for the AGL computations.The AGL computation of the height-based diagnostics instead computed a height above mean sea level. In areas of high terrain, even with a 6.5 K/km lapse rate, large errors near the intended surface would have occurred. This has now been corrected.

  • A few commonly used variables (i.e., precipitable water, relative humidity, total rain, total liquid rain, potential temperature) are included in the output of traditional fields. By setting diag_nwp2=1 (&diags), one can get these variables.

  • Fixed a bug that causes AFWA diagnostics option to fail or hang at the time of writing output. Note also that the full set of AFWA diagnostics only works with a small number of microphysics options at the moment.

  • After hrrr_cycling was introduced to WRF, the stochastic solutions were not bit reproducible when compared to the solutions that used a restart. Also, with the newer GNU compilers, the size of the seed array exceeded the vertical dimensions in WRF (the vertical dimension was originally used as an easy and safe value). Both problems have been addressed.

  • For the real program, the order of the vertical interpolation option is now always set to a linear interpolation for meteorological fields that are supposed to be positive definite (for example, moist process constituents and number concentration). However, the RH field in the real program continues to use a user-defined vertical interpolation option.

  • OBSGRID input data may now be used with V4.1 real.exe, regardless of the version of OBSGRID.

WRFDA

New Features:

  • Himawari AHI radiance data assimilation
    Wang, Y., Z. Liu, S. Yang, J. Min, L. Chen, Y. Chen, and T. Zhang, 2018:
    Added value of assimilating Himawari-8 AHI water vapor radiances on analyses
    and forecasts for "7.19" severe storm over north China. J. Geophys. Res. Atmos.,
    123, https://doi.org/10.1002/2017JD027697.

  • 3DVAR and hybrid-3D/4DEnVar now work with moist potential temperature.

Updated Features:

  • Update WRFDA to handle RTTOV v12.1. New emissivity formulations are enabled as well as HDF-only emissivity atlas files. Compiling WRFDA for RTTOV now requires an HDF5 library.

  • The code-base of CRTM carried with WRFDA is updated from v2.2.3 to v2.3.0.
    Starting with V4.0, CRTM coefficient files are NOT included in any of the WRF or WRFDA tar files. There are two ways to get this data.
    1. A subset of coefficient files can be download from
    http://www2.mmm.ucar.edu/wrf/users/wrfda/download/crtm_coeffs.html
    2. The full set of CRTM coefficients may be downloaded from:
    ftp://ftp.emc.ncep.noaa.gov/jcsda/CRTM/REL-2.3.0/crtm_v2.3.0.tar.gz
    For external users, replace the link ‘var/run/crtm_coeffs’ with the directory
    contained in the appropriate tar file. A copy is also available under ~wrfhelp
    on the NCAR cheyenne supercomputer (for internal use).

  • Enhance the gts_omb_oma diagnostics by appending time slot information after the level information.

  • Allow pseudo ob at not-first time slot for 4DVAR.

  • New WRFDA namelist options for specifying errors for radar rhv(rrn/rsn/rgr)

  • New WRFDA namelist options for turning on/off some diagnostic output

  • Fix to allow assimilation of SSMI TPW contained (prior to Nov 2009) in prepbufr obs files when ob_format=1 and use_ssmiretrievalobs=.true.

Note: See http://www2.mmm.ucar.edu/wrf/users/wrfda/updates-4.1.html for a full list of updates. For more information about WRFDA, visit the WRFDA Users home page
http://www2.mmm.ucar.edu/wrf/users/wrfda/index.html

WRF-Chem

  • Correction to the mapping of MEGAN emissions of A-PINENE and B-PINENE for the CRIv2R5 gas-phase chemical scheme (those constituents were swapped).

  • It was found that the dust (and sea salt) mass balance is violated in routine settling (module module_gocart_settling.F). In particular, the total mass of the dust (and sea salt) is increasing, when the dust (and sea salt) settles from the upper to the lower layers. In the code the dust (and sea salt) mixing ratios are transferred between the layers, which is not correct, since the air density is changing. The discretization scheme was replaced to the one, which accounts for air density change and allows to conserve the dust (and sea salt) total mass.

  • Added CH4 and CO2 initialization to RACM_SOA_VBS_KPP (chem_opt=108) and RACM_SOA_VBS_KPP_AQCHEM (chem_opt=109). Added NH3 and CH4 to plumerise transport for biomass_burn_opt=1. Updated CO2 to 400 ppm.

  • The sub-grid cloud effect to the optical depth in radiation (cu_rad_feedback = .true.) requires setting additional time-averaged variables for cumulus physics (cu_diag = 1). This option now works with the Grell-Devenyi (GD) cumulus scheme, joining Grell-Freitas (GF), Grell 3D (G3), and Kain–Fritsch Cumulus Potential (KF-CuP). These changes only impact WRF-Chem runs.

HWRF

  • Operational 2018 HWRF updates to the WRF model, including:
    (i) RRTMG cloud overlap options
    (ii) Ensemble perturbations to PBL scheme (GFSEDMF)

Note: The active HWRF developers are using a fork in the NCAR organization (gitHub.com/NCAR/HWRFdev). Anyone interested in HWRF development should contact hwrf-help at ucar dot edu.

WRF-Hydro

Assets 2
Loading