请选择 进入手机版 | 继续访问电脑版
爱气象,爱气象家园! 

气象家园

 找回密码
 立即注册

QQ登录

只需一步,快速开始

新浪微博登陆

只需一步, 快速开始

搜索
查看: 19852|回复: 43

WRF-V4.0发布,也许wrf的使命完成了

[复制链接]

新浪微博达人勋

发表于 2018-6-9 15:34:20 | 显示全部楼层 |阅读模式

登录后查看更多精彩内容~

您需要 登录 才可以下载或查看,没有帐号?立即注册 新浪微博登陆

x
本帖最后由 andrewsoong 于 2018-6-10 15:36 编辑

WRFv4.0悄悄公布了,据说这是最后一个版本。更新如下:http://www2.mmm.ucar.edu/wrf/users/wrfv4.0/updates-4.0.html
WRF Model Version 4.0: UPDATES  WRF Model Update  
The WRF model tar file has been updated to Version 4.0 on June 8, 2018.
For known problems in V4.0, please click here.  The WRF Pre-Processing System (WPS) has been updated to Version 4.0.  
Note: V4.0 WRF can use wrfinput and wrfbdy files from earlier versions of the model. But one must use appropriate options to do so (see below). If one would like to try the latest options, one must re-run program real.exe from this version of the code.  As always, be cautious when using new options.  If you are interested in seeing how V4.0 has been tested, click here.
New in Version 4.0:  Dynamics:  
* The hybrid sigma-pressure vertical coordinate introduced in V3.9 is now the default. It can be turned of by using namelist option hybrid_opt=0. See doc/README.hybrid_vert_coord for details.  
* Temperature variable is now moist theta by default. This has been an option since V3.7, controlled by namelist use_theta_m.  Physics  Microphysics:  
* P3 (Predicted Particle Property scheme) microphysics: Added an option for including the second free array for ice. Added a new data file p3_lookup_table_2.dat. Activated by using mp_physics = 52. (Thanks to Morrison of NCAR)  
* Thompson-Eidhammer microphysics (mp_physics = 28): Added a surface dust emissions scheme (near copy of GOCART) to link into the ice-friendly aerosol variable. This option is activated by namelist option dust_emis = 1. (Thanks to Thompson, M. Xu and T. Eidhammer of NCAR)
Cumulus/Microphysics:  
* An option is added use CESM-NCSU (North Carolina State University) prescribed RCP4.5 climatological aerosol data (spatially and temporally varying) to interact with cloud simulated by MSKF (cu_physics=11) and a special aerosol Morrison scheme (mp_physics=40). A subgrid-scale cloud microphysics scheme of Song and Zhang (2011) is added in MSKF, and the aerosol activation and ice nucleation parameterizations of Song and Zhang (2011) are added into a version of Morrison scheme (mp_physics = 40). The cloud produced by the MSKF is passed onto radiation. Radar reflectivity from the microphysics in MSKF can be output as mskf_refl_10cm. The options are activated by namelist aercu_opt (= 0, no aerosol interaction; = 1, interaction with MSKF only; = 2, interaction with both MSKF and Morrison). The aerosol concentration can be changed by a namelist aercu_fct (default value is 1.). The aerosol comes in on 30 pressure levels, and has 10 types. The input data can be found on Cheyenne in /glade/p/work/wrfhelp/WRF_files, and on the web at http://www2.mmm.ucar.edu/wrf/src ... erosol_Data.tar.gz. A user needs to link one of these files to CESM_RCP4.5_Aerosol_Data.dat before running. (Thanks to Timothy Glotfelty, Patrick Hawbecker, and Kiran Alapaty of EPA)  
Cumulus:  
* KSAS (cu_physics=14): A scale-aware verson of NSAS. Should be run with shcu_physics = 4, the NSAS shallow scheme, which is part of the NSAS scheme. The old NSAS scheme has been moved to cu_physics = 96. (Thanks to Korea Instituted of Atmospheric Prediction Systems or KIAPS and Jihyeon Jang of NCAR).  Land-Surface Model:  
* Noah-MP:  - Gecros crop model (Genotype-by-Environment interaction on CROp growth Simulator, Germany) is added to NoahMP and can be activated by option opt_crop = 2. (Thanks to J. Ingwersen of U. Hohenheim and M. Barlage of NCAR) - Added capability for Noah-MP to read soil composition data and use pedotransfer (PT) function to determine soil properties (opt_soil in namelist). PT function coefficients are controlled from MPTABLE. Users can add additional PT functions controlled by opt_pedo in namelist. Special version of GEOGRID.TBL and datasets added in WPS. (Thanks to M. Barlage of NCAR)  - Variables for NoahMP is now in a separate registry file, registry.noahmp (Thanks to M. Barlage of NCAR)  Radiation:  
* RRTMK (ra_sw_physics=14, ra_lw_physics=14): A version of RRTMG improved by KIAPS: G-packed McICA was sped up without losing accuracy, and two-stream approximation (TSA) for scattering process is revised. Overall speed of the code isn't changed. (Thanks to Korea Instituted of Atmospheric Prediction Systems.  Reference: Baek, S. (2017), A revised radiation package of G-packed McICA and two-stream approximation: Performance evaluation in a global weather forecasting model, J. Adv. Model. Earth Syst., 9, doi:10.1002/2017MS000994)  Improvements and Bug fixes:  Physics:  RAP/HRRR Physics: (Thanks to Joe Olson, Tanya Smirnova, and others at  GSD/NOAA)  
* MYNN PBL (bl_pbl_physics=5) and MYNN surface layer (sf_sfclay_physics=5):  (a) Enhancement of mass-flux scheme, tweak of mixing length and stratus component of subgrid clouds. Changed default option to EDMF and mixing length option 2. (b) SPP fixes: SPP turned off for subgrid cloud decay;  and missing height information for length scale is added. (c) MYNN surface layer: Changed 2 m exchange coefficient calculation which affects 2-m diagnostics, but not model solution; (d) Switched to use THv for TH
* calculation to be consistent with RIB calculation and further relaxed some limits on PSIT, PSIX, and UST following Jimenez et al. (2012). Small impact. Using Li et al. (2010, BLM) for RIB -> z/L mapping. Changed default z0,zt,zq over water to be COARE 3.0 (was 3.5 in v3.9).  
* RUC LSM (sf_surface_physics = 3):  (a) Changed empirical formulation of frozen precipitation density to match better observed snow accumulation on the ground;  (b) Applied snow "mosaic" approach only when temperature at the first atmospheric level is above 271K. This change prevents from two warm skin temperature over shallow snow;  (c) Modified 2-m diagnostics of temp and mixing ratio to not use 'flux' method.   
Microphysics  
* NSSL scheme: (Thanks to Ted Mansell of NSSL)  - Added a new default ice crystal fall speed formula (icefallopt = 3), which has faster speeds for small ice particles. The main effect is on anvil clouds to help them settle a bit faster. The old behavior can be recovered with icefallopt=1;  - Added an option to change ice crystal and snow fall speeds by an arbitrary factor, but default factors are 1.0 (no impact with default settings); - Added a calculation of reflectivity-weighted snow fall speed instead of using mass-weighted (for size-sorting limiter); - Added a check for large supersaturation w.r.t. liquid at ice-only temperatures and nucleate/condense if needed; - Added option to limit accretion of droplets smaller than "exwmindiam" by setting a "reserve value" that is excluded from accretion so that a minimum value of mass and number remains. (Default is no limit, i.e., exwmindiam = 0.0 ); - Added two options for rime density on graupel/hail (irimdenopt). Default is original Heymsfield and Pflaum formula. Extra options are for Cober and List (1993) and Macklin; - Added a rime density calculation for rain accreted by graupel instead of always assuming high density; - Added a calculation of a separate rate for number of rain drops produced by snow melting rather than using the snow number loss rate; - Separated rain-ice freezing from flag ifrzg, and is now controlled by ifiacrg (no impact -- for sensitivity study only) - Snow aggregation coefficient set back to 1.0 (increases aggregation rate); - Made consistent use of dtpinv to replace divisions by dtp (no impact).  
* WDM6 (mp_physics = 16): Added dependency of qcr (cloud and rain water conversion rate) on land and ocean. The lower limit of nccn is set to clean air value over ocean, and this improves effective radius calculation and radiation feedback (From KIAPS).  
* Morrison schemes (mp_physics = 10, 40): A new namelist, morr_rimed_ice, is added to control graupel or hail option for these schemes. This was controlled by hail_opt. The default value is set to use hail (1). (From H. Morrison)  Planetary boundary layer and surface layer physics:  
* Q2 diagnostics: limited to 5% over lowest model level QV to avoid unrealistic spikes at sunrise and sunset times.  
* ACM2 (bl_pbl_physics=7): used height instead eta values to compute layer structure. This is more consistent with MPAS, but also gives some improvement for 2 m T in WRF (Thanks to R. Gilliam and J. Pleim of EPA)  
* MM5 surface layer (sf_sfclay_physics = 1, 91): added a option to use Czil = 0.1 by setting iz0tlnd = 2. This provides a way to change the sensitivity of exchange coefficient.  
* MYJ suface layer (sf_sfclay_physics = 2): Allowed to work with LES by providing frictional velocity for momentum from the scheme. ARW only.  Land Surface physics:  
* NoahMP: a patch communication is needed for the groundwater model with generic dataset. The bug was introduced in 3.9. (Thanks to M. Barlage of RAL/NCAR).  
* Noah: output variable CANWAT is incorrectly scaled on output by 1000. No effect to model solutions (Thanks to P. Campbell of EPA).  
* Several variables, RA (aerodynamic resistance), RS (stomatal resistance), WSPD, LAI_mosaic, RS_mosaic, are diagnosed from Noah, Noah-mosaic and NoahMP and can be added to the output for CMAQ applications (EPA). (Thanks to P. Campbell of EPA)  
* Urban physics: Added variable TSK_RURAL for BEP/BEM with Noah to prevent possible model failure when aggregated TSK is disaggregated (Thanks to Alberto Martilli of CIEMAT, Spain)  
* PX LSM: removed arbitrarily high values of vegetation fraction and leaf area index computed internally for rangeland, scrubland etc. in western US. This improves the moist and cool biases and high precipitation biases. Also added data for 21 category MODIS. (Thanks to R. Gilliam of EPA)  
* SSiB LSM: fixed a bug in soil layer initialization.  
Ocean:  
* sf_ocean_physics = 1: Allowed simple mixed layer ocean model to be turned off in DFI, and turned back on in forward integration.  
Radiation:  
* Two fixes for using CAM climatological ozone (o3input=2) with RRTMG: 1) if one uses physics suite specification in V3.9, then o3input is mistakenly reset to 0 for all domains (This means the constant ozone profile is used); 2) if one uses a nest, the nest didn't use any ozone data. This bug was introduced in V3.9 as well.  
* Clear-sky surface diagnostics for direct and direct normal surface solar radiation are added (Thanks to Pedro Jimenez of NCAR).  
* Use of GHG: The model will now stop if input file CAMtr_volume_mixing_ratio is not found, rather than reverting to default values. (Thanks to Claire Carouge of ARC Centre of Excellence for Climate Systems Science, Australia)  
Others:  
* Gravity wave drag (gwd_opt=1): 1) Gravity wave drag generalized for all projections. Winds are rotated to lat/long at beginning of GWD code so that orographic data from geogrid can be used accurately. Tendencies are rotated back afterwards. The stop when the grid is rotated too much relative to lat/lon, which was introduced in V3.9, is removed. 2) Fixed a problem with excessive orographic stress in unstable conditions (Thanks to M. Toy of NOAA and Hyun-Joo Choi of KIAPS).  
* WRF-Fire updates. (Thanks to Domingo Munoz-Esparza of RAL/NCAR)  - Proper parallelization of the entire level-set algorithm using MPI - 3rd-order Runge-Kutta time integration scheme for the level-set equations - 3rd- and 5th-order weighted essentially non-oscillatory (WENO) schemes for the discretization of the advection term in the level-set equations [also 2nd- and 4th-order schemes are available] - A hybrid-order level-set method with locally reduced artificial viscosity (high-order over a local band - Reinitialization PDE for the level set (also using hybrid-order discretization) - Transport and dispersion of fire-generated smoke as a passive tracer - Diagnosed flame length - Option to extrapolate mid-flame height winds to reduce over-coupling of atmospheric fields impacted by the fire - Initialization from observed fire area - Scott & Burgan (40) fuel model crosswalk to Anderson (13) - Several bug fixes - Modified test/em_fire/namelist.input* files.  
* Hailcast option (hailcast_opt = 1): Added haildt to avoid computing at every model time step; Added the use of the adiabatic liquid cloud water profile to account for typical convective-allowing resolutions not fully resolving the precipitation-free bounded weak echo region; the output variable is changed to maximum hail size rather than the mean and stand deviation; The model will stop if this option is used in a domain where a cumulus scheme is activated (Thanks to Becky Adams-Selin of AER and Conrad Ziegler of NSSL/NOAA).
* Spectral nudging of water vapor: added and controlled by namelist. A capability to place an upper-layer limit on spectral nudging of QV and T via a new namelist variable "ktrop" is added too. (Thanks to Tanya Spero of National Exposure Research Lab).  
* Obs-nudging: Fixed a bug for variable reserf when model levels are greater than 100 (Thanks to B. Reen of Army Research Laboratory)  * output_diagnostics = 1: Fixed a potential issue in climate extreme computation where roundoff errors may result in negative value for square root computation. (Thanks to C. Carouge of ARC Centre of Excellence for Climate Systems Science, Australia)  
* Physics sharing with MPAS: MPAS directives are added to a few physics options (new Tiedtke CPS, YSU PBL, Noah LSM, WSM6) so that potentially it will allow MPAS to get the latest versions of WRF physics.
Initialization/Input/Output:  
* Added a new way to computed vertical levels, which has a smooth variation of dz (see User's Guide, Chapter 5 for details). This method is the default and can be changed by the new namelist auto_levels_opt  (= 1 gives the old set of levels). Together with namelists dzbot, dzstretch_s, dzstretch_u (new) and max_dz, one can adjust to give you different number of vertical levels.  
* force_use_old_data: Set to true, if using wrfinput data from older versions of the model. This is a new namelist in &time_control. When using old input data, one should also set hybrid_opt =0 and use_theta_m = 0 in &dynamics, since the defaults for V4 are hybrid_opt =2 and use_theta_m = 1. If this option isn't active, checks in the model will make suggestions to you.  
* Because of the re-organization of WPS static input data, checks are put in program real to make sure the required input for a given option is available. Otherwise the program will stop. Older versions of WPS files are exempt from this check.  
* A general module_initialize_ideal.F now handles ideal cases quarter_ss, squall2d_x, squall2d_y, b_wave, les, seabreeze2d_x, convrad, hill2d_x and grav2d_x. The case is determined by namelist variable ideal_case (added to namelist template in each ideal case directories). Compile-time selection is not changed, since it still needs to know which test directory to link executables to.  
* force_ideal.nc in test/em_scm_xy/: String 'V4' is added to the TITLE of global attribute so that it can be used by V4 model.  * real and wrf: in V3.9, checks were added for mixed layer ocean and input, gwd and input, urban and input to catch problems with using those options but no input data. But checks also stop the model if there is input for a particular option, but the option was turned off. This is fixed in V4.0.  
* Urban fraction initialization: To use new NLCD FRC_URB2D input file (see WPS updates) from WPS/geogrid program properly in the model, a match between this data and other datasets (such as MODIS or USGS landuse, which may cover the non-urban areas) is put in.  
* The default input SNOALB is changed from old 0.15 deg NCEP data to new 0.05 deg MODIS snow albedo data. If one uses usemonalb = .true. (default value is false), the default dataset from geogrid is changed from 0.15 deg NCEP data to new 0.05 deg MODIS data.  
* RSL files: Removed non-fatal messages from these files.  
* A mechanism to output a small selected fields is added. This writes to auxiliary stream 1. It writes total pressure, total geopotential height, temperature, pressure, rotated u and v (earth relative), wind speed, mslp. This can be activated by using namelist option diag_nwp2 = 1. This option will be developed further in the coming months.  
Dynamics:  
* diff_6th_opt was fixed for conservation. The option can be turned off if the terrain slope exceeds a namelist specified threshold value (diff_6th_slopeopt=1, and diff_6th_thresh=0.1).  
* Options are added to turn on/off 2nd and 6th order diffusion for all 4D arrays.  
* Diffusion for lat/lon grid: corrected map-scale factors for two terms. (Thanks to GSD/NOAA).  Namelist Changes:  
* Added a new namelist template, namelist.input.pbl-les, in test/em_real directory for running model from km to LES scales.  
*  Simplied to reduce user errors: removed many variables those values cannot be changed. Examples are num_soil_layers, spec_zone, relax_zone (can be derived from spec_zone and spec_bdy_width), {start, end}_{minute,second}, debug_level,  
* e_vert in all real-data namelist templates are increased to 33 based on the new default method to compute eta levels. If you set eta_levels, there is no restrictions. But caution must be used to ensure there are adequate number of vertical levels.  
* damp_opt for real-data templates are changed from 0 to 3.  
* num_metgrid_levels in all real-data namelist templates are changed from 27 (number of old GFS data levels) to 32 (new GFS data levels).  
Removed:  
* SLOPECAT field: removed from WPS and model.  
* directory dyn_exp/: removed.  
* run_restart.tar in test/em_b_wave/: removed.
Software:  
* The top model directory is now WRF/.  
* The default build is now netCDF4 with compression if netCDF4 is built with HDF5 library. This allows WRF netCDF input and output files written with compression. If the desired library doesn't exist, it reverts to classic netCDF. It is noted that the compression may be slower than non-compression writes, but the advantage is it writes files with smaller sizes.  
* Determine the resultant x- and y-direction computational patch sizes based on the full domain size and the number of MPI ranks used in each direction. If the patch size < 10 grid cells in either direction, the model will stop.  
* Support for OpenMPI version 1 is removed..  
* Cleaned up configure and compile scripts.

评分

参与人数 1金钱 +2 收起 理由
洋溢 + 2

查看全部评分

密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-9 15:43:41 | 显示全部楼层
{:5_275:}
密码修改失败请联系微信:mofangbao
回复

使用道具 举报

新浪微博达人勋

发表于 2018-6-9 17:40:39 | 显示全部楼层
{:eb301:}{:eb301:}
密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-9 18:37:58 | 显示全部楼层
开始变分网格咯
密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-10 09:44:43 | 显示全部楼层
后面谁会替代上来呢?
密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-10 10:12:53 | 显示全部楼层
而我才刚刚开始接触wrf
密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-10 10:12:57 | 显示全部楼层
而我才刚刚开始接触wrf
密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-10 10:28:48 | 显示全部楼层
宋大神,你也不排一下版
密码修改失败请联系微信:mofangbao

新浪微博达人勋

 楼主| 发表于 2018-6-10 15:36:42 | 显示全部楼层
尽头的尽头 发表于 2018-6-10 10:28
宋大神,你也不排一下版

刘菊教训的是
密码修改失败请联系微信:mofangbao

新浪微博达人勋

发表于 2018-6-10 16:42:32 | 显示全部楼层
好像初始场数据格式要有新变化?
密码修改失败请联系微信:mofangbao
您需要登录后才可以回帖 登录 | 立即注册 新浪微博登陆

本版积分规则

Copyright ©2011-2014 bbs.06climate.com All Rights Reserved.  Powered by Discuz! (京ICP-10201084)

本站信息均由会员发表,不代表气象家园立场,禁止在本站发表与国家法律相抵触言论

快速回复 返回顶部 返回列表