Introduction to WRF
WRF Model Overview
The Advanced Research WRF (ARW-WRF) 歷經多年大氣學術社群及相關產業社群開發。WRF 模式堪稱一套靈活、最先進的 (最尖端的) 大氣模擬系統，在平行計算平台上運行高效且具可移植性。WRF 模式的應用範圍涵蓋各種尺度，從數公尺到大氣層的廣闊範圍皆能使用，在地理研究、氣候研究、天氣預報等領域發揮著重要作用。最新版本的原始碼可在 WRF GitHub Repository ( wrf-model/WRF ) 取得。
不僅如此，WRF 模式還允許使用者針對不同情境，選擇合適的物理參數化方案，例如地表模型、雲微物理方案、邊界層參數化方案和積雲參數化等，使模擬的结果更加貼近真實大氣狀況。
Workflows
這是 WRF 模式簡要的執行流程圖。理想個案模擬(中上深綠)可以只安裝/使用 WRF 模式即可，真實個案模擬及預報需要使用 WPS 先處理好靜態地形資料及三維氣象場資料(左中二項淺綠)，觀測資訊可以透過Nudging 等方法調整氣象場資訊(中下藍色)，亦能以資料同化系統 (Data Assimilation System; 右下藍色) 將觀測資訊適當地放入模式中。
WPS：將全球模式的網格資料處理成符合 WRF 區域模式網格設定的等壓面層場並結合靜態地理資料。
REAL：初始化，將 WPS 產出的等壓面資料處理成 WRFV3.X 預設的地勢追隨座標 (Eta；terrain-following coordinate) 或 WRFV4.0+ 預設的 Hybrid Vertical Coordinate (HVC) 初始場及邊界條件。 HVC 在越靠近地表時越接近地勢追隨座標，越到高空則越接近等壓面。相關技術細節請參閱：details 。
WRF：模式的主要部分，由目前的解和邊界條件解出下一個時間步長的動力和熱力方程組的數值解。
New Features & Options Since WRFV3.9
WRFV4.5+：
A new package to compute two-moment prognostics for graupel/hail and a predicted density graupel category is added in the Thompson scheme (mp_physics=38). Other updates to the scheme include a change to the Y-intercept relationship for one-moment graupel and replacement of air temperature for wet-bulb temperature in riming and mixed phase processes (Jensen et al., in review). This code requires the datafile "qr_acr_qg_mp38V1.dat" to run, which can be found on the file system for the NCAR community HPC "Cheyenne" under /glade/work/wrfhelp/WRF_files/ and online at http://www2.mmm.ucar.edu/wrf/src/wrf_files/. Details
A scale-aware or grid-distance-dependent option is added to the new Tiedtke scheme. The code is based on the paper by Wang (2022) with two minor changes removed. The option becomes active automatically when grid sizes is less than 15 km. PR1806 and PR1840
A new k-epsilon-theta^2 PBL scheme, including two additional prognostic equations for dissipation rate and temperature variance, has been included in the WRF model (see Zonato et al.,2022:A New K–ε Turbulence Parameterization for Mesoscale Meteorological Models. Monthly Weather Review, DOI: https://doi.org/10.1175/MWR-D-21-0299.1). This scheme is designed for finer grids. It works with sfclay=2(time step = 90), sfclay=1(time step = 60), diff_opt=1, km_opt=4). Details
Add support to build WRF on ARM64 (aarch64) architecture systems with the NVIDIA HPC SDK Compilers (nvfortran,nvc). Details
Provided AMD AOCC Compiler support at the time of configuring WRF for AMD Architectures including Zen1, Zen2 and Zen3.
WRFV4.4+：
The capability to couple WRF and CMAQ (an air quality model developed at the US EPA) is added. As a coupled model, WRF-CMAQ can be run in a one-way fashion (i.e., meteorological information will be transferred to CMAQ directly to drive its calculation), or in a two-way fashion, (i.e., the aerosol information from CMAQ will be fed back to the RRTMG radiation physics in WRF). This is available beginning in WRFV4.4 and CMAQ v5.3.3.3. Refer to the EPA's WRF-CMAQ Model webpage for instructions.
link The ability to write compressed NetCDF4 files in parallel, via NetCDF V4.7.4+ is added. Performance is slower than pnetcdf, but can be notably faster than regular NetCDF on parallel file systems. As expected, compression provides files significantly smaller than pnetcdf generates.
WRF's Special