WRF-Fire

From openwfm
Jump to navigation Jump to search
This is a page with basic information about a software. Please see Template for creating software pages for what should be here.

WRF-Fire combines the Weather Research and Forecasting model (WRF) with a fire code implementing a surface fire behavior model, called SFIRE, based on semi-empirical formulas calculate the rate of spread of the fire line (the interface between burning and unignited fuel) based on fuel properties, wind velocities from WRF, and terrain slope. The fire spread is implemented by the level set method. The heat release from the fire line as well as post-frontal heat release feeds back into WRF dynamics, affecting the simulated weather in the vicinity of the fire. The fire code is written in Fortran 90 following WRF coding conventions. It is integrated as a physics option, called from WRF as a subroutine. It calls WRF libraries for utilities such as I/O and communication between MPI processes. The fire code executes on a part of the domain, called a tile (in WRF nomenclature). All communication between the tiles is in the caller; thus, one time step requires multiple calls to WRF-Fire.

WRF-Fire is essentially a reimplementation of CAWFE with the weather model replaced by WRF and with the fire spread implemented by the level set method in SFIRE. The subroutines for the computation of the rate of spread and for the insertion of the heat flux to the atmosphere are taken from CAWFE. Click on the following poster for the origins and a summary of current capabilities of WRF-Fire:

Agu-poster-8mp.gif

Distribution

WRF-Fire is public domain software, released under the WRF public domain notice and disclaimer.

Our goal is to transfer the development code to the WRF download yearly. We are making the development version publicly available in hope that it will be useful; the code has been developing quite rapidly, with important features and an occasional bug fix added between WRF releases.

Development version

This version also contains a number of additional tools such as Matlab scripts for diagnostics, and the latest features and bug fixes. See How to get WRF-Fire and How to run WRF-Fire for installation instructions. The head of the master branch should contain tested stable code at any time.

From WRF release

WRF-Fire is included in WRF since version 3.2, released on April 2, 2010.

See the NCAR WRF-Fire page for further information.

Contact

  • Jonathan Beezley, Janice Coen, Jan Mandel

File format

All input, output, and restart files (with the complete model state) are in NetCDF format.

Programming language and environments

Fortran 90 with CPP preprocessor. Part of the WRF code is generated by C programs from a description in the registry.

Documentation

Technical description

User's guides

Support

Email support

Questions regarding the development version of WRF-Fire can be sent to the maling list at http://mailman.ucar.edu/mailman/listinfo/wrf-fire (subscription required), or email the developers directly. When you write us, please:

  • Identify the version (WRF version if from the WRF release, the first line of the output of git log if your code is from the git repository)
  • Make sure you test the issue first on the code exactly as you have received it and the code version is up to date. If you obtained the code by git (highly recommended) as described in How to get WRF-Fire,
    • git diff should return no output
    • please verify that you are on the latest commit on the master branch from the repository.
  • Recompile the code from scratch after typing clean -a first
  • Send us sufficient information to identify and reproduce the problem if needed (output of wrf compilation, namelist.input, output from the run)

Please contact wrfhelp@ucar.edu with any questions regarding the version of WRF-Fire in the official WRF release.

Mailing list

For further support, WRF-Fire announcements, questions, and discussions, please subscribe to the WRF-Fire mailing list at http://mailman.ucar.edu/mailman/listinfo/wrf-fire.

Wiki

We welcome contributions and discussion on the pages of this wiki. Please see the Main Page for How to get an account. The WRF-Fire wish list is recommended for requests and Talk:WRF-Fire wish list for further discussions regarding future developments of WRF-Fire.

Publications

See WRF-Fire publications.

Contributors

  • Janice Coen (NCAR) developed the physics components of the fire model in CAWFE (Coen (2005) and Clark et al. (2004)), in particular the fire spread rate and the heat flux insertion modules, which were adopted into WRF-Fire with no substantial changes.
  • Ned Patton (NCAR) ported the fire code from (Clark et al., 2004) and interfaced it with WRF (Patton and Coen, 2004).
  • Minjeong Kim and Jan Mandel (University of Colorado Denver) have identified a version of the level set metod suitable for fire spread, and developed a prototype Matlab code.
  • Jan Mandel (University of Colorado Denver) is the lead programmer of WRF-Fire. He has implemented the fire spread by the level set method as a parallel WRF-compliant code with assistance from Jonathan Beezley and Minjeong Kim, starting from Ned Patton's WRF interface and using the subroutines for the spread rate computation and the insertion of heat fluxes into the atmosphere from CAWFE.
  • John Michalakes (formerly NCAR, now NREL) modified WRF to support refined grids (submesh) for the fire code.
  • Jonathan Beezley (University of Colorado Denver) has further modified WRF to support the fire software, provided the software engineering infrastructure, has set up and maintains the git repository, maintains synchronization with WRF changes, and developed the modified version of WPS for WRF with the fire model to enable the use of real data.
  • Volodymyr Kondratenko (University of Colorado Denver) has improved memory handing in the computation of remaining fuel.
  • Adam Kochanski (University of Utah) has contributed variable atmospheric bubble initialization and suggested the walking line ignition. He is currently leading the validation effort.
  • Kara Yedinak (Washington State University) has improved in fire initialization. See discussion

See also

External links