WRF/WPS comes with a standard distribution of geogrid input data. The data is available globally at a resolution of at most 30 arc seconds (a little less than 1 km in NS-direction). For WRF-Fire, however, two additional variables at higher resolution are needed: ZSF for terrain elevation and NFUEL_CAT for the categories of the fuel model, by default the 13 categories of the Anderson model. The source data for these variables needs to be manually downloaded, converted and added to the geogrid configuration. The description here is based on the notes on the WRF-Fire User Guide on the OpenWFM Community Wiki.
The additional input data should be available on a resolution as close as possible to that of the fire grid (5-10m, typically). 1” data is acceptable (30 m in NS-direction), 1/3” data is better. (1” latitude corresponds to approx. 30 m, but 1” longitude in Interior Alaska is reduced by a factor of the cosine of the latitude, to approx. 13 m.) In the following, NS-extensions will be indicated.
The USGS maintains a list of freely available elevation data (US and globally) including links to download tools and sites.
For Alaska, the options are:
Unfortunately, the otherwise interesting Shuttle Radar Topography Mission (SRTM) 1” data is not available above the 60th degree latitude.
The ASTER DEM option appears to be the easiest to use. The 1” NED should integrate the same data (as it is a product composed of the best available data), and downloading GeoTIFF files from the National Map site turns out to be not always straightforward: even when the option is offered, sometimes GridFloat files are provided.
1” fuel category data for the entire US (including Alaska) are available from the USGS Landfire data distribution site. The steps to download are:
Other than the Anderson model, the 40 categories from the Scott and Burgan model are available as well.
At this stage, the potentially multiple GeoTIFF source files for the ZSF and NFUEL_CAT variables need to be converted into the intermediate format used by
WPS/geogrid.exe. Fortunately, the authors of WRF-Fire have provided a utility program for this purpose, with high-level and detailed instructions how to install and use it.
Before it can be used, however, it is necessary to combine the downloaded GeoTIFF files of each dataset into a single file. The easiest way to do this is via the free gdal_merge.py utility from the Geospatial Data Abstraction Library (GDAL). On Mac OS X, GDAL can be installed via Homebrew. Using the Python utilities requires setting the
PYTHONPATH variable – pay close attention to the installation script outputs (or
brew info gdal).
Now the single GeoTIFF file can be converted to the geogrid intermediate format as described on the OpenWFM WRF-Fire User Guide, for example:
~/util/convert_geotiff.x -c 13 -w 1 -u "fuel category" -d "Anderson 13 fuel categories" myfuelcats.tiff
The output will be a set of directories containing binary files and a file called
index containing the metadata. The index file should be checked to compare to the one in the WRF-Fire User Guide.
The elevation and fuel data are then copied each to its own directory, and the location has to be provided to
geogrid.exe together with instructions for how to generate the variables. As an example setup, the default geog data, met data (narr) and additional geog data can be placed in directories off the same level as one's WRF installations (if several versions are used). Here, high resolution data for separate domains are placed each in its own directory and the currently used data set is linked symbolically:
pacman9:wrf_fire waigl$ ls -lAF total 44 drwxrwxr-x 4 waigl uafsmoke 4096 Oct 10 17:26 geog/ lrwxrwxrwx 1 waigl uafsmoke 19 Oct 30 21:41 geog-extra -> geog-wrffire-alaska/ drwxrwxr-x 2 waigl uafsmoke 4096 Oct 19 13:49 geog-wrffire-alaska/ drwxrwxr-x 5 waigl uafsmoke 4096 Oct 30 21:41 geog-wrffire-testcase/ drwxrwxr-x 2 waigl uafsmoke 12288 Oct 2 17:47 narr/ drwxr-xr-x 8 waigl uafsmoke 4096 Oct 6 13:09 WRFF-GH-20121006/ drwxr-xr-x 4 waigl uafsmoke 4096 Sep 25 17:09 WRFV3.3.1/ drwxrwxr-x 4 waigl uafsmoke 4096 Oct 10 16:51 WRFV34/ drwxrwxr-x 3 waigl uafsmoke 4096 Oct 10 15:42 WRFV341/ pacman9:wrf_fire waigl$ ls -lAF geog-extra/ total 0 lrwxrwxrwx 2 waigl uafsmoke 25 Oct 6 13:08 dem_data/ lrwxrwxrwx 2 waigl uafsmoke 30 Oct 6 13:08 landfire_data/ lrwxrwxrwx 1 waigl uafsmoke 25 Oct 19 13:18 ned_data -> dem_data/
geog_data_path = ../../geog/. And the
GEOGRID.TBL is edited to add the additional variables on the fire subgrid as follows:
=============================== name=NFUEL_CAT priority = 1 dest_type = categorical dominant_only = NFUEL_CAT z_dim_name=fuel_cat fill_missing = 14.0 interp_option = default:nearest_neighbor abs_path=../../geog-extra/landfire_data subgrid=yes ============================== name=ZSF priority = 1 dest_type = continuous df_dx=DZDXF df_dy=DZDYF smooth_option = smth-desmth_special; smooth_passes=1 interp_option = default:average_gcell(4.0)+four_pt+average_4pt abs_path=../../geog-extra/ned_data subgrid=yes ==============================
Note in particular the
fill_missing = 14.0 and
interp_option = default:nearest_neighbor: The fuel data will have missing values in locations where there is no fuel, in particular for rivers and other water surfaces. With more sophisticated interpolation schemes, bogus data may be created in these cells. It is better to fill those cells with a pre-selected missing value (14, when using the 13 Anderson fuel behavior model categories) than to interpolate or let WPS set its default value (10^20).
Examples of elevation data for a 30×30 km domain around the 2012 Dry Creek fire south-east of Fairbanks: