A WUDAPT-to-WRF python tool that injects WUDAPT's Local Climate Zone information into WRF.
pip install w2w
Install from GitHub:
pip install git+https://github.com/matthiasdemuzere/w2w
- Check out its help:
w2w --help-
Try with the provided sample:
Sample data used here can be downloaded from the repository in the
sample_datafolder. After clicking on the file you can download it.
w2w ./sample_data lcz_zaragoza.tif geo_em.d04.nc- Deploy using your own data:
w2w INPUT_DIRECTORY YOUR_LCZ.TIF YOUR_GEO_EM.d0X.NC-
A geo_em.d0X.nc file (produced by WRF's WPS geoegrid.exe), for the inner WRF model domain in which you would like to use the LCZ-based information.
-
A Local Climate Zone map (lcz.tif) that is slightly bigger than the domain of the geo_em.d0X.nc file. There are a number of ways to obtain an LCZ map for your region of interest:
- Extract your domain from the continental-scale LCZ maps for Europe (Demuzere et al., 2019) or the United States (Demuzere et al., 2020). For more information, see here. Make sure you use the version ending with _epsg4326.tif.
- Check if your region of interest is already covered by the many LCZ maps available in the LCZ Generator submission table.
- Use the LCZ Generator to make an LCZ map for your region of interest. In case the geo_em.d0X.nc domain is larger than ~ 2.5 x 2.5°, the LCZ Generator will fail. In that case, please contact Matthias Demuzere for support.
- Your LCZ .tif and geo_em*.d0X.nc files should both live in the INPUT_DIRECTORY.
- Also, this INPUT_DIRECTORY should be writeable by the user.
- Also the geo_em.d0[0 to X].nc file(s) of the parent domain(s) should be available in the INPUT_DIRECTORY. This is needed because the
w2w.pyroutine will check whetherNUM_LAND_CATis set to 41 in all these parent domain files. If that is not the case, this will be fixed by writing out adjusted geo_em.d0[0 to X]_41.nc files. - In case you use an LCZ map produced by the LCZ Generator, by default
-lcz_band 1will be used, which is the best-quality gaussian filtered LCZ map (see Demuzere et al. (2021) for more info). - Once the adjusted geo_em.d0X.nc files are created (geo_em.d01_41.nc, ..., geo_em.d0X_NoUrban.nc, geo_em.d0X_LCZ_extent.nc, geo_em.d0X_LCZ_params.nc), make sure to rename them (e.g. rename geo_em.d01_41.nc to geo_em.d01.nc, or geo_em.d04_LCZ_params.nc to geo_em.d04.nc) before using them as input to the metgrid.exe module. See documentation for more info.
- The outputs of this tool have only been tested with the most recent WRF version 4.3.x. So we advise you to work with this version as well, which is now able to ingest the urban LCZ classes by default.
- Additional arguments to be used:
-b --built-lcz = LCZ classes considered as urban (DEFAULT: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
-l --lcz-band = Band to use from LCZ file (DEFAULT: 0). For maps produced with LCZ Generator, use 1
-f --frc-threshold = FRC_URB2D treshold value to assign pixel as urban (DEFAULT: 0.2)
-n --npix-nlc = Number of pixels to use for sampling neighbouring natural land cover (DEFAULT: 45)
--lcz-ucp = Specify a custom lookup table for the LCZ-based Urban Canopy Parameters
-
Using a custom lookup table for the LCZ-based urban canopy parameters
- you can create a custom csv-file based on this format and use it specifying it with the
--lcz-ucpflag. For the example:w2w ./sample_data lcz_zaragoza.tif geo_em.d04.nc --lcz-ucp path/to/custom_lcz_ucp.csv - A table must have this this format (indentation optional). The table displayed below is the default table.
- you can create a custom csv-file based on this format and use it specifying it with the
,FRC_URB2D ,MH_URB2D_MIN ,MH_URB2D ,MH_URB2D_MAX ,BLDFR_URB2D ,H2W
1 ,0.95 ,25 ,50 ,75 ,0.5 ,2.5
2 ,0.9 ,10 ,17.5 ,25 ,0.55 ,1.25
3 ,0.85 ,3 ,6.5 ,10 ,0.55 ,1.25
4 ,0.65 ,25 ,50 ,75 ,0.3 ,1
5 ,0.7 ,10 ,17.5 ,25 ,0.3 ,0.5
6 ,0.6 ,3 ,6.5 ,10 ,0.3 ,0.5
7 ,0.85 ,4 ,5 ,6 ,0.75 ,1.5
8 ,0.85 ,3 ,6.5 ,10 ,0.4 ,0.2
9 ,0.3 ,3 ,6.5 ,10 ,0.15 ,0.15
10 ,0.55 ,5 ,10 ,15 ,0.25 ,0.35
11 ,0 ,0 ,0 ,0 ,0 ,0
12 ,0 ,0 ,0 ,0 ,0 ,0
13 ,0 ,0 ,0 ,0 ,0 ,0
14 ,0 ,0 ,0 ,0 ,0 ,0
15 ,0.95 ,0 ,0 ,0 ,0.05 ,0
16 ,0 ,0 ,0 ,0 ,0 ,0
17 ,0 ,0 ,0 ,0 ,0 ,0
A citable documentation is in preparation for The Journal of Open Source Software. The draft version can be accessed here.
An important objective of WUDAPT, the World Urban Database and Access Portals Tools community project, is to generate urban canopy information and provide the (open-source) tools to facilitate urban-focused modelling studies (Ching et al., 2018).
Since the work of Brousse et al. (2016), the level-0 WUDAPT information, the Local Climate Zone maps, have been used increasingly in WRF, the community “Weather Research and Forecasting” model. Their original guide and code on how to use WUDAPT information into WRF (originally designed for WRF v3.2) is available here. Note that this tool was first assigning the LCZ mode to each WRF grid cell, and only afterwards assigning corresponding morphological, radiative and thermal properties to this modal LCZ class. This is done differently in w2w, see below.
As of spring 2021, WRF v4.3.x is able to ingest LCZ information by default (previous versions required manual WRF code changes by the user). See more details on "Updates of WRF-urban in WRF 4.3: Local Climate Zones, Mitigation Strategies, building materials permeability and new buildings drag coefficient" here. Because of this, we decided to simultaneously built an improved WUDAPT-to-WRF routine, to make the translation of LCZ-based parameters better and more simple. As before, the LCZ-based urban canopy parameters generally follow the values provided by Stewart and Oke (2012) and Stewart et al. (2014).
The procedure in this new w2w tool is different from the former tool. Morphological parameters are assigned directly to the high-resolution LCZ map, and only afterwards aggregated to the WRF grid. In this way, the method produces a unique value of the different urban morphology parameters for each model cell. This was found to be more efficient in reproducing urban boundary layer features, especially in the outskirts of the city (Zonato et al., 2020), and is in line with the WUDAPT-to-COSMO routine (Varentsov et al., 2020). Other radiative and thermal parameters are for now still assigned to the modal LCZ class. More details on the procedure and its assumptions will soon be available here.
The project is licensed under the MIT license.
Contributions to w2w are welcome! This is how:
-
Bugs: If you find a bug, please report it by opening an issue. if possible, please attach the complete error/Traceback, the
w2wversion used, and other details like theWRFversion. -
Fixing Issues: If you want to contribute by fixing an issue, please check the issues: contributions are welcome for all open issues especially those with labels
bug,help wantedorgood first issuefor easy contributions. -
Enhancement: New features and modules are welcome! You can check the issues: contributions are welcome for open issues with labels
enhancementandhelp wanted.
- Matthias Demuzere: Lead developer
- Daniel Argueso: WRF expert
- Andrea Zonato: WUDAPT-WRF expert and liaison
- Jonas Kittner: Python wizard
- We appreciate the feedback and suggestions provided by Alberto Martilli and Oscar Brousse, lead developers of the original Fortran-based WUDAPT-to-WRF fortran package.
- Thanks to Alberto Martilli and Alejandro Rodriguez Sanchez for allowing us to use of their Zaragoza case-study files as sample data.