Skip to content

The wrfpp post-processor

What is wrfpp?

wrfpp is a Python package that facilitates the analysis of WRF and WRF-Chem outputs.

Installation

Tip

We strongly encourage you to work within some kind of isolated Python environment (eg. venv or conda)

You can install wrfpp and its dependencies with pip. It is recommended to install a tagged version (eg. "v1.0"):

pip install wrfpp[all]@git+https://github.com/WRF-Chem-Polar/wrfpp.git@v1.0

To install the latest development version, simply omit the tag:

pip install wrfpp[all]@git+https://github.com/WRF-Chem-Polar/wrfpp.git

You can choose not to install optional dependencies by omitting [all] in the commands above. Some features of wrfpp will not work in this case.

Versioning

For reproducible computing, please only use commits tagged with a version number (eg. "v2.3"). These uniquely and unequivocally define specific versions of wrfpp. This is not true for any version number that has the ".dev0" suffix (eg "v2.3.dev0").

Basic usage

First open a WRF or WRF-Chem output file with wrfpp:

import wrfpp

ds = wrfpp.open_dataset("wrfout.nc")

wrfpp adds wrf-specific functionalities to xarray datasets via a tool called an accessor.

For all practical purposes, the ds object should behave exactly like any other xarray Dataset:

print(ds.variables)

Derived variables

One of the main features of wrfpp is so-called derived variables. These are DataArray objects, just as any other dataset variable, but they are calculated on the fly by wrfpp instead of being read directly from the underlying WRF file.

For instance, WRF output files typically contain a variable T that is the potential air temperature relative to some reference value:

avg = ds.T.values.mean()
units = ds.T.units
print(f"Average relative potential temperature = {avg} {units}")
Average relative potential temperature = 43.795982360839844 K
However, air temperature is typically not found in WRF output files. Thanks to wrfpp's magic, we can use air_temperature as if it were a native variable:

avg = ds.air_temperature.values.mean()
units = ds.air_temperature.units
print(f"Average air temperature = {avg} {units}")
Average air temperature = 234.40847778320312 K

Thanks to this transparent syntax, one can write code that works agnostically on both native and derived WRF variables:

variables = ("T", "air_temperature", "QVAPOR", "relative_humidity")
for v in variables:
    data_array = getattr(ds, v)
    avg = data_array.values.mean()
    units = data_array.units
    print(f"{v}: {avg} {units}")
T: 43.795982360839844 K
air_temperature: 234.40847778320312 K
QVAPOR: 0.00043676584027707577 kg kg-1
relative_humidity: 31.046321868896484 %

The following table lists all derived variables available at this time:

Name Description
grid_cell_area Grid cell area, accounting for projection deformation
altitude_asl Grid cell altitude above sea level
altitude_agl Grid cell altitude above ground level
altitude_asl_c Grid cell centrepoint altitude above sea level
altitude_agl_c Grid cell centrepoint altitude above ground level
box_dz Grid box vertical extent
potential_temperature Potential temperature
atm_pressure Atmospheric pressure
air_temperature Air temperature
density_of_dry_air Dry air density
relative_humidity Relative humidity
accumulated_precipitation Accumulated total precipitation
liquid_water_path Liquid water path
cloud_liquid_water_path Cloud liquid water path
ice_water_path Solid water path
cloud_ice_water_path Cloud solid water path
aer_number_conc_nonact Non-activated aerosol number concentration
aer_number_conc_act Activated aerosol number concentration
aer_number_conc_total Total aerosol number concentration
fraction_activated_aerosol Number-based fraction of aerosols that are activated