Method to calculate a gridded, prior emissions estimate for methane across Australia.
This repository is matched with downloadable input data so that it will run out of the box.
Copy the .env.example
file to .env
and customise the paths as you need.
In order to download the GFAS emissions data, credentials for the Copernicus Atmospheric Data Store (ADS) API are required. Instructions for registering for the ADS API and setting up the credentials are provided at ADS Docs.
Step-by-step:
- Register for an ECMWF account
- While logged in to ECMWF, register your account with ADS
- Accept the ADS terms and conditions
- Accept the License to use Copernicus products, by visiting the Download tab of the dataset you wish to use and scrolling to the Terms of use section: https://ads.atmosphere.copernicus.eu/datasets/cams-global-fire-emissions-gfas?tab=download
Note: the ADS API is different from the CDS (Climate Data Store) API even though they are both parts of the Copernicus program and share the same credentials file.
To get started, you will need to make sure that poetry is installed. The Open Methane prior can be installed from source into a virtual environment with:
make virtual-environment
The Makefile
contains the set of commands to create the virtual environment.
You can read the instructions out and run the commands by hand if you wish.
To download all the required input files, run:
make download
This will download input files that match the data in .env
,
so you have a working set to get started with.
The downloaded files will be stored in data/inputs
by default.
The domain of interest for the prior is defined using an input domain netCDF file.
The format of the input domain is based on the CMAQ domain file format. Note that CMAQ uses a
staggered grid
where some quantities are defined at the center of a grid cell, whereas other quantities are defined
at the edges of a grid cell. This circumstance is represented in ROW_D = ROW + 1
.
This input file should contain the following variables:
LAT
LON
LANDMASK
LATD
LOND
The contents of the default domain is shown below:
>>> ncdump -h prior_domain_aust10km_v1.0.0.d01
netcdf prior_domain_aust10km_v1.0.0.d01 {
dimensions:
TSTEP = 1;
ROW = 430;
COL = 454;
LAY = 1;
ROW_D = 431;
COL_D = 455;
variables:
float LAT(TSTEP, ROW, COL);
LAT:_FillValue = NaNf;
LAT:long_name = "LAT";
LAT:units = "DEGREES";
LAT:var_desc = "latitude (south negative)";
float LON(TSTEP, ROW, COL);
LON:_FillValue = NaNf;
LON:long_name = "LON";
LON:units = "DEGREES";
LON:var_desc = "longitude (west negative)";
float LANDMASK(TSTEP, ROW, COL);
LANDMASK:_FillValue = NaNf;
LANDMASK:long_name = "LWMASK";
LANDMASK:units = "CATEGORY";
LANDMASK:var_desc = "land-water mask (1=land, 0=water)";
float LATD(TSTEP, LAY, ROW_D, COL_D);
LATD:_FillValue = NaNf;
LATD:long_name = "LATD";
LATD:units = "DEGREES";
LATD:var_desc = "latitude (south negative) -- dot point";
float LOND(TSTEP, LAY, ROW_D, COL_D);
LOND:_FillValue = NaNf;
LOND:long_name = "LOND";
LOND:units = "DEGREES";
LOND:var_desc = "longitude (west negative) -- dot point";
// global attributes:
:DX = 10000.f;
:DY = 10000.f;
:TRUELAT1 = -15.f;
:TRUELAT2 = -40.f;
:MOAD_CEN_LAT = -27.644f;
:STAND_LON = 133.302f;
:XCELL = 10000.;
:YCELL = 10000.;
:XCENT = 133.302001953125;
:YCENT = -27.5;
:XORIG = -2270000.;
:YORIG = -2165629.25;
}
As part of the OpenMethane project, we have provided a domain file for a 10km grid over Australia.
This file will be downloaded with the other layer inputs (see Input Data) using the default configuration values.
These two commands are set up so that not all generated files have to be deleted manually
Delete all files in the intermediates
and outputs
directory with
make clean
Or delete all files in intermediates
, outputs
, and inputs
directory with
make clean-all
To calculate emissions for all layers, run omPrior.py
with a start and end date:
poetry run python scripts/omPrior.py 2022-07-01 2022-07-01
or use the make target
make run
This takes a while to process (~10 minutes) with the vast majority of that time spent on the layers
in omAgLulucfWasteEmis.py
.
To skip re-projecting raster layers (you only need to do this once for every time you change the raster input files),
add the --skip-reproject
option.
You can run and re-run individual layers one-by-one. Just run each file on it's own (GFAS and Wetlands require a start and end date as below):
poetry run python src/layers/omWetlandEmis.py 2022-07-01 2022-07-02
Outputs can be found in the data/outputs
folder. The emissions layers will be written as variables to a copy of the
input domain file, with an OCH4_
prefix for the methane layer variable names. The sum of all layers will be stored in
the OCH4_TOTAL
layer.
The name of the layered output file will be om-prior-output.nc
.
The data/processed
folder will contain any re-projected raster data, and any files downloaded or generated in the
process.
Many sectors are taken from data sets used in by Saunois et al (2020) (doi:10.5194/essd-12-1561-2020)
- Livestock: Enteric fermentation emissions generated by CSIRO Ag. and Food using livestock census data and UNFCCC emissions factors
- Electricity: Uses OpenNEM facility data to spatialise the Aust. Gov UNFCCC electricity emissions
- Agriculture: Agricultural emissions apart from livestock taken from the Agricultural emissions of the NGGI and spatialised according to the agriculture land-use mask
- Fugitives: Facility-level data from ACF (more info?)
- Industrial: Spatialises the industrial sector of the NGGI according to nighttime lights
- Stationary: Spatialises the stationary energy sector of the NGGI according to nighttime lights
- Transport: Spatialises the transport sector of the NGGI according to nighttime lights
- Waste: Spatialises the NGGI waste emission according to the landuse map
- LULUCF: Spatialises the LULUCF emission from the NGGI according to the landuse map
- FIRE: daily emissions from the Global Fire Assimilation System (Kaiser et al., 2012, doi:10.5194/bg-9-527-2012)
- wetland: Monthly wetland emissions from the diagnostic ensemble used in Saunois et al. 2020 and described in Zhang et al. (2023 under review)
- Termite: Termite emissions used in Saunois et al. 2020 supplied by Simona Castaldi and Sergio Noce
data/inputs
This folder should contain all the required input files, which should be referenced in the.env
file at the root. A set of input files has been included in the repository so that it functions out of the box (see Input Data), but you can add your own data here.data/inputs/domains
The domain of interest is stored in this folder (see domain info).data/intermediates
This folder contains any intermediate files generated through the process. Everything within this folder should be ignored.data/outputs
Outputs files will be saved here.
To carry out the steps described above in a Docker container, first build the Docker image with
make build
Then run the commands to download the input data in the docker container
docker run --rm -v </your/path/to/openmethane-prior>:/opt/project openmethane-prior python scripts/omDownloadInputs.py
Replace the python files according to the commands in the Makefile for the other steps.
The ruff-fixes target runs a series of ruff commands to format the code, check and fix linting issues, and then format the code again to ensure that all formatting and fixes are applied.
make ruff-fixes
The test target will run all the tests
make test