era5grib

Convert NCI ERA5 archive data to GRIB format

Using era5grib

To use era5grib you will need to be a member of the hh5 (conda), rt52 (ERA5) and zz93 (ERA5-Land) projects at NCI

Load the conda environment with

module use /g/data/hh5/public/modules
module load conda/analysis3

Then run the application.

Common use cases

The most common use cases are creating initial conditions for either the Unified Model or WRF directly from ERA5 and ERA5-Land data. For example:

era5grib -f um_era5land --time 20200101T1200 --target qrparm.mask --output out.grib

will create a GRIB file named out.grib for a Unified Model LAM configuration that can be passed to the reconfiguration program to create initial conditions for a model run commencing at 1200 UTC on 1st January 2020 using ERA5-Land data supplemented by ERA5 data over ocean points; or

era5grib -f wrf_era5land --namelist namelist.wps --geo geo_em.d01.nc --output GRIBFILE.AAA

will create a GRIB file named GRIBFILE.AAA using the start time specification from namelist.wps and the domain from geo_em.d01.nc for use in generating WRF initial conditions using ERA5-Land data supplemented by ERA5 data over ocean points. The -f option specifies a YAML configuration file, and can either be the path to a file or the key for one of the four known configurations:

• wrf_era5land - Creates a GRIB1 file with all fields necessary for WPS initial conditions from regridded ERA5-Land + ERA5 data.
• wrf_era5 - Creates a GRIB1 file with all fields necessary for WPS initial conditions from only ERA5 data.
• um_era5land - Creates a GRIB1 file with all fields necessary for UM reconfiguration initial conditions from regridded ERA5-Land + ERA5 data.
• um_era5 - Creates a GRIB1 file with all fields necessary for UM reconfiguration initial conditions from only ERA5 data.

The YAML files corresponding to these known configurations can be found in the config subdirectory of the package.

Legacy run options

In order to maintain backwards compatibility with earlier versions of era5grib, legacy run modes where the 'model' keyword (wrf or um) is the second argument are supported.

era5grib wrf --namelist namelist.wps --geo geo_em.d01.nc --output GRIBFILE.AAA

or

era5grib um --time 20200101T1200 --target qrparm.mask --output era5.20200101T1200.grib

The wrf and um options are equivalent to passing -f wrf_era5land and -f um_era5land respectively. The --no-era5land flag is respected only in this variation, and will change the configuration files to -f wrf_era5 and -f um_era5 respectively. --no-era5land is ignored if the -f option is specified.

Advanced use cases.

Most of the time, simply converting netCDF ERA5/ERA5-Land data will be sufficient for constructing initial conditions for UM and WRF experiments. However, era5grib permits custom experiments without having to manually manipulate GRIB output files. All necessary configuration can be set in the configuration file. For the following examples, the configuration file will be named custom.yaml, and era5grib can be invoked as:

era5grib -f custom.yaml --time 20200101T1200 --target qrparm.mask --output out.grib

for the UM, or

era5grib -f custom.yaml --namelist namelist.wps --geo geo_em.d01.nc --output GRIBFILE.AAA

for WRF. For many cases, a small modification to a known configuration will be required. In this case, the includes key is generally placed first in a custom configuration file. The following custom.yaml is a valid configuration:

includes: wrf_era5

And is exactly equivalent to running era5grib -f wrf_era5 .... Any further options in custom.yaml will override options set by a configuration specified in includes. Running era5grib with the --debug flag or including log_level: info or log_level: debug in a custom configuration will have it print its working configuration.

Example - Replace an ERA5 field from a WRF run with a new field.

In this case, a custom skin temperature field has been generated, and we wish to replace the ERA5 skin temperature field with this new custom field. custom.yaml will be as follows:

includes: wrf_era5
custom_fields:
  skt: /path/to/custom/skt.nc

The includes key tells the configuration to inherit all settings from the wrf_era5 known configuration, and the custom field will take preference over the skt field provided by ERA5. The custom field must contain either one time stamp, the timestamp specified by the --time or --start argument, or hourly timestamps spanning the times specified in the --start and --end arguments. It must also have the same name as the field(s) it is replacing, and must have latitude and longitude dimensions (though variatons of those are acceptable e.g. LAT, LATITUDE, etc.)

In this case, the custom field must be derived from ERA5 data or match the ERA5 grib. In order to accept a custom field on any grid, so long as it contains the full geographical region specified the file passed in the --target or --geo arguments, custom.yaml must also provide a regrid key:

includes: wrf_era5
custom_fields:
  skt: /path/to/custom/skt.nc
regrid: era5

Example - Replace an ERA5 field, then replace the land values of that field with ERA5-Land data for a UM run

In this case, we're still using a custom skin temperature field, however, we only want it to persist over the ocean, and we want to use ERA5-Land data over the land. In this case, we want the custom field to take precedence over the fields from ERA5, however, we want the ERA5-Land fields to take precedence over the custom fields. To do this, we use the catalogues key which can modify the order in which the various data sources are handled.

includes: um_era5land
custom_fields:
  skt: /g/data/v45/cc6171/3JuneERA5_SKT_with_box_and_buffer.nc
catalogues:
  - era5_land
  - custom_fields
  - era5

Note the custom_fields list entry under catalogues. This is a special 'catalogue' (defined by the custom_field_catalogue_key configuration key, custom_fields by default) that instructs era5grib to process custom fields. Data from each of the data sources is handled in the order of the catalogues list, so data from ERA5-Land will be processed first, then the custom fields, then ERA5 last. Processing stops when a domain is considered complete, so for land-only fields (e.g. soil temperatures), only ERA5-Land data is considered. For global fields, however, processing will continue down the list of catalogues. Therefore, the skin temperature field will be considered complete after the custom field has been loaded. The remainder of the global or ocean-only fields will then be satisfied by the ERA5 dataset.

Example - Selecting arbitrary fields

In this case, we are extracting a series of ERA5 fields for neither the UM nor WRF, meaning that an entirely different set of fields needs to be specified. The minimum specification in this case is the fields key, which must contain single-levels and pressure-levels keys.

fields:
  single-levels:
    - 10u
    - 10v
    - sst
    - lsm
    - z
  pressure-levels:
    - u
    - v
    - z

This is necessary as the field z appears in both single level and pressure level ERA5 data with different meanings. In this configuration, both ERA5 and ERA5-Land datasets will be used, to use only ERA5, add the catalogues key to the configuration with:

fields:
  single-levels:
    - 10u
    - 10v
    - sst
    - lsm
    - z
  pressure-levels:
    - u
    - v
    - z
catalogues:
  - era5

Finally, era5grib can be used as a custom ERA5 field accumulator by setting the output format to netcdf instead of GRIB:

fields:
  single-levels:
    - 10u
    - 10v
    - sst
    - lsm
    - z
  pressure-levels:
    - u
    - v
    - z
catalogues:
  - era5
format: netcdf

Command line reference

era5grib takes the following command line options

[model]
    Legacy entry point. Must be one of um or wrf and must be the first argument to era5grib

-f, --file[=]NAME
    YAML configuration file

-o, --output[=]NAME
    Name of the output GRIB/netCDF file

--time[=]ISOTIME
    Timestamp for output file

--namelist[=]NAME
    Read start and end times from WPS namelist. Ignored if --time is specified

--start[=]ISOTIME
    Start time for multi-time output file. Ignored if --namelist or --time is specified

--end[=]ISOTIME
    End time for multi-time output. Ignored if --namelist is specified

--geo[=]NAME
    Geogrid file for trimming domain for WPS ungrib input. Ignored if [model] is um

--target[=]NAME
    UM file on target grid for trimming domain for UM reconfiguration. Ignored if [model] is wrf

--format[=]FORMAT
    Output file format. Must be one of grib or netcdf. This argument takes precedence over the format specified in the configuration file.

--debug
    Set the log level to debug. This argument takes precedence over the log level specified in the configuration files.

--[no]era5land
    Enable/disable ERA5-Land catalogue. Ignored if -f is specified.

--[no]polar
    Include all longitudes. Ignored if -f is specified.

Configuration reference

The following is a reference for the configuration parameters of era5grib. Where a default is listed, this refers to the contents of config/default.yaml. Where a default is not listed, the parameter is optional.

Output specification

fields.<dataset> List[str]:
List of fields from <dataset> to extract from each of the named intake catalogues and/or custom field files. <dataset> must correspond to an index in one or more of the specified intake catalogues. All fields must be present in their respective datasets, or the application will exit with an error. For ERA5 and ERA5-Land, valid datasets are single-levels and pressure-levels. If fields is not provided, the application will exit with an error.

custom_fields Dict[str,str]:
A mapping of fields to include external to the specified catalogues where the key is the name of the field and the value is the file in which the field will be found. Only netCDF is accepted for custom fields, and the field name within the file must correspond to the key used. The field in the file must have latitude and longitude coordinates (or equivalent) and must contain the entire domain as defined by the WRF geogrid file or UM mask file. The field must contain a single time point, or, for output containing multiple time steps, it must contain every time step expected by the output file. Used for customising initial conditions by e.g. inserting climatologies.

custom_field_flags Dict[str,str]:
Defines the realm that each custom field applies to. Valid values are global, land_only and ocean_only.

land_only List[str]:
A list of fields in all catalogues that are only defined over land. Fields in this list are assumed to be defined on a single level. Used to determine if a field read on a realm other than global is fully defined:      Default: [ stl1, stl2, stl3, stl4, swvl1, swvl2, swvl3, swvl4 ]

ocean_only List[str]:
A list of fields in all catalogues that are only defined over the ocean. Fields in this list are assumed to be defined on a single level. Used to determine if a field read on a realm other than global is fully defined:
     Default: [ ci, msl, sst ]

static.<dataset> List[str]:
A list of fields in all catalogues that are static in time. These fields are handled differently when combining a dataset with more than one time value. Defaults:

• None
• static.single-levels: [ lsm, z ]

equivalent_vars Dict[str,str]:
A mapping of fields names that are treated as equivalent.
    Default: { 10u: u10, 10v: v10 }

format str:
Final output format. Allowed values are grib for GRIB1 format and netcdf.
    Default: grib

data_types int:
Size in bytes of floating point output data types.
    Default: 32

Intake catalogues

catalogue_path List[str]:
Path to the yaml files that contain intake catalogue data. The paths will be searched in the order they are listed.
    Default: [ /g/data/hh5/public/apps/nci-intake-catalogue/catalogue_new.yaml, ]

catalogues List[str]:
Names of catalogues to search through when looking for fields. The catalogues will be searched in the order they are listed. A special name, defined by the custom_field_catalogue_key configuration value, determines the order of custom fields relative to catalogues.
    Default: [ era5land, era5 ]

catalogue_flags.<catalogue>:
Configuration applied to the catalogue <catalogue>. These are optional, defaults will be provided if any flags are not specified. A dictionary can be provided here, but the preferred method of configuring catalogue_flags is through namespaces separated by . characters. Individual options described below:

catalogue_flags.<catalogue>.realm str:
Define the domain over which the data in this catalogue is defined. Valid values are global, land_only or ocean_only.
Defaults:

• global
• catalogue_flags.era5land.realm: land_only

catalogue_flags.<catalogue>.product_type str:
The product_type to filter on.
    Default: 'reanalysis'

catalogue_flags.<catalogue>.sub_collection_pref str: The subcollection of <catalogue> to search for fields before searching through the main collection. Defaults:

• None
• catalogue_flags.era5.sub_collection_pref: era5-1

catalogue_flags.<catalogue>.chunks Dict[str,int]:
Chunk specification for data from <catalogue> loaded in xarray.
Defaults:

• catalogue_flags.era5land.chunks: {time: 1, level: 37, latitude: -1, longitude: -1}
• catalogue_flags.era5.chunks: {time: 12, level: 5, latitude: -1, longitude: -1}

land-mask.<catalogue> str:
The name of the land mask field in <catalogue> - used in certain cases for combining fields over multiple realms.
    Default: land-mask.era5: lsm

metadata_catalogue str:
Name of catalogue from which to draw GRIB parameters of the fields to be loaded.
    Default: ecmwf.grib_parameters

metadata_mapping Dict[str,str]:
A map of keys in the metadata catalogue to map to attributes of fields in the final dataset. Metadata keys not listed in this option will not be present in the final dataset.
    Default: { table: table2Version, code: indicatorOfParameter, standard_name: cfName, ecmwf_name: name, ecmwf_shortname: shortName, units: units }

custom_field_catalogue_key str:
A placeholder value to denote the order that custom fields will be processed with respect to intake catalogues:
    Default: custom_fields

Regridding and coordinates

regrid str:
If multiple grid sizes are detected in the input, this value of this key determines the target grid specification for the final output. Valid values are any of the supplied catalogue names, or the paths to any custom field files. If not present or set to None, no regridding will be attempted and the application will exit with an error if multiple grid sizes are detected.
    Default: era5

regridding Dict[str,str]:
Reference parameters for regridding. Used to load a reference field in the event that a field meeting the regridding target specification has not been already loaded.
    Default: { ref_field: 2t, ref_date: '20010101' }

regrid_options str:
Select from the two available methods for handling NaN's in land_only or ocean_only fields when being merged onto global fields. Valid values are:

• weight_file - Use a pre-determined weight file that fills ocean points with extrapolated land points. Only available when regridding from era5land to era5. Bitwise-reproducible with legacy era5grib application.
• interpolating - Use the following method of applying scipy interpolate_na function for missing values:

- If domain corners are NaN, fill with average value of field
- lat_first = field.interpolate_na(dim=latitude).interpolate_na(dim=longitude)
- lon_first = field.interpolate_na(dim=longitude).interpolate_na(dim=latitude)
- return (lat_first + lon_first / 2)

    Default: weight_file

polar bool:
Include all longitudes.
    Default: False

Application internal configuration

includes str:
Either a path to another era5grib configuration file, or the tag of one of the four provided configurations that will be included when reading the current configuration file. Values specified in a configuration in an include statement will be overwritten by those from the current file.

dataset_tags Dict[str,str]:
Internal tags used to track different types of dataset.
    Default: { single-levels: surf, pressule-levels: pl }

log_level str or int:
Application logging level as specified by the Python logging How-To guide.
    Default: warning