source:
flex_extract.git/For_developers/Sphinx/source/Installation/local.rst
@
b1674ed
Last change on this file since b1674ed was b1674ed, checked in by Anne Philipp <anne.philipp@…>, 4 years ago | |
---|---|
|
|
File size: 15.6 KB |
Local mode installation
.. toctree:: :hidden: :maxdepth: 2
Local mode - dependencies
The installation is the same for the access modes member and public.
The environment on your local system has to provide these software packages and libraries, since the preparation of the extraction and the post-processing is done on the local machine:
Python part | Fortran part |
|
Prepare local environment
The easiest way to install all required packages is to use the package management system of your Linux distribution which requires admin rights. The installation was tested on a Debian GNU/Linux buster and an Ubuntu 18.04 Bionic Beaver system.
# On a Debian or Debian-derived sytem (e. g. Ubuntu) system you may use the following commands (or equivalent commands of your preferred package manager): # (if not already available): apt-get install python3 (usually already available on GNU/Linux systems) apt-get install python3-eccodes apt-get install python3-genshi apt-get install python3-numpy apt-get install gfortran apt-get install fftw3-dev apt-get install libeccodes-dev apt-get install libemos-dev # Some of these packages will pull in further packages as dependencies. This is fine, and some are even needed by ``flex_extract''. # As currently the CDS and ECMWF API packages are not available as Debian packages, they need to be installed outside of the Debian (Ubuntu etc.) package management system. The recommended way is: apt-get install pip pip install cdsapi pip install ecmwf-api-client
Note
In case you would like to use Anaconda Python we recommend you follow the installation instructions of Anaconda Python Installation for Linux and then install the eccodes package from conda with:
conda install conda-forge::python-eccodes
The CDS API (cdsapi) is required for ERA5 data and the ECMWF Web API (ecmwf-api-client) for all other public datasets.
Note
Since public users currently don't have access to the full ERA5 dataset they can skip the installation of the CDS API.
Both user groups have to provide keys with their credentials for the Web API's in their home directory. Therefore, follow these instructions:
- ECMWF Web API:
- Go to MARS access website and log in with your credentials. Afterwards, on this site in section "Install ECMWF KEY" the key for the ECMWF Web API should be listed. Please follow the instructions in this section under 1 (save the key in a file .ecmwfapirc in your home directory).
- CDS API:
- Go to CDS API registration and register there too. Log in at the cdsapi website and follow the instructions at section "Install the CDS API key" to save your credentials in a .cdsapirc file.
Test local environment
Check the availability of the python packages by typing python3 in a terminal window and run the import commands in the python shell. If there are no error messages, you succeeded in setting up the environment.
# check in python3 console import eccodes import genshi import numpy import cdsapi import ecmwfapi
Test the Web API's
You can start very simple test retrievals for both Web APIs to be sure that everything works. This is recommended to minimise the range of possible errors using flex_extract later on.
ECMWF Web API
Please use this piece of Python code for Member user: | Please use this piece of Python code for Public user: |
from ecmwfapi import ECMWFService server = ECMWFService('mars') server.retrieve({ 'stream' : "oper", 'levtype' : "sfc", 'param' : "165.128/166.128/167.128", 'dataset' : "interim", 'step' : "0", 'grid' : "0.75/0.75", 'time' : "00/06/12/18", 'date' : "2014-07-01/to/2014-07-31", 'type' : "an", 'class' : "ei", 'target' : "download_erainterim_ecmwfapi.grib" }) |
from ecmwfapi import ECMWFDataServer server = ECMWFDataServer() server.retrieve({ 'stream' : "enda", 'levtype' : "sfc", 'param' : "165.128/166.128/167.128", 'dataset' : "cera20c", 'step' : "0", 'grid' : "1./1.", 'time' : "00/06/12/18", 'date' : "2000-07-01/to/2000-07-31", 'type' : "an", 'class' : "ep", 'target' : "download_cera20c_ecmwfapi.grib" }) |
CDS API
Extraction of ERA5 data via CDS API might take time as currently there is a high demand for ERA5 data. Therefore, as a simple test for the API just retrieve pressure-level data (even if that is NOT what we need for FLEXPART), as they are stored on disk and don't need to be retrieved from MARS (which is the time-consuming action):
Please use this piece of Python code to retrieve a small sample of ERA5 pressure levels:
import cdsapi c = cdsapi.Client() c.retrieve("reanalysis-era5-pressure-levels", { "variable": "temperature", "pressure_level": "1000", "product_type": "reanalysis", "year": "2008", "month": "01", "day": "01", "time": "12:00", "format": "grib" }, "download_cdsapi.grib")
If you know that your CDS API works, you can try to extract some data from MARS.
Please use this piece of Python code to retrieve a small ERA5 data sample as a member-state user! The Public user do not have access to the full ERA5 dataset!
import cdsapi c = cdsapi.Client() c.retrieve('reanalysis-era5-complete', { 'class' : 'ea', 'expver' : '1', 'stream' : 'oper', 'type' : 'fc', 'step' : '3/to/12/by/3', 'param' : '130.128', 'levtype' : 'ml', 'levelist': '135/to/137', 'date' : '2013-01-01', 'time' : '06/18', 'area' : '50/-5/40/5', 'grid' : '1.0/1.0', 'format' : 'grib', }, 'download_era5_cdsapi.grib')
Local installation
First prepare the Fortran makefile for your environment and set it in the setup.sh script. (See section :ref:`Fortran Makefile <ref-convert>` for more information.) flex_extract comes with two makefiles prepared for the gfortran compiler. One for the normal use makefile_fast and one for debugging makefile_debug which is usually only resonable for developers.
They assume that eccodes and emoslib are installed as distribution packages and can be found at flex_extract_vX.X/Source/Fortran, where vX.X should be substituted with the current version number.
Caution!
It is necessary to adapt ECCODES_INCLUDE_DIR and ECCODES_LIB in these makefiles if other than standard paths are used.
So starting from the root directory of flex_extract, go to the Fortran source directory and open the makefile of your choice to modify with an editor of your choice. We use the nedit in this case.
cd flex_extract_vX.X/Source/Fortran
nedit makefile_fast
Edit the paths to the eccodes library on your local machine.
Caution!
This can vary from system to system. It is suggested to use a command like
# for the ECCODES_INCLUDE_DIR path do: $ dpkg -L libeccodes-dev | grep eccodes.mod # for the ECCODES_LIB path do: $ dpkg -L libeccodes-dev | grep libeccodes.so
to find out the path to the eccodes library.
Substitute these paths in the makefile for parameters ECCODES_INCLUDE_DIR and ECCODES_LIB and save it.
# these are the paths on a current Debian 10 Testing system (May 2019) ECCODES_INCLUDE_DIR=/usr/lib/x86_64-linux-gnu/fortran/gfortran-mod-15/ ECCODES_LIB= -L/usr/lib -leccodes_f90 -leccodes -lm
The Fortran program called calc_etadot will be compiled during the installation process.Therefore the name of the makefile to be used needs to be given in setup.sh.
In the root directory of flex_extract, open the setup.sh script and adapt the installation parameters in the section labelled with "AVAILABLE COMMANDLINE ARGUMENTS TO SET" like shown below.
.. code-block:: bash :caption: 'Example settings for a local installation.' :name: setup.sh ... # ----------------------------------------------------------------- # AVAILABLE COMMANDLINE ARGUMENTS TO SET # # THE USER HAS TO SPECIFY THESE PARAMETER # TARGET='local' MAKEFILE='makefile_fast' ECUID=None ECGID=None GATEWAY=None DESTINATION=None INSTALLDIR=None JOB_TEMPLATE='job.template' CONTROLFILE='CONTROL_EA5' ...
Afterwards, type:
$ ./setup.sh
to start the installation. You should see the following standard output.
# Output of setup.sh WARNING: installdir has not been specified flex_extract will be installed in here by compiling the Fortran source in <path-to-flex_extract>/flex_extract_v7.1/Source/Fortran Install flex_extract_v7.1 software at local in directory <path-to-flex_extract>/flex_extract_v7.1 Using makefile: makefile_fast gfortran -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -I. -I/usr/local/include -fdefault-real-8 -fopenmp -fconvert=big-endian -c ./rwgrib2.f90 gfortran -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -I. -I/usr/local/include -fdefault-real-8 -fopenmp -fconvert=big-endian -c ./phgrreal.f90 gfortran -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -I. -I/usr/local/include -fdefault-real-8 -fopenmp -fconvert=big-endian -c ./grphreal.f90 gfortran -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -I. -I/usr/local/include -fdefault-real-8 -fopenmp -fconvert=big-endian -c ./ftrafo.f90 gfortran -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -I. -I/usr/local/include -fdefault-real-8 -fopenmp -fconvert=big-endian -c ./calc_etadot.f90 gfortran -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -I. -I/usr/local/include -fdefault-real-8 -fopenmp -fconvert=big-endian -c ./posnam.f90 gfortran rwgrib2.o calc_etadot.o ftrafo.o grphreal.o posnam.o phgrreal.o -o calc_etadot_fast.out -O3 -march=native -Bstatic -leccodes_f90 -leccodes -Bdynamic -lm -ljasper -lemosR64 -fopenmp ln -sf calc_etadot_fast.out calc_etadot lrwxrwxrwx. 1 <username> tmc 20 15. Mär 13:31 ./calc_etadot -> calc_etadot_fast.out