Ocean Color Environment on Server
Python, SeaDAS&OCSSW, SNAP
Python
At the beginning, I accidentally installed some packages in base, which can be deleted. Try to run in pysoc environment.
$ mkdir -p miniconda
$ cd miniconda
$ wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
$ bash Miniconda.... .sh
$ conda create --name pysoc python=3.7
$ conda activate pysoc
$ conda install -c conda-forge py6s
$ conda install seaborn
$ conda install -c conda-forge xarray dask netCDF4 bottleneck
$ conda install -c conda-forge cartopy
$ conda install -c conda-forge pyresample
$ conda install -c anaconda basemap
$ conda install -c conda-forge basemap-data-hires
$conda install opencv-pythonAfter installation, it’s necessary to test py6s
$ python
>>> from Py6S import *
>>> SixS.test()
6S wrapper script by Robin Wilson
Using 6S located at <PATH_TO_SIXS_EXE>
Running 6S using a set of test parameters
The results are:
Expected result: 619.158000
Actual result: 619.158000
#### Results agree, Py6S is working correctlyIf it shows this, it’s correct.
Jupyter
For convenience in writing programs, I installed jupyter
$ pip install jupyterlab
$ conda install -c conda-forge nodejs
$ conda install -c conda-forge python-lsp-server r-languageserver
$ jupyter labextension install @jupyterlab/toc
$ conda install -c conda-forge jupyterlab-drawio
$ pip install jupyterlab_latex
$ pip install lckr-jupyterlab-variableinspector
$ conda install -c conda-forge jupyterlab_execute_time
$ conda install -c conda-forge nb_conda_kernels
$ jupyter lab build
$ jupyter labextension enable allSince this is on a remote server, some configuration is needed
$ jupyter lab --port=8888 --no-browser --allow-rootOn your own computer
$ ssh -N -f -L 8880:localhost:8888 root@123435Finally, just add the URL shown on the server in your browser.
Remember to ask the programmer to open the firewall.
SeaDAS
Basic Environment
$ sudo yum -y install gcc
$ sudo yum -y install git
$ sudo yum -y install curl
$ sudo yum -y install wgetSince it’s RedHat, there’s no apt-get.
$ sudo yum -y-repository ppa:webupd8team/java
$ sudo yum -y update
$ sudo yum -y install oracle-java8-installer
$ sudo yum -y install oracle-java8-set-default
$ java -versionShows
java version "1.8.0_311"
Java(TM) SE Runtime Environment (build 1.8.0_311-b11)
Java HotSpot(TM) 64-Bit Server VM (build 25.311-b11, mixed mode)$ wget https://oceandata.sci.gsfc.nasa.gov/SeaDAS/installer/7.3.1/seadas_7.3_linux64_installer.sh
$ bash seadas...sh
This is the last version that doesn’t require GUI installation, no wonder so many papers keep using 7.3.
Where should SeaDAS be installed? [/opt/seadas-7.3]
A strange thing
This is the first time I discovered that Linux paths work this way
cd / enters the root directory
This directory is shared by all users
While cd ~ enters my own directory
Learned something new
OCSSW
The most important part
$ wget https://oceandata.sci.gsfc.nasa.gov/ocssw/install_ocssw.py
$ chmod +x install_ocssw.py
$ ./install_ocssw.py --install-dir=/root/SeaDAS/seadas-7.3/ocssw --aqua --seawifs --goci
$ mv /opt/seadas-7.3 ~/SeaDAS
$ vim ~ .bashrcEnter
export SEADASPATH=/root/SeaDAS/seadas-7.3
export PATH=$PATH:$SEADASPATH/binwhere: [SeaDAS_install_dir] is the directory where you installed SeaDAS.
Save after finishing.
At this point, I suddenly found that I couldn’t find those installed files in the OCSSW path. After searching, it turned out they were installed to / instead of ~. The download location above can change the path but I forgot to change it.
Had to move them over again.
Test command
$ source .../OCSSW_bash.env
$ which l2genShows
/root/SeaDAS/seadas-7.3/ocssw/bin/l2genThat means it’s done.
Then I added a more complete configuration in bashrc
export SEADASPATH=~/SeaDAS/seadas-7.3
export PATH=$PATH:$SEADASPATH/bin
export PYTHON_PROGRAMS=/root/miniconda3/bin/python
export PYTHONPATH=$PYTHONPATH:$PYTHON_PROGRAMS:$PYTHON_PROGRAMS/utilities
export LOCAL_RESOURCES=$PYTHON_PROGRAMS/local_procesing_resources
export OCSSWROOT=$SEADASPATH/ocssw
export OCSSW_DEBUG=0 # set to 1 for debugging
source $OCSSWROOT/OCSSW_bash.envFile Transfer
Previously I used Putty or FileZilla directly, but the company programmer said to use SCP
$ scp SourceFile user@server:directory/TargetFileTo transfer back
$ scp user@server:directory/SourceFile TargetFileAtmospheric Correction
Due to NASA data format updates, it’s slightly different from the previous script
#Example script to process L1A files up to L2, put in same directory as
# your L1A files
# Run script by typing on Terminal Command Line:
# bash l1a_to_l2.sh input.txt
# input.txt contains list of files [ create by: ls -1 *L1A_LAC* > input.txt ]
LIST=$1
if [ -z "$LIST" ]
then
echo "No input file list supplied"
exit 1
fi
if [[ ! -a $LIST ]]; then
echo "$LIST does not exist!"
exit 1
fi
mkdir -p done
mkdir -p l2_lac
for FILE in $(cat $LIST);
do
echo "Working on $FILE"
echo ${FILE}
# get file basename (no file extension)
BASE=`basename $FILE .L1A_LAC`
echo ${BASE}
GEOFILE=${BASE}.GEO
echo ${GEOFILE}
L1BFILE=${BASE}.L1B_LAC
echo ${L1BFILE}
L2FILE=${BASE}.L2_LAC.x.hdf
echo ${L2FILE}
ancfile=${BASE}.L1A_LAC.x.hdf.anc
echo "Creating GEOFILE $GEOFILE"
modis_GEO.py -v $FILE -o $GEOFILE
echo "Creating L1B file"
modis_L1B.py -v $FILE $GEOFILE
echo "Creating anc file"
getanc.py -v $FILE
echo "Create L2 file"
l2gen ifile=$L1BFILE geofile=$GEOFILE ofile=$L2FILE par=$ancfile l2prod="Kd_490 Rrs_nnn angstrom aot_869 chlor_a ipar nLw_nnn nflh par pic poc rhos_nnn" aer_opt=-2
echo "Cleaning up"
rm -v ${BASE}.L1B* $GEOFILE $ancfile
#mv -v $FILE done/
mv -v $L2FILE l2_lac/
doneThings Went Wrong
After running a test data, I found it couldn’t run.
Had to reinstall.
And this network was unstable.

Classic 404
After checking, I found their official website was under maintenance.
Things Went Right Again
$ bzip2 -d *.bz2
$ ls -1 *L1A_LAC* > input.txt
$ bash l1_to_l2.sh inupt.txt