Notebooks for melt onset research.
- Download and install Miniconda3 from
http://conda.pydata.org/miniconda.html.
cd ~/Downloads
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh bash Miniconda3-latest-Linux-x86_64.sh
Miniconda is a program that will allow you control what are called conda "environments," which just means that you can set up individualized python installations with difference sets of packages. It does all of the package management for you, so it will know what combination of packages to install for cetbtools, and jupyter and all the tool you will want to use. Most importantly, it keeps this installation completely separate from any other tools you may have installed that use their own special version of python. So you will be playing "nice" and not stepping on your other tools.
The latest Miniconda version currently is for Python 3.9, this is fine.
You will download a file called something like:
Miniconda3-py39_4.9.2-Windows-x86_64.exe
Go ahead and run it to install Miniconda.
Accept all defaults during the installation, except for one:
Check to make sure you have a latest conda installed with
conda update -n base -c defaults conda
As of 8/8/2022, the latest version of conda is 4.13.0.
2.a. Windows: Get the miniconda shell window by doing:
Windows menu > All Programs > Anaconda Prompt (miniconda3).
2.b. non-Windows: Open a terminal on your machine.
- Set up the list of "channels," which are locations to look online for the packages we will need, and the priority order that these locations will be searched. (If you had a previous .condarc, it might need to be deleted or appended.) The first conda config command that is run will always include defaults channel.
Add necessary channels (order is important here):
conda config --add channels conda-forge
conda config --add channels nsidc
You should see the following when checking your channels:
conda config --show channels
channels:
- nsidc
- conda-forge
- defaults
- Create an environment with the cetbtools and jupyter
First, confirm that you only have the base environment:
conda info --envs
should return:
# conda environments:
base * C:<your home directory>\miniconda3
Now install a package called mamba in the base environment that will speed up a lot of the rest of the process:
conda install mamba
Now use mamba to create a new environment that we will use for our project:
mamba create -n cetb cetbtools matplotlib scipy jupyter basemap seaborn
(Starting summer 2024, pandas is complaining about future need to install pyarrow for pandas 3.x and later. For now, add pyarrow to the conda env manually. I think once pandas is pinned to 3.x, the pyarrow will come along with it.)
Alternatively, some extra packages that are needed for geotiff clipping operations:
mamba create -n cetb cetbtools matplotlib scipy jupyter basemap seaborn geopandas rasterio pycrs
This should get cetbtools 1.7.x--this is the latest as of August 2022.
Optional, only if it's needed to get a very specific version of cetbtools:
Windows:
mamba create -n cetb cetbtools matplotlib scipy jupyter basemap proj4 seaborn
Non-Windows:
mamba create -n cetb cetbtools matplotlib scipy jupyter basemap seaborn
as of August 2022, apparently mamba won't do the magic on OSX, use conda if working on a Mac.
N.B.:
- jupyter is needed for running jupyter notebooks
- seaborn is a graphics package that does some things much much better than matplotlib
- basemap is needed for some of the displays and maps that we are displaying in the notebooks.
N.B.: confirm that matplotlib and scipy are both coming from conda-forge
If you run "conda env list" again will will see a second entry, for cetb. The asterisk will be positioned in the environment that is "current."
- Switch to the new conda env that we just created:
conda activate cetb
Note that your command prompt has changed from (base) to (cetb).
This took a little while to run on my machine, be patient. Mine generated some non-fatal warning messages prepended with "SafetyError:..." I think it's OK to ignore them, I'm checking on that.
- Test the installation. At the command prompt:
python -c "from cetbtools import algorithms; help(algorithms)"
should return an array of with methods in the algorithms module.
- Start jupyter notebook. The directory where you start the jupyter notebook server will be the top directory in the jupyter notebooks, so I recommend cd'ing to it if it's a long way from your home directory.
Optional:
cd
At the command prompt, start the jupyter notebook server:
jupyter notebook --ip=0.0.0.0 --no-browser
Your browser should start up with jupyter server running in new tab.
- Open a blank notebook, and try doing:
import cetbtools
this should do the import without generating an error message.
help(cetbtools)
this should display the package help message
from cetbtools import ease2conv help(ease2conv)
this should display the ease2conv help message
Congratulations, you have successfully installed the cetbtools python package on your local machine.
To list all the conda envs you have created:
conda info --envs
To remove a conda env:
conda remove --name <my_env> --all
To use the ipynb debugger for understanding what the scripts are doing:
Add the following line to one of the script files in the place you want to set a breakpoint:
import pdb; pdb.set_trace()Save the python script.
Then (this is important) go back to the notebook, restart the kernel and clear all outputs and then recompile all cells up to the point where you are calling the function in the python script. This will ensure that you are now importing the script with the set_trace call in it.
When you execute this cell, the python debugger will start an ipython debugger ("ipdb") subshell in the jupyter notebook. To do things in this cell, you can use the usual
l : list the code here l : list the code centered on this the usual jupter commands to print variables, etc
Remember that this is a subshell to the jupyter notebook, so to execute commands here you use (NOT shift-).
To exit the ipdb subshell, use
exit()