Requires python3.7 or later.
Create a virtual environment and activate it.
pip install virtualenv
python3 -m virtualenv venv
source venv/bin/activate
Install dependencies listed in requirements.txt.
pip install -r requirements.txt
We observed that the weather data values might be missing for some dates and locations, for which WOFOST throws an error. Therefore we updated the pcse source code to interpolate the missing weather values by using the values from the previous days.
To use this imputation method for weather, you can simply change venv/lib/python3.7/site-packages/pcse/base/weather.py with the weather.py in this repo.
To view jupyter notebooks running on a remote machine on your local browser, you can use port forwarding.
For convenience, first create aliases for these commands on your ~/.bashrc.
On the remote machine:
alias jp="jupyter notebook --no-browser"
On your local machine:
alias port_forward="ssh -N -f -L localhost:XXXX:localhost:YYYY user@remote"
alias show_ports="netstat -vanp --tcp"
Source your .bashrcs on both machines. Now can view the ports in use by the command show_ports.
Choose any available XXXX port on local machine and YYYY port on the remote machine.
After aliasing, run the following commands to start the jupyter notebook at port=YYYY on the remote machine, and to forwards ports such that you can view the notebook on your browser at localhost:XXXX
On the remote machine:
jp --port=XXXX
On the local machine:
port_forward
You will need to repeat the above two lines every time you disconnect from the remote machine. To avoid killing the notebook when ssh connection is broken, you can use tmux.