The data was downloaded, preprocessed for the task and stored in .tif files. It is available at Google Drive as archive. Prior to the running a docker container (see next section), the project directory should be organized as follows (tree depth is limited by 4):
.
├── configs
│ ├── eval
│ | └── eval_config.yaml
│ ├── process
│ | └── process_config.yaml
│ ├── train
│ | └── train_config.yaml
│ └── main_config.yaml
├── data
│ ├── 01-raw
│ | ├── boundary
│ | ├── Asia
│ | | ├── Climate
│ | | ├── LC_Type2.tif
│ | | └── ELv.tif
│ | ├── Rice_map
│ | ├── SEAsia
│ | | ├── Climate_future
│ | | ├── LC_Type2.tif
│ | | └── ELv.tif
│ | └── FAOSTAT
│ ├── 01-prepared
│ ├── 02-models
│ ├── 03-results
│ └── 04-feed
├── README.md
├── environments
│ ├── Dockerfile
│ ├── poetry.lock
│ └── pyproject.toml
├── notebooks
│ ├── crops_statistics.ipynb
│ ├── notebook_yield.ipynb
│ ├── notebook_esg.ipynb
│ └── rice_mask.ipynb
└── src
├── load_data.py
├── models
├── plotting.py
├── rice_mask.py
└── utils.py
Pay attention the available years range in that source:
From repo folder run:
docker build -t arable:1.0 environments/.
docker run -it -v <CODE FOLDER>:/app -v <DATA FOLDER>:/app/data -w="/app" arable:1.0
sh download.sh
-- it will download all the necessary data from google drive, unpack it, delete the archived dataScripts in repo folder allow to perform main steps of algorithm. To run them properly, set the required parameters in all files (including subfolders) of 'config folder'.
Alternatively, the same steps are implemented interactevely in notebook_esg.ipynb
Moreover, notebook_yield.ipynb performs modeling of fertilizers consumption and rice yield in countries of Southeastern Asia. More respecifically, it
'src' folder contains load_data.py
, plotting.py
and utils.py
modules for loading data, plotting and auxilary procedures.