This package aims to provide machine learning (ML) functions for performing comprehensive soil and groundwater data analysis, and for supporting the establishment of effective long-term monitoring. The package includes unsupervised ML for identifying the spatiotemporal patterns of contaminant concentrations (e.g., PCA, clustering), and supervised ML for evaluating the ability of estimating contaminant concentrations based on in situ measurable parameters, as well as the effectiveness of well configuration to capture contaminant concentration distributions. Currently, the main focus is to analyze historical groundwater datasets and to extract key information such as plume behaviors and controlling (or proxy) variables for contaminant concentrations (Schmidt et al., 2018). This is setting a ground for integrating new technologies such as in situ sensors, geophysics and remote sensing data.
This development is a part of the Advanced Long-Term Monitoring Systems (ALTEMIS) project. In this project, we propose to establish the new paradigm of long-term monitoring based on state-of-art technologies – in situ groundwater sensors, geophysics, drone/satellite-based remote sensing, reactive transport modeling, and AI – that will improve effectiveness and robustness, while reducing the overall cost. In particular, we focus on (1) spatially integrative technologies for monitoring system vulnerabilities – surface cap systems and groundwater/surface water interfaces, and (2) in situ monitoring technologies for monitoring master variables that control or are associated with contaminant plume mobility and direction. This system transforms the monitoring paradigm from reactive monitoring – respond after plume anomalies are detected – to proactive monitoring – detect the changes associated with the plume mobility before concentration anomalies occur.
The latest package can be downloaded from: https://pypi.org/project/pylenm/
More information on the project can be found here: https://altemis.lbl.gov/ai-for-soil-and-groundwater-contamination/
It is recommended to install the package and work in a virtual environment.
Read more here to learn how to install conda.
conda create --name pylenm_env python=3.8
conda activate pylenm_env
Working with Anaconda, you might need to install jupyter for Anaconda to identify this env as a jupyter environemnt.
pip install jupyter
Install directly using pip as mentioned on the PyPI page.
pip install pylenm
-
Clone the repository
git clone https://github.com/ALTEMIS-DOE/pylenm.git cd pylenm
-
Install the package
pip install .
Aurelien O. Meray, Savannah Sturla, Masudur R. Siddiquee, Rebecca Serata, Sebastian Uhlemann, Hansell Gonzalez-Raymat, Miles Denham, Himanshu Upadhyay, Leonel E. Lagos, Carol Eddy-Dilek, and Haruko M. Wainwright Environmental Science & Technology 2022 56 (9), 5973-5983 DOI: 10.1021/acs.est.1c07440
These notebooks use the refactored version of the pylenm package - pylenm2.
This refactored version reorganizes the functions into a more semantically separated modules.
To use this version, import pylenm2 instead of pylenm after installation.
The function hirarchy is shown in pylenm2 README.
1 - Basics
2 - Unsupervised learning
3 – Water Table Estimation & Well Optimization
4 – Tritium Spatial Estimation
5 – Proxy Estimation (SC~Tritium)
6 - LOWESS Outlier removal
7 - Miscellaneous
Sample data used for these notebooks is stored in the data directory.
1 – Basics
2 - Unsupervised learning
3 – Water Table Estimation & Well Optimization
4 – Tritium Spatial Estimation
5 – Proxy Estimation (SC~Tritium)
The data used in the demonstration notebooks above can be downloaded here.
Aurelien Meray
Haruko Wainwright
Himanshu Upadhyay
Masudur Siddiquee
Savannah Sturla
Nivedita Patel
Kay Whiteaker
Haokai Zhao
Haokai Zhao
Satyarth Praveen
Zexuan Xu
Aurelien Meray
Haruko Wainwright