💊 SourceApp 🚱

This repository contains material accompanying the paper by Chrapkiewicz et al. (2024) Apportioning sources of chemicals of emerging concern along an urban river with inverse modelling, published in Science of the Total Environment.

At the centre of our workflow is the faster-unmixer software by Barnes & Lipp (2024). It can be installed following the guidelines below.

Structure of the repository

• data: contains publicly available data sufficient to reproduce the results,
• examples: contains a minimal working example, MWE (see 'Installation' below for how to run it).

The example uses two input files (provided under data):

• flow directions across Thames basin (in ESRI D8 format)
• an Excel file with concentrations of contaminants of emerging concern measured across Thames basin in London by Egli et al. (2023), see: https://doi.org/10.1016/j.envint.2023.108210.

Installation

To install the software required to run the MWE, follow the steps below. The commands are supposed to be run in a Unix-like terminal (tested in bash and zsh shells), unless otherwise specified. Make sure you have anaconda installed before you proceed.

1. Clone the repository and navigate to its directory:

git clone git@github.com:kmch/SourceApp.git
cd SourceApp

2. Create a conda environment using the provided yaml file:

my_env=mwe; conda env create --name $my_env --file=mwe.yaml && conda activate $my_env

From now on, all the commands need to be run in this conda environment. Make sure that the python command points to the executable of this environment, not the system-wide Python installation. Check this by running:

which python

It should output a path looking more or less like this:

$HOME/anaconda3/envs/mwe/bin

where $HOME will be your home path.

3. Install the autocatchments package, following the guidelines on https://github.com/AlexLipp/autocatchments/tree/name listed below:

git clone git@github.com:AlexLipp/autocatchments.git
cd autocatchments
pip install -e .

4. Install the faster-unmixer package:

git clone --recurse-submodules https://github.com/r-barnes/faster-unmixer/ 
cd faster-unmixer
git submodule update --init --recursive
pip install -e .

5. Run the MWE:

cd SourceApp/examples
python mwe.py

Open the generated PNG files. The top-right subplot of each 6-subplot figure is supposed to be blank.

Troubleshooting

If you experience troubles with Python failing to import some modules, check if they are installed correctly by running:

conda activate $my_env && conda list | grep riverpy

where riverpy should be replaced by the module you want to check. If the above command returns an empty line, this means the module is not installed in your environment. To solve this problem, make sure that the python command points to the executable of this environment (see the step 2. of this section).

If you encounter any run-time errors from the calls to funmixer, it may mean the version you cloned has some modifications that made it incompatible with the rest of the software. In this case, change to the tested version of the package by running:

cd faster-unmixer
git checkout 939c9f89f27ccefc4a172eb065c0d6f1f0de0438

Citing

If you use any part of this repository in your research, please cite the following paper:

• Chrapkiewicz, Kajetan and Lipp, Alex and Barron, Leon Patrick and Barnes, Richard and Roberts, Gareth, Apportioning sources of chemicals of emerging concern along an urban river with inverse modelling. Science of the Total Environment. DOI:10.1016/j.scitotenv.2024.172827.