This repository contains the contents to run the case study described in the manuscript "A computational framework for agent-based assessment of multiple environmental exposures". Users on Linux operating systems should be able to execute the model script.

How to install

A few steps are required to run the case study.

1. You will need a working Python environment, we recommend to install Miniforge in case you have no Conda package manager installed yet. Follow their instructions given at:

   https://conda-forge.org/download/

   before you continue.

2. Open a terminal and browse to a location where you want to store the course contents.

3. Clone this repository

   git clone https://github.com/computationalgeography/paper_agent_based_exposure_assessment.git

   or download and uncompress the zip file of the repository.

4. Navigate to the paper_agent_based_exposure_assessment folder:

   cd paper_agent_based_exposure_assessment

   and create the required Python environment:

   conda env create -f environment/environment.yaml

The environment file will create a environment named casestudyutrecht using Python 3.10. In case you prefer a different name you need to edit the environment/environment.yaml file.

The user guide and short reference on Conda can be found here.

Download the input data

Required additional input data can be downloaded from Zenodo, e.g. with

wget https://zenodo.org/records/13913079/files/input_data.zip

Extract the downloaded zip file and its contents into the paper_agent_based_exposure_assessment folder.

How to run

Activate the environment in the command prompt:

conda activate casestudyutrecht

Execute the script run.sh. It will first run 10 realisations each for the homemaker (weekday and weekend) and commuter profiles. Note that running the simulations can take a while, the progress will be printed. After completion, overall exposure estimates are calculated for NO₂, PM_2.5 and noise using 5 workdays and 2 weekend days. Six CSV output files with mean and standard deviation values for each agent will be written to the current working directory.

The current setup simulates 1000 agents. If you want to use more agents you can set the query_home_where in the config.py file to a larger value. To use a different number of realisations change the REALISATIONS entry in the script run.sh.

Questions or issues

The most recent version of the modelling framework can be found in the development repository of this project. Please file issues there or contact the corresponding author Oliver Schmitz if you have questions or need support in applying the framework.