Mapping environmental injustices within the U.S. prison system

This is the working repository for NASA-funded research mapping environmental injustices within the U.S. prison system. The project is currently in Phase 2, building on the initial Phase 1 analysis (see Project Versions below).

This repository hosts the workflow used to carry out the spatial analysis, from data retrieval to producing the final dataset that is hosted (here)[https://doi.org/10.5281/zenodo.8306892]. Below are further details on the repository structure, how to use the code base, and data sources.

Forward any questions reguarding this project and code base to Caitlin Mothes (ccmothes@colostate.edu)

Project Versions

This repository contains analyses from two phases of NASA-funded research:

Phase 1 (v2023-1) - NASA ROSES-21 A.49 (Award No 80NSSC22K1465) - Published: August 31, 2023 - DOI: 10.5281/zenodo.8306856 - Status: Initial environmental justice analysis for U.S. state and federal prisons - Archived code: GitHub Release v2023-1

Phase 2 (in progress) - NASA ROSES-23 A.47 (Award No 80NSSC25K7033) - Updated datasets and extended analyses - Release and DOI forthcoming

Citation: For the overall project, use the concept DOI: 10.5281/zenodo.8306855. For specific versions, cite the version-specific DOI listed above.

File Organization

NASA-prison-EJ/
├── README.md
├── setup.R
├── LICENSE
├── .gitignore
├── NASA-prison-EJ.Rproj
├── process_indicators.R
├── process_indicators.Rmd
├── process_components.Rmd
├── R/
├── python/
├── process_data/
├── analysis/
├── figures/
├── outputs/
├── jobs/
├── doc/
├── working_scripts/
├── phase1/                    (ARCHIVED - Phase 1 complete)
│   ├── R/
│   ├── python/
│   ├── process_data/
│   ├── analysis/
│   ├── figures/
│   ├── outputs/
│   ├── jobs/
│   ├── doc/
│   ├── process_indicators.R
│   ├── process_indicators.Rmd
│   ├── process_components.Rmd
│   └── README.md
└── data/                      (Local only - not in git)
    ├── phase1/
    │   ├── raw/
    │   └── processed/
    └── phase2/
        ├── raw/
        └── processed/

Folder Descriptions

• docs/ Project-specific references and metadata.

• data/ This folder is stored locally (see Github_Collab_Guide.docx in docs/ for methods on how to set up a local data folder that syncs to a cloud-hosted folder). All raw, unprocessed data lives in raw/ and all datasets processed throughout the workflow (using code in process_data/) go to processed/ . GEE-processed datasets also live in this processed/ folder.

• process_data/ Scripts to process raw data (when needed). Outputs of these scripts go into data/processed/

• R/ R code for analysis, each formatted as an individual function for each indicator calculation (i.e., all indicators not requiring Earth Engine data sets).

• python/ Python code for Google Earth Engine analyses.

• jobs/ Job scripts to run background jobs for each component function (See 'process_components.Rmd' for executing these).

• outputs/ Where all outputs from each indicator function, component function and final data frame calculation are stored. Each file is set up to include the date of creation when saved to this folder.

• analysis/ R code for all post-hoc analysis of final data set.

• figures/ Figures produced from analysis scripts.

• working_scripts/ Old or in progress scripts outside of current workflow. External collaborators can disregard this folder.

File Descriptions

There are a few important files in the root directory:

• setup.R Executing this script will install and load all necessary packages, source all functions, and create a local 'outputs/' directory if it does not already exist.

• process_indicators.R / .Rmd This file (saved as both an R script and R Markdown for user preference) works through executing indicator calculations individually.

• process_components.Rmd This file walks the user through calculating indicators as grouped component scores. It also has an option for executing the component functions as background jobs (as some may take days to full execute due to large datasets and scale of processing).

General workflow

The main files described above give users the option to run individual indicators or process the entire component scores all at once, and each work through creating a final data frame as the last step.

This workflow is designed to work on ANY spatial polygon sf object. While this project specifically studies state and federal US prisons, the code can be applied to any other boundary shapefile, greatly expanding the utility of this code base and allowing for future researchers to analyze other areas of interest (e.g., extend to jails and juvenile detention centers, or calculate measures for other communities/neighborhoods to assess comparisons with risks faces by prisoners).

While some functions and scripts retrieve the datasets directly from the code base, some need to be downloaded manually. The table below provides links to download each of the datasets used in this project, and direct users to the processing script used to clean the raw data download (if necessary).

Data sources and indicator methods

A full metadata table with data sources, links, indicator methods and data metadata can be found in docs/indicator_metadata.md