ADS506 Final Project: Global Warming Trajectories

Purpose and Overview

This project explores the link between anthropogenic CO₂ emissions and global temperature change. It combines historical climate data, statistical time-series models, and an interactive Shiny application that lets users test emission scenarios and see their projected temperature impacts. The repository also contains a Quarto document outlining the exploratory data analysis (EDA) and modeling workflow.

Key Findings

• Berkeley Earth monthly anomaly data show a pronounced shift from long term stable temperature conditions in the early record to persistent warming over the last ~40 years, reinforcing the acceleration of climate change.
• STL decomposition of the anomaly series highlights a strong long-term trend layered on top of seasonal variation, underscoring the need to model both components when attributing changes to human activity.
• ARIMA models combined with CO₂ concentration (ppm) and ENSO signals, provide the backbone for temperature projections in the app, enabling scenario comparisons between baseline emissions and user-adjusted trajectories.

Repository Structure

• shiny_app/: Shiny UI/server code for the Global Warming Trajectory Simulator (app.R, legacy prototype app_main.R).
• data/: Preprocessed RDS files used by the app and modeling pipeline (e.g., temperature conversions, ENSO-lagged predictors, country baselines).
• appendix.qmd and appendix.pdf: Quarto analysis notebook and rendered PDF with the EDA and modeling narrative.
• data_preprocess_scripts/: Scripts for transforming raw inputs (not run automatically by the app).

How to Reproduce the Analysis

1. Restore the R environment

   install.packages("renv")  # if needed
   renv::restore()

2. Regenerate the appendix (optional) Knit appendix.qmd to reproduce the EDA/modeling figures and tables:

   quarto::quarto_render("appendix.qmd")

   The compiled report is available as appendix.pdf for quick reference.

Running the Shiny App Locally

1. Ensure the data RDS files in data/ are present (they are tracked in the repo).
2. From the project root, start the app:

   shiny::runApp("shiny_app")

3. A browser window will open with three main tabs:
   • Historical Explorer – Plot historical global temperature anomalies, total CO₂ emissions, or CO₂ concentration; optionally view STL decompositions; hover for exact values.
   • Forecast & Scenarios – Configure forecast horizon (5–20 years), ENSO conditions, and emission adjustments (percentage growth or absolute changes). Compare user scenarios to the precomputed baseline temperature path.
   • Impact Report – View monthly or yearly averages of baseline vs. scenario temperatures, cumulative emission deltas, and download the results as CSV.

Data and Modeling Notes

• Temperature inputs come from Berkeley Earth global temperature anomalies converted to absolute temperatures for seasonality-aware modeling.
• Emission baselines use 2024 territorial and land-use change estimates (~43.27 Gt/year) to anchor forward scenarios; country-level starting points are read from country_2024_baseline_Gt.rds when available.
• Forecasts leverage ARIMA models that couple CO₂ concentration, ENSO, and seasonal Fourier components to project temperature anomalies and highlight differences between baseline and user-defined emission pathways.

Building/Deployment Tips

• The Shiny app runs entirely from precomputed RDS datasets; no external API calls are required.
• For deployment to a Shiny server or Posit Connect, bundle the shiny_app/ directory and ensure the data/ folder is accessible at the project root (or adjust paths accordingly via here()).
• The project uses renv for dependency management; deploying with an renv.lock-aware service will reproduce the tested package set.

Navigation Cheatsheet

• Explore tab: Choose variable → optional STL toggle → adjust date slider → hover to see values.
• Forecast tab: Pick scope (global or country), horizon, ENSO state, and adjustment mode → click Run Scenario → hover lines to inspect exact temperatures; summary value boxes show temperature change and cumulative emission difference.
• Impact Report tab: Switch granularity to view monthly or yearly aggregates and download the scenario comparison.