Building a case for human-induced global warming

Introduction

The aim of this project is to show that the atmospheric CO₂ concentration as a result of human-related activities is causing the average global temperature to rise. We have found evidence, from available data, to support this fact by showing that a correlation exists between temperature anomaly and CO₂ concentration and that no such correlations exist with other climate-related measurements.

Once this evidence is established, our focus is turned to a predictive model of the temperature anomaly and CO₂ concentration. We develop a simple model based on differential equations, and calculate the best-fitting parameters using a least-squares optimization algorithm on the sum of squared residuals between the model and the data. Finally, a Bayesian statistical approach is used to fit the model to the data and obtain appropriate 95% credible intervals.

Directory structure

data contains all data used throughout the project.

• downloads contains the original data before processing.
• processed contains the processed data.
• images contains any generated images.

scripts contains all code written throughout the project.

• preprocessing.ipynb contains the code for pre-processing the downloaded data.
• cross_correlations.R performs the cross-correlation analysis.
• temperature_map.R creates various temperature maps.
• RStanODEModel.R and ode.stan performs Bayesian temperature modelling.
• main.ipynb main Python script collating the analyses.

Data

All of the data used is freely available.

• Temperature data was accessed from the NASA GISSTEMP v4 dataset, which consists of monthly anomaly estimates on a 2°×2° grid from 1880 to present.
• CO2 data was accessed from the NOAA GML dataset, which consists of monthly atmospheric CO2 concentrations at the Mauna Loa Observatory in Hawaii from 1958 to present.
• Volcanic activity data was accessed from the Global Volcanism Program dataset, which details all recorded eruptions in recent history.
• Solar irradiance data was accessed from the NOAA CDR dataset, which contains yearly averaged solar irradiance values from 1880 to present.

This data is preprocessed before analysis.

Requirements and Installation

The Python programming language (version 3.11) is used for most of the analysis in this project, with the R programming language (version 4.3) and Stan probabilistic programming language (version 2.26.1) being used for certain tasks.

• Jupyter notebooks are used for all Python code. The easiest way to install Jupyter notebooks is through the Anaconda platform.

• The standard Python tools for statistical data analysis are used. These include the pandas, numpy and scipy packages. The preprocessing of the data requires the xarray and netCDF4 Python packages. Python packages can be installed using conda install package_name if using Python through the Anaconda platform, or using pip install package_name otherwise.

• The ggplot2, ggmap, testcorr, rstan and HDInterval packages are required for running the R and Stan scripts. R packages can be installed using install.packages("package_name").

Usage

The recommended usage is as follows:

1. Download the contents of data and scripts.
2. Install all necessary software and packages.
3. Run preprocessing.ipynb to prepare the downloaded data for analysis.
4. Run cross_correlations.R to perform the correlation analysis.
5. Run RStanODEModel.R to perform the Bayesian temperature modelling.
6. Run temperature_map.R to generate useful temperature plots.
7. Run main.ipynb to view the analyses and relevant plots together.

License

Authors

Adam Watt
Seán O'Neill