Contributor: Dr. Jonathan Bloom, Senior Lecturer
Additional Contributions:
- Mohammad Reza Karimi, Postdoctoral Instructor. Mathematics
- Paolo Gianni, Postdoctoral Associate. Earth, Atmospheric and Planetary Sciences, Selin Group
- Emmie Le Roy, Graduate Student. Earth, Atmospheric and Planetary Sciences, Selin Group
In this project, we created and integrated climate science applications into 18.05: Introduction to Probability and Statistics. Jonathan Bloom co-designed and led 18.05 in Spring 2025 with 123 students. Mohammad Resa Karimi co-taught the course. The course description is as follows:
A unified introduction to probability, Bayesian inference, and frequentist statistics. Topics include combinatorics, random variables, (joint) distributions, covariance, central limit theorem; Bayesian updating, odds, posterior prediction; significance tests, confidence intervals, bootstrapping, regression. Students also develop computational skills and statistical thinking by using R to simulate, analyze, and visualize data; and by exploring privacy, fairness, and causality in contemporary media and research. Flipped subject taught in a Technology Enabled Active Learning (TEAL) classroom to facilitate discussion, group problem solving, and coding studios with ample mentorship.
The primary new materials are three coding projects in R that connect climate science data and concepts to statistical concepts, models, and thinking. Students attend the weekly coding “studio” in the 2–5 PM window on Fridays, taking about an hour to complete the studio. All code is confirmed to run on R 4.3.2 or later. We also added climate applications to two problem sets and an exam. Students loved these applications so we intend to use these materials in 18.05 going forward.
Studio 3: Geospatial climate data and correlation
This studio explores trends and correlations in geospatial gridded data (monthly average temperature) over the past century.
Studio 8: The El Niño-Southern oscillation and significance testing
This studio explores the relationship between the El Niño-Southern oscillation in Pacific ocean temperature and rainfall in the Indian subcontinent, and whether this relationship is significantly changing over the last century. In the process of creating the studio, we realized that students would need additional background earlier in the week to be prepared to fully understand the concepts and complete the studio efficiently. So we held off on deploying it in the spring of 2025 in favor of a more comprehensive approach the following spring.
Studio 12: Climate change attribution and multivariate regression
This studio explores how temperature is driven by natural (solar) and anthropogenic (greenhouse gases, aerosols) forcing, using multivariate linear regression to fit these factors to the observed temperature data and estimate (attribute) how each has contributed to global warming.
Problem Set 5, Problem 2: Modeling the probability of extreme weather events
Over the past 20 years, the high temperature in Boston on February 29 has averaged 39.9 degrees Fahrenheit, with a standard deviation of 12.0 degrees. In problem 2, students use R to model, visualize, and interpret the probability of extreme temperature events based on three underlying distributions for the daily high temperature (uniform, normal, and Laplace distributions).
Problem Set 8, Problem 4: Climate change in Massachusetts, t-tests, and linear regression
Students use R to model the average annual temperature in Fahrenheit in Massachusetts for each year from 1895 to 2023, and to determine if the temperature has significantly increased. The data is in climate-data-MA.csv, which is sourced from NOAA’s Statewide Time Series.
Exam questions not provided in order to preserve them for the future.
