The problem

Which areas of Montana are most vulnerable to extreme heat?

As a portfolio project I created a Heat Vulnerability Index (HVI) for Montana at the census tract level. This work closely follows the method developed in 2023 by James Shope, PhD with the New Jersey Climate Change Resource Center and tests the robustness of that approach by additionally performing two variations of Multi-Criteria Decision Analysis using Monte-Carlo simulations.

What I found particularly compelling and useful is how the NJ team developed a practitioner-focused vulnerability index that can be broken down into its three component categories (Exposure, Sensitivity, and Adaptive Capacity) that together constitute vulnerability to heat.

You can view the interactive Montana HVI map below. I’ve also created a detailed Storymap that dives deeper into the methodology and comparisons with the multi-criteria decision analyses, plus three short introduction videos below that explain my methodology & sensitivity analyses. In light of that I’ve left this post intentionally brief.

Approach

I created a short series of videos walking through the project’s methodology.

Part 1: overall methodology introduction.

Part 2: analyzing the validity of the methodological assumptions using Multi-Criteria Decision Analysis

Part 3: a bit more of the census tract-level results from the MCDA sensitivity analysis.

Skills

  • Identify, test, and build upon existing climate resilience methodologies.

  • Programmatically gather and analyze data from a variety of sources. In this project I used R, Python, and Google Earth Engine to gather data via APIs from the EPA, U.S. Census American Community Survey, CDC PLACES program, USGS Landsat program, USFS, and the National Land Cover Database.

  • Test methodological assumptions using Monte Carlo simulations and different weighting scenarios in multi-criteria decision analysis.

  • Build useful data tools like interactive maps and Storymaps to display data analyses and underlying methods.

Tools & platforms

  • R Studio

  • Python

  • Google Earth Engine

  • ArcGIS (Online, Storymaps)

Results

The biggest takeaway from this analysis is that the multi-criteria decision analyses that are built on the work from the New Jersey Climate Change Resource Center provide evidence supporting the robustness of the approach I employ for mapping vulnerability across Montana. Across 10,000 Monte Carlo simulations of differently-weighted indicator variables, the additional analyses generally estimate vulnerability in the same way as the NJ approach, which helps us have confidence that the assumption of equally-weighted indicator variables that the NJ approach is built upon is indeed valid.

Secondly, comparing three different approaches for determining vulnerability demonstrates how much the assumptions underlying estimations of vulnerability matter. While together these three approaches generally agree on how vulnerability should be assigned, clear differences do emerge for census tracts in the hottest and fastest-warming areas of the state depending on the approach used.

A huge hat tip goes out to Dr. Shope and the New Jersey team for inspiring me to dive into this, to Nick Silverman (who provided analytical assistance with the 2017 and 2021 Montana Climate Assessments) for his guidance, and to Brian Norderud at the Montana Department of Public Health & Human Services.

Previous
Previous

Extreme Heat in Montana

Next
Next

Energy Insights — a building energy use & forecasting app