Warriors Garden Solutions

Understanding Correlation to Decode Climate Triggers of the Maui Wildfires

Christopher Ramos • September 18, 2023

How climate parameters likely contributed to the wildfires around Maui in August 2023

Wildfires can be incredibly destructive natural disasters. As researchers, we wanted to analyze the complex interplay between various environmental parameters that may have created prime wildfire conditions, specifically, we analyzed the conditions that contributed to the Kula Fire on Maui. 

In our analysis, we utilize correlation coefficients, a key technique in machine learning. But, what exactly is a correlation coefficient and what does it tell us?


An Introduction to Correlation Coefficients

A correlation coefficient (r) quantifies the strength of the linear relationship between two continuous variables, ranging from -1 to 1. The closer r is to -1 or 1, the stronger the correlation. The sign of r indicates whether the relationship is positive or negative.


- Positive Correlation: r close to 1 means, that as one variable increases, the other also increases. For example, electricity usage typically correlates positively with temperature, since people use more air conditioning when it's hotter.


- Negative Correlation: r close to -1 means as one variable increases the other decreases. For instance, think of the relationship between ice cream sales and winter coat sales. As ice cream sales go up (assuming warmer weather), winter coat sales usually go down, and vice versa.


- No Correlation: r close to 0 means the variables are not linearly related.


*Instances are possible where nonlinear relationships yield coefficients of zero.

Correlation Matrix for Climate Analysis 


We constructed a correlation matrix to analyze connections between temperature, humidity, wind speed, precipitation, and other climate factors in Maui during early August 2023 when the Kula fire occurred. 


The matrix calculates r for each pair of variables. It provides a bird's eye view of the relationships between the different parameters, highlighting patterns that may have driven fire risk. 


For example, we found:

- Strong positive correlation (r = 0.94) between temperature and potential evaporation. Hotter temperatures increased evaporation rates from vegetation, transforming the vegetation into tinder.

- Strong negative correlation (r = -0.91) between humidity and potential evaporation. As moisture evaporated from the ground humidity plunged, further drying grasses and shrubs.

- Gusty winds positively correlated with high temperatures, accelerating the spread of the fire.

Climate Correlation Matrix Kula Fire 1-14 August 2023

Advantages of Correlation Matrix Analysis

The correlation matrix gave us a powerful statistical technique to identify and quantify connections between the climate variables. It allowed us to:


Rapidly analyze many variable pairs to spotlight highly correlated drivers.


Use r values to pinpoint the strongest relationships (such as between temperature and evaporation).

Identify positive and negative interactions that either aggravated or mitigated fire risk.

Establish a mathematical foundation to investigate dynamics between variables.

Kula Fire Climate Scatter Plot Matrix 1-14 August 2023

A scatter plot correlation matrix is an invaluable tool for visualizing the relationship between variables, it provides a clear and concise graphic representation of the correlations and helps to identify patterns and trends in the data.

Key Takeaways

Correlation coefficients are a vital machine learning tools to measure the strength of linear relationships between variables. 


Constructing a correlation matrix enabled us to unravel the complex climatic forces behind the Kula wildfire.


Our analysis statistically quantified how extreme heat, plunging humidity, gusty winds, and drought formed a perfect storm for catastrophic fire conditions.

Follow Us

Keep up with our latest news


Share by: