Discovering Patterns in Global Wildfires

A GIS Approach to Measuring Change

Australia. Brazil. California. Canada. Siberia.

What do these places have in common? All too recently (or regularly) they have been ravaged by wildfires, either natural or anthropogenic in origin.

Not only does it seem like these fires are getting larger, more deadly, and more frequent, it seems like they're becoming more common across the globe.

How can we distinguish actual patterns from perceptions?

In this Story Map, we'll be examining patterns in global wildfires using data from NASA's Aqua satellite. Aqua and its space sister Terra have been orbiting the planet for around 20 years. The satellites' MODIS sensor has been used to track changes in the ocean, on land, and in the atmosphere.

And since MODIS can also detect small thermal signals, it can also be used to track changes in wildfires.

To be clear, I'm lumping a lot of different fire types into the term "wildfires" to include things like prescribed burns, agricultural fires, biomass burning, and true wildfires.

Examining Regional Patterns

The burning of the Amazon in the summer of 2019 captured the world's attention. Let's take a deeper dive into the same MODIS wildfire data to examine patterns in space and time across the Amazon.

Are there meaningful patterns in this data? How can we get objective, data-driven insights into wildfire frequency and trends?

A GIS Analysis of Wildfire Patterns in the Amazon

To simplify the data and patterns, we can also aggregate the millions of points into hexbin summaries.

Next Steps

I'll be updating this Story Map with analyses of other regions and also integrating population data from Living Atlas to being to assess areas of risk.

Using spatial analysis tools in ArcGIS Pro, such as Create Space Time Cube by Aggregating Points, and visualizing the results can allow you to easily repeat such investigations of wildfire activity in other parts of the world.

Methodology

The NASA FIRMS data for MODIS includes multiple sources of thermal detections. All but the "presumed wildfires" (Type 0) have been removed from the analysis. Additionally, only Aqua observations during the daytime have been included. Since confidence level accuracy varies around the planet, all have been included.