Wildfire Risk & Human Boundaries

A discussion of H3 vs WUI Blocks, in the context of wildfire risk assessment

Athena Intelligence
6 min readJan 23, 2024

Athena Intelligence’s cutting-edge wildfire risk model is an algorithm characterized by its conditionality, geospatial nature, and development through machine learning. It was created to analyze extensive sets of disaggregated wildfire and environmental data, forming profiles that assess the likelihood of a given land area hosting a fire. Every 30-square-meter location (pixel) is assigned a distinct alphanumeric code, traceable back to the source data with a time sequence.

A graphical overview of Athena’s model

While the algorithm integrates information from the US Forestry Services’ Rocky Mountain Research Station and published research on wildfire development and progression, Athena’s emphasis lies in leveraging this information for businesses and other large organizations. Consequently, map boundaries (depicted by the black oval) are integrated into the algorithm and constitute a component of the information included in the output.

The algorithm output is present to customers in various formats such as pixels, raster files, polygons tailored for ArcGIS use, JSON files, numerical expressions of probabilities, as well as Green/Yellow/Red scores for property risk underwriting. Additionally, Voice of the Acre® provides a comprehensive 6-color scoring system for assessing landscape risk, specifically designed for the needs of utilities and municipalities.

In order for landscape-level insights to be valuable to organizations, the information must be visually presented on maps, taking into account human boundaries like a utility’s service territory or a county boundary for disaster planners.

Are Map Lines Arbitrary? Considering the WUI

Before delving into H3, WUI blocks, and the advantages and disadvantages of consistent geospatial partitions, it’s essential to take a step back and examine what is already understood about the Wildland Urban Interface (WUI).

Ignition events resulting from lightning can happen in various locations, yet the majority are caused by human activities such as improperly extinguished campfires, discarded cigarettes, sparks from electrical equipment, children playing with matches, and instances of arson, among others. The progression of small fires into wildfires typically occurs in uninhabited areas. Subsequently, consequential wildfires are generally those that transition from wildlands into areas occupied by humans.

Analysis of multiple wildfires: Graph on the left shows the risk potential 5 miles of the ignition event. Graph on the right shows what type of land actually burned.

Consequently, all wildfire risk models use data related to the Wildland Urban Interface (WUI). These models take into account factors such as the distance between buildings and wildlands, the characteristics of the border between them, and the density of the community near the WUI. Although areas with dense structures may entail a higher economic value at risk, they are typically less prone to losses from a wildfire.

Regions featuring a distinct boundary between vegetation and the community, forming an interface, usually benefit from the presence of an active government entity involved in wildfire mitigation activities. This differs from a community characterized by an intermix of vegetation. For instance, consider two neighborhoods, both situated at the same distance from a Wildland Urban Interface (WUI) area. The community on the right, with a clear interface, tends to have lower risk compared to the intermix community on the left. In an intermix community, embers from the WUI can potentially reach and ignite vegetation within the neighborhood, heightening the risk.

What may appear as an arbitrary line on a map, like the boundary of a park or the transition from a low-density neighborhood with numerous trees to an area densely populated with structures, can carry significant meaning in the context of wildfire risk. These demarcations often signal crucial distinctions in the factors influencing the likelihood and potential impact of wildfires in a given area.

H3, a Geospatial Index that Partitions Maps into Hexagonal Units

The human mind naturally seeks patterns and consistency, a requirement mirrored in software. Hexagons, triangles, and squares are all polygons suitable for tiling.

Hexagons exhibit distinctive features; from the center of a hexagon, the centers of all adjacent tiles are equidistant. The expanding rings of neighbors in hexagons approximate circles. Mapping software utilizing H3 is notably elegant, capitalizing on these unique characteristics of hexagons.

H3 stands out as a geospatial indexing system, employing hexagonal cells to partition the world. This open-source system operates under the Apache 2 license and has gained prominence through promotion by Uber Technologies. Particularly effective in urban and geospatial applications, H3 was purposefully designed to seamlessly integrate disparate datasets, making it well-suited for machine learning applications dealing with geospatial data. Notably, H3 facilitates simpler analysis of movement in various contexts.

The core library of H3geo.org encompasses essential functions that enable the conversion of latitude and longitude coordinates into the corresponding H3 cell. It also provides capabilities for locating both the center of the cells and the bounding geometry. The GitHub community actively contributes to the development and improvement of H3, and the repository can be accessed at the following link: uber/h3.

WUI Blocks

In contrast to the systematically spaced and elegant H3, WUI blocks, which sometime align with census blocks and other times aggregate up into a census block. Census blocks roll up into larger census tracks, are indicative of population density.

The outputs generated by Athena Intelligence’s wildfire risk model, which consolidate the granular pixels, are available in either H3 or WUI blocks. Despite the computational advantages associated with H3 data, new customers are frequently initially surprised by Athena’s recommendation of the irregularly shaped WUI blocks.

Empirically, the team at Athena has observed, both in historic wildfire footprints and real-time wildfire observations, a peculiar tendency for wildfire perimeters to align with WUI block boundaries. Reflecting on the Wildland Urban Interface (WUI), wildfire behavior, and human activity, the team has recognized that in rural areas, census and WUI block boundaries often align with features like roads, highways, streams, ridges, or other landscape changes that create natural fire boundaries.

This phenomenon is less prevalent in urban areas and occurs even less frequently in densely populated suburban zones. The effect is particularly pronounced in areas characterized by low population density.

The WUI provides some of the best information about wildfire risk, based on proximity to people. H3 offers some of the best information about the landscape, with no reference to humans and their impact on the land. Census blocks combine both, but are closer to the WUI in information, reflecting population density, roadways and other impacts of humans on the land.

Consequently, Athena encourages new municipal or utility customers to initially explore WUI block data at a discounted rate, recognizing the valuable insights it can offer, especially in areas where these natural fire boundaries play a significant role.

This does not occur in urban areas and occurs less frequently in dense suburban zones. The effect is quite pronounced in areas with low population density.

Athena Intelligence is a data vendor with a geospatial, conditional, profiling tool that pulls together vast amounts of disaggregated wildfire and environmental data to generate spatial intelligence, resulting in a digital fingerprint of wildfire risk.

For power companies, this helps them assess, a year in advance, where a consequential fire is most likely. For insurance or financial services companies, this assists in underwriting, portfolio risk optimization and loss cost probabilities for reserving.

Contact us at info@Project-Athena.com and follow us here.



Athena Intelligence

Athena Intelligence weaves vast amounts of disaggregated environmental data. Drop us a line (Info@Project-Athena.com), or visit www.athenaintel.io