Allen-Michael Moench Vector GIS 1040, Front Range Community College 5/8/24

Predicted Wildfire Behavior and Risk near Gold Hill, Colorado

Abstract

The town of Gold Hill, Colorado exists in an environment prone to wildfires and wildfire-related destruction. Understanding how wildfire ignites, spreads, and travels across the landscape is essential to protecting the structures and values at risk in this type of “Wildland Urban Interface” environment. This project conducts a predictive analysis of possible wildfire behavior and risk to structures in Gold Hill. To accomplish this analysis, data was used that related to predictive wildfire behavior criteria such as fuels, weather, and topography. The data was manipulated using geoprocessing tools and analyzed based on the author’s own experience and study as a wildland firefighter to identify areas of high “risk” and potential extreme fire behavior. “Risk” in this case is a metric which will be explained below. The Results show that Gold Hill is particularly at “risk” from lightning and human-ignited fires which begin in light fuels such as grass or shrubs, on steep slopes, on South aspects, or during strong wind events. Areas where these factors operate together represent particularly high “risk” areas where fire is more likely to start, more likely to spread once it starts, and more likely to subsequently grow in size and intensity. The areas of highest risk identified by this study include the saddle at the East end of town, the ridge also to the East, the dispersed camping to the West, and the steep slopes to the North and South of the plateau or ridge on which the town sits.

Introduction

Wildfires pose a significant risk to structures in Colorado’s front range, especially in a zone called the WildlandUrban Interface (WUI), where dispersed homes and settlements have pushed further and deeper into territories that were previously considered to be wilderness. Fire is a natural process of the landscape and ecosystem in these areas, however humans still chose to build in these locations which can put their communities at risk. The town of Gold Hill is an example of such a location, where locals have settled into what was once an old mining town and is now a hub of musical entertainment, outdoor recreation, and retreat. Gold Hill was nearly burned to the ground in the Fourmile fire of 2010, and multiple other historic fires have taken place nearby, sometimes with devastating effects to the local communities. Gold Hill is considered by many to be an idyllic and historic mountain paradise, however the local environment’s tendency towards devastating wildfires puts its structures and its community in a position of considerable risk.

In order to mitigate this risk, it is essential to understand the ways in which fire behaves, and particularly how it travels across the landscape. This study assesses the three major factors that influence wildland fire behavior: fuels, weather, and topography. It does this by comparing datasets which model elements of these three factors such as slope and elevation (topography), and fuel model type (fuels). This paper will discuss major geographic areas near Gold Hill where fires could be ignited, and areas where fire is likely to spread or travel across the landscape. Methods

Analysis began with an extensive review of data sources, intended to locate information regarding fuels, weather, and topography. Sources that were identified as having relevance to this project included Digital Elevation Models, maps of fire protection districts, Boulder County’s Community Wildfire Preparedness Plans (CWPP), a wildfire risk viewer produced by an analyst from Lefthand FPD, sources regarding historic wildfires, and in particular datasets from the Landfire website https://www.landfire.gov/, which contained useful raster datasets. These raster datasets provided the foundation for much of the analysis.

After downloading, vector and raster layers were added to a new map frame in ArcGIS pro. All vectors and rasters were projected to the NAD 1983 UTM Zone 13 N using the project tool and the project raster tool, respectively. Data layers were symbolized using appropriate color schemes so that relevant patterns could be observed. Several data layers were found not to be useful, including a DEM from Boulder County that consisted of numerous .img files. An attempt was made to use the mosaic tool to combine these into a single raster, however this proved ineffective. The possibility of georeferencing a USGS topographic map of Gold Hill was also briefly considered, however deemed to be impossible due to time constraints.

Next, a new map frame was opened and the most useful .tif rasters were added. The Clip tool was run on these to limit them to an approximately 78 square kilometer box centered on Gold Hill. In order to manipulate the data as vector data, the Raster to Polygon tool was run on each .tif file. To each of these new polygon files, two new text fields were added, “􀄩hav” and “risk”. The author used personal experience and study as a firefighter to fill out these fields, also referencing Anderson’s 13 fire behavior models (https://www.fs.usda.gov/rm/pubs_int/int_gtr122.pdf). Information was included that detailed expected behavior based on fuel model, slope, and vegetation cover. “Risk” was determined by category based on the likelihood of a fire to ignite or spread rapidly in a particular fuel model, accelerate up a steep slope, or based on the possibility of crown fires or spo􀆫ng.

Next, new feature classes were created based on areas that were considered to be “high risk”. These were fire models in the grass and shrub types, since these represent the finest fuels and the highest probability of iginition, and steep slopes, since these can dramatically increase the rate of a fire’s spread. The fuel model layer was then clipped to the slope layer using the clip tool, and the feature was exported to show areas where a steep slope and a “high risk” fuel model overlap. These are the areas with above average risk, where fires could be ignited and spread quickly. Results and Discussion

The raster layers downloaded and displayed in the map frame provide useful reference information for observing the characteristics of the landscape and making predictions of expected fire behavior and risk. For example, the Digital Elevation Model can be used to see the saddle at the East end of Gold Hill’s downtown area, which would tend to funnel flames in the event of a fire. The ridge to the East of the town is also visible, and is a feature which could attract lightning. Other important features are the steep-walled valleys to the North and South, which could catch fire in the event of an alignment with a strong west wind (such as the “chinook” winds which blow over the Rocky Mountains from the West). The fuel model map is highly informative, since fire would move differently in light fuels such as grass and shrub (most prevalent on the south aspect of the valley to the town’s south) versus timber and heavy fuels, which are more prevalent to the North of the town. Conclusions

It is impossible to predict with certainty where any fire will ignite, or how it will travel. However, the results of this analysis place the greatest risk to the town on a fire that begins to the West during a wind event, perhaps ignited by dispersed camping in that area. Such a fire could rapidly spread through the grass and shrub fuel models to the South of the town (on the South aspect of the slope that drops down towards Fourmile Canyon). Lightning strikes on the ridge also pose a possible threat, as do uphillclimbing fires ignited in Lefthand Canyon to the North of the town. Such a fire could be devastating to the historic buildings in the downtown area, as they are situated in a natural saddle which would tend to funnel the heat and energy of a fire. Raster-based imagery from the Landfire website provides a good allaround resource for response planning and mitigation efforts, should these be desired in the future.

Lessons Learned

The project provided ample opportunities for beginning to learn techniques for manipulating rasters. The raster to polygon tool was particularly useful. This was also an excellent opportunity to dive into fire data, and see what types of sources exist as well as which ones are most useful. Given the opportunity to continue this research in the future, there are many directions it could be taken. For one thing, the methodology could be applied to other geographical areas. Given more time, I would like to locate and track down sources of lightning and wind data, and spend more time comparing and working with the various datasets.

Works Cited

• Wildfire Risk Viewer (Rod’s project) https://co-pub.coloradoforestatlas.org/#/
• Landfire map viewer: https://www.landfire.gov/getdata.php o Us_230 Existing Vegetation Cover o Us_230 Existing Vegetation Height o Us_230 Existing Vegetation Type o Us_230 13 Fire Behavior fuel models o Us_230 40 Fire Behavior fuel models o Us_220 Aspect o Us_220 Elevation o Us_220 Slope (Degrees) o Us_220 Slope (Percent Rise) o Us_220 Operational Roads
• Colorado GIS clearinghouse data: fire protection districts https://geodata.colorado.gov/datasets/05808f28d9414845b7cbf9d4f55eb4ec_0/explore?location=40.091573%2C-105.363012%2C9.60
• Boulder CO Wildfire History https://opendatabouldercounty.hub.arcgis.com/datasets/61f20f4a64274969a9e740eda5c62de7_0/explore?location=40.075323%2C-105409900%2C10.84
• Boulder County Wildfire zones https://opendatabouldercounty.hub.arcgis.com/datasets/d2cad1928b74496ea8185e6ccc8328cf_0/explore
• Boulder County Community wildfire protection plan (CWPP) - Forestry Management Projects
• Boulder County Community wildfire protection plan (CWPP) – Community Boundaries
• BoCo roads: https://opendatabouldercounty.hub.arcgis.com/datasets/f8292cbf379e4df7b9b8f62e21120ea7_0/explore?location=40.054336%2C-105.395172%2C13.46
• Boulder County DEM: https://opendatabouldercounty.hub.arcgis.com/documents/1e9edf8038d04a􀄩b0d34f99ccbb4f10/explore
• Digital Elevation Model: https://srtm.csi.cgiar.org/download
• ArcGIS Living Atlas: National Risk Index Census Tracts, Colorado Census 2020 Redistricting blocks
• 7.5min DEM: http://www.webgis.com/terr_pages/CO/dem75/boulder.html