Finding an Optimal Location for a New CalFire Helibase

In 2020 alone, we have been facing an overwhelming amount of wildfires burning throughout California. According to  CalFire's statistics on the 2020 fires ,

• over 9,000 incidents have been reported
• more than 4 million acres have been burned
• over 10,000 structures have been damaged or fully destroyed

These numbers are extremely high and devastating to look at, and the numbers can only go up from here. 

There have been many causes and factors that contribute to these fires, many of which can be attributed to climate change. Some examples of forces that have contributed to the fires include Diablo and Santa Ana winds in Northern California and Southern California respectively, both of which are dry and strong winds, the lighting strikes that occurred in Northern California mid-summer, and even arson or accidents caused by humans. Many families and animals have been displaced, and firefighters are putting their lives at risk in order to prevent the fires from spreading and destroying more areas. We must find ways to combat these fires more effectively and efficiently and prevent more things from being destroyed and lost. 

Our goal is to find the best location for CalFire to establish a new helicopter base. Helicopters are typically used to transport tanks of water or fire retardant as well as transport personnel in a quicker manner, which all help to slow down the spread of a fire. Helicopters can also be used for short-haul rescue.

After analyzing where current helicopter bases are located in  Southern California  and  Northern California , we narrowed our possible locations down to 2 regions: around Bakersfield and around Plumas National Forest. Both of these regions seemed the furthest away from existing helicopter bases, which is why we chose them. However, looking at the map, in which we have chosen to not display vegetation of cover types water, urban, herbaceous, barren, and agriculture, we chose the Plumas National Forest region over the Bakersfield region because the Bakersfield region didn’t really have much fuel to burn in comparison to all the conifer forest/woodland, hardwood forest/woodland, mixed conifer/hardwood, and shrub near the national forest. To be more specific, we chose to analyze Plumas, Nevada, Placer, Yuba, and Tehama Counties, all of which are either a part of or close to Plumas National Forest. 

Within these counties, we analyzed environmental and structural factors, which included large bodies of water, slopes, highways, existing helicopter bases, existing fire stations, and road service/access. To model these factors, we used the following analyses:

1. Network Analysis: to determine the areas that are well-serviced and those that are poorly-serviced by existing fire stations.
2. Suitability Analysis: to combine the network analysis with other factors affecting the suitability of a location, and use this to narrow down the list of possible sites. 
3. Surface Modeling: to find specific areas in suitable sites that of low slope and avoid residential areas.

 Link to full image of the flowchart. 

Projection: UTM Zone 10N

Datum: North American Datum of 1983 (NAD 1983)

For our network analysis, we wanted to find areas that were well-serviced and those that were poorly-serviced by fire stations in Yuba, Nevada, Placer, Plumas, and Tehama Counties. We used the roads and highway data provided by these counties, selected their fire stations as our facilities, and calculated the service area based on travel time in minutes, with breaks of 5, 10, 15, 20, 25, and 30 minutes.

From the map, we can see which areas are well-serviced in green and those that are poorly-serviced in red. It is also evident that there are certain areas that cannot be serviced, possibly due to the lack of roads in those areas. These areas seem to be in the forest area as evidenced by our base map and are areas that may not be highly populated but may be prone to fires.  

We also used service areas of 0-5, 5-10, and 10-15 minutes as a constraint map with weights of -3, -2, and -1 respectively because we wanted to avoid areas that were well-serviced and focus on areas that were further away from fire stations.

For the suitability analysis, we considered what might be needed by both the ground facility as well as the helicopters themselves. We found several factors, both good (“opportunities”) and bad (“constraints”), that would affect the suitability of a location for this purpose. 

While there are only two opportunities, they are important ones, and the final weightings of our model reflect that. Firstly, the ground facility would need to be built on a relatively flat area, because anything too steep would increase construction costs of landing pads and building foundations. Additionally, while pumping water into helicopter buckets is definitely possible, it can be much faster to simply submerge the bucket in a large body of water, such as a lake or reservoir. For this reason, we also looked for areas that were a short flight-time away from a body of water big enough to support such an operation. 

There were many things to stay away from as well. First, we needed to avoid choosing a site that was literally on the water, or very close to it. So, we made a constraint out of the same water bodies that we used for an opportunity above. Similarly, we cannot build on top of or very near to big roads and highways; this is for safety, security, and freedom to build the base in whatever shape is necessary. Thirdly, it would be redundant to build the base too close to the existing firefighting infrastructure, so we made a constraint layer using the location of firefighting facilities. Finally, as an extension of the third constraint, it would also be redundant to build the base in areas easily reachable by ground units. So, using the network analysis from before, we made a constraint layer to reflect that idea as well.

By combining each of these layers as described in the flow chart, we were able to create a very detailed suitability analysis of the area in question, and identify the optimal sites in green for such a facility to be constructed upon. We were also able to identify the least optimal sites in red, which indicated areas we should avoid. Looking only at the areas bounded by the "Counties of Interest" lines, there seems to be potential sites in Plumas County.

While surface modeling was also used in our suitability analysis, we also used it in our final analysis to narrow down potential locations for a helibase site. According to the National Wildfire Coordinating Group Standard for Helicopter Operations, helibase sites should avoid areas with over 11% or 6° of slope. A permanent helibase should also avoid residential areas due to disturbance by low-flying helicopters. Based on our conclusions from the network analysis and suitability analysis, we concluded that we would conduct surface analysis in only Plumas County. In this analysis, we used data from USGS National Map Viewer, U.S. ACS Data, and Plumas County Geospatial Portal.

From the American Community Survey 2019 5-Year Estimates, we found the population density of the 6 census tracts by normalizing population over area. However, we found that the census tracts were each a large portion of the county. To understand where people live in Plumas County, we combined residential land use data with the population density to estimate the density of each residential area in Plumas County. We were able to use those areas as a constraint to siting the helibase.

To create a surface model, we downloaded DEM files from USGS that covered the Plumas County and created slope surfaces. Next, we clipped the slopes with boundaries from Plumas County. Since residential areas should be avoided, we overlaid the residential areas on the surface model. We concluded that there are 2 areas in Plumas County that are suitable for siting a helibase. 

Originally, we planned on analyzing a bigger region that consisted of Placer, Nevada, Sierra, Yuba, Butte, Plumas, and Tehama Counties. However, while collecting data, we discovered that Sierra County and Plumas County required us to purchase their data, and Butte County required us to request their data, which we didn’t know how long that would take. Luckily, we were able to find other data that we could use for Plumas County, but we struggled to find data for Sierra and Butte Counties elsewhere and decided to cut out these counties from our regional scope. This caused us to lose some potential locations for our new helicopter base but also came with the benefit of lessening the data collection load, especially because these counties were small and data was hard to find. 

Nonetheless, the sheer volume of data was still an issue, even after cutting some out as described above. Running raster operations on the very high-resolution DEMs, while rewarding, was very slow, sometimes taking up to an hour to process, and often even crash the program. These crashes resulted in a lot of repeated processing, eating up time that could have been used for further analysis.