California Wildfires (Part 1): Mapping the hardest hit neighborhood
By Greg Crutsinger, Drone Scholars
I’m not a disaster drone guy. I am actually an ecologist by training.
I started in the drone industry because I wanted a better way to map some large tree experiments I was running (measuring leaves by hand is tedious and time-consuming). I quickly saw the potential of drones as a transformative technology and was all in. I left the safety and predictability of a tenure-track faculty job in academia for a leading drone company in the San Francisco Bay area. It’s not a secret that the past few years have been tumultuous in the drone world. Since joining the industry, I have been through two drone hardware companies and a software company. I recently founded Drone Scholars to travel less and focus on vegetation mapping more…not emergencies or fires.
Like many in the Bay Area, I woke up on Monday, October 9th to the news of wildfires in Sonoma County and the smell of smoke in the air. The destruction and loss of life I read about were heartbreaking, particularly for Santa Rosa. It wasn’t until a few days later that I got a call from the local sheriff’s department whom I had consulted with on a few prior occasions. Their UAV team was on the ground and could use some volunteer assistance on processing imagery.
What was supposed to be a quick mapping project of a mobile home park spilt over to Coffey Park, one of the largest and hardest hit neighbourhoods with several thousand homes lost. Street after street reduced to ashes. And not just houses. Cars, gardens, trees, toys….everything gone. It is terrible what this community has experienced.
The sheriff’s department had received an emergency Certificate of Authorization (CoA) from the FAA to fly within the Temporary Flight Restriction (TFR) to support that county. However, the initial permission was at only 100 ft above ground level because of the manned aircraft entering and leaving the area as part of the fire monitoring. Ultimately, it would take the UAV team three days, up to four DJI Phantom 4s at a time, two dozen flights, 10,000 photos, and 50 gigabytes of data to cover just over 350 acres of neighbourhood homes. The officers did all the flying. I assisted in creating a patchwork of overlapping missions and in processing the results.
During those three days, there were many of the standard drone issues. Glitchy apps would crash or fail to record photos. Batteries were wasted and flights needed to be reflown. At the end of each day, we sat around tiredly waiting for thousands of files to transfer onto different laptops and thumb drives. Volunteers would come around with drinks and food enabling us to continue mapping. It was amazing to watch the Sonoma community pulling together.
Ultimately, the officer’s dedication to collecting vast amounts of data for the county helped make everything come together. We had mapped the majority of one of the hardest hit neighborhoods in the most destructive wildfires in California history down to the square inch. You can see the map of Coffey Park that has been made publically available here:
https://www.mapbox.com/bites/00382/#18/38.4763/-122.74861
The next hurdle was the processing. I tried to do this in sections over several days. My home office becomes a hub of crunching data, uploading photos from drives, and sharing finished portions of maps with the coordinating agencies. It quickly became clear my computer couldn’t handle the dataset. It was too big and my laptop was simply unable to keep up. The San Francisco office of Pix4D helped merge the full dataset together into one large orthomosaic.
The remaining challenge was to get the dataset into a manageable format to view online. After reviewing a number of options, I called Mapbox in San Francisco who graciously donated technical assistance. Within 24 hrs, they had a map with all the streets overlaid and each individual address visible when you click on a home.
This map also shows camera icons linked to 360 panoramic photos collected from the free Hangar 360 for DJI app (see Part 2 of this story for details).
In total, it took 6 days to map, process, and get all of the data to the coordinating agencies. Sonoma County released the maps publically to help streamline the process of allowing residents to return to the Coffey Park neighbourhood. The Sonoma County GIS team has since incorporated the drone datasets into a larger, publicly-available map using ESRI products.
These maps include some additional areas that were collected by other UAV teams. Also included are satellite data of impacted homes in neighbourhoods were not mapped by drone. It is estimated that at least 8,400 homes and buildings across three counties were destroyed by four different wildfires and the death toll is currently 43 individuals.
This experience reminded me that the industry has a ways to go before there is seamless workflow. Successful mapping requires the right drone flying the right version of both drone firmware and 3rd-party app software on the right phone/tablet. There are lots of tiny microSD cards to lose. A generator is needed to keep drone and phone batteries charged in the field. Processing huge datasets can take serious computing power and several days. Sharing the results quickly and effectively requires GIS tools to overlay multiple data layers and a good internet connection. I’m sure the UAV teams in aftermath of hurricanes in Texas, or Puerto Rico or other emergencies (e.g. earthquakes in Nepal) have had similar experiences.
My takeaway is that there are still many technical issues that don’t go smoothly under normal circumstances, let alone a disaster. Perhaps there are more simplistic approaches that could be used in the near term to complement full-scale mapping for emergencies (see Part 2 for one approach tested).
This article is simply my notes as an observer during the wildfires. My opinions are my own and do not reflect those of the agencies involved. Thank you to the Pix4D, Mapbox, and Hangar teams for all of the technical assistance. Thank you to thousands of local, state, and federal agencies, as well as all the volunteers, in their heroic efforts during these tragic fires and the long road of recovery to come.
Meanwhile, I’m back to mapping vegetation and training others to use multispectral imagery at Drone Scholars. I’ll be leaving disaster management to the professionals.
