More Efficient Post-earthquake Surveys: Assessing Damage Virtually with Drones and Street-level Imagery
In 2016 I travelled to Ecuador after the devastating Musine earthquake as part of the Earthquake Engineering Field Investigation Team (EEFIT). Over the years, EEFIT missions have used film and then digital cameras with GPS stamping to record the post-earthquake situation. On this mission we wanted to explore if advances in technology could enhance mission capabilities even further so we set out to investigate how drones and 360° cameras can be used to improve the findings of post-earthquake surveying.
The equipment we used
EEFIT has been using drones on missions since 2015. In Ecuador we used the drone to photograph landslides, map large-scale ground failures, and observe building failures from a safe distance.
It was the first time on an EEFIT mission that a 360° camera (Ricoh Theta S) was employed. We attached the camera to a selfie stick and captured street-level photos both walking and driving.
This is how we mounted the 360° camera and the selfie stick on a car (©Harriette Stone)
Me walking and capturing with a 360° camera mounted on a selfie stick (©EEFIT)
The result of one of the walks I did with the 360° camera mounted on a selfie stick
The purpose of our study
One of the aims of earthquake engineers and insurance companies after a disaster is to understand the building damage over the whole region: more data on more buildings in the affected region help us to improve our understanding of the vulnerability of buildings. In our study we wanted to explore the potential of virtual damage surveys (where the level of earthquake damage to buildings is assessed) using street-level images as this would help us to gather significantly more information about the damage over the impacted region.
The EEFIT engineers completed damage surveys on the ground along various routes throughout the region, and 360° street-level images were taken along all of these routes. Once the 360° images had been uploaded to Mapillary, an experienced engineer based in the UK completed virtual damage surveys for the same routes. Our hope was that the results from the ground would match the virtual results.
We also used the drone to take aerial images of damaged streets to assess the viability of observing additional information about damaged structures. Additionally, we tested using drone videos to build a 3D model of a damaged building in Portoviejo.
What we found out
The results from the ground and from the virtual surveyor were similar. There was a good match for data collected on the construction type and the number of storeys of buildings along the routes. Additionally, the virtual survey identified whether buildings had generally low or high levels of damage very well, however, more refined levels of smaller damage were more difficult to identify virtually (i.e. when the buildings was only damaged slightly, or if non-structural damage was the only failure). It’s not unexpected, as this usually needs to be identified with closer inspection, which is difficult to provide with street-level imagery.
Aerial images from drones can have numerous benefits to surveying missions. 3D models of an area created from a series of drone images enables engineers to identify damage and failure mechanisms much more clearly than with 2D photos. However, challenges such as flight restrictions, transportation of the drone, and pilot training need closer consideration and differ between countries.
Despite the challenges, the comparative results here show a strong link between ground surveys and virtual surveys. When quantity of data is important, virtual surveys could offer an effective option. With advances in online tools, such as Mapillary and OpenStreetMap, ‘citizen-scientists’ (or citizen engineers!) and volunteer engineers could analyse large amounts of data to rapidly get an idea of post-earthquake damage levels to different types of buildings.
How to improve future missions
Many of the issues with the virtual survey with the 360° camera were due to poor image quality or the lack of photo spheres taken regularly and close enough to each other. Additionally, buildings were often obscured by objects such as trees, walls or vehicles, making judgements difficult for a virtual surveyor. Ground surveyors can often avoid these obstructions by finding an alternative viewpoint. This can, however, be improved with a few adjustments in the process.
- Improve image quality by using more advanced (and expensive) cameras (the model used for this experiment was one of the cheapest available on the market).
- Collect additional viewpoints by decreasing the distance between images. This will increase the data storage burden in the field but will increase the abilities of the virtual surveyors to observe buildings and make the required judgements.
- When views are obstructed by objects on the street, capture images even more frequently. The optimal distance between images is usually between 10-12 m, however, on more obstructed streets (e.g. tree-lined), this could be reduced to allow more chance of useful images. Perhaps a more appropriate distance would be between 6-8 m.
With using drones, a major challenge in some places is understanding the local regulations and obtaining the necessary permissions. The next issue is training competent, experienced pilots to fly the drones safely. When these criteria are met, drone imagery can offer a very valuable alternative viewpoint in addition to street-level imagery and surveys.
As technology advances, it is important that the earthquake engineering community continues to embrace the relevant and useful technological options, employing them in future missions to more effectively learn about the impacts of earthquakes. It will also be valuable to work alongside the platform developers, such as Mapillary, to enhance post-disaster missions in the future.
The complete research report can be found here.
/Harriette Stone, Earthquake Engineer, University College London
If you want to discuss how you can leverage street-level imagery for earthquake post-surveying, get in touch.