Unmanned Aerial Vehicles (UAVS) – aka drones – are all over the news. As they get lighter and more affordable, the applications for their use widen. Brian Ritter, an ISA Certified Arborist, Registered Forester, and current PhD candidate in Forest Resources at Clemson University, studied their applications in urban forest inventories, summarizing his research in a thesis titled “Use of Unmanned Aerial Vehicles (UAV) for Urban Tree Inventories.” In his research, Brian examined the scope of drones for urban forestry assessment, evaluated their strengths relative to traditional inventory techniques, and explored direct applications. I spoke to Brian and his adviser, Dr. Christopher Post, about their research. Our conversation has been edited and condensed. –LM
Leda Marritz [LM]: You studied how drones could be used to complete an urban tree inventory. What did you hope to assess?
Brian Ritter [BR]: My goal in the research was to develop a method that would reduce resources and time needed for collecting tree inventory/metrics across the landscape. Within the urban forest, resources are often limited and tree inventories generally have long temporal periods between them. The UAV offered an opportunity to test its application in developing a method to capture tree inventory and at least one forest metric (tree height). The result was a method that reduced time and manpower to capture a level 1 risk assessment of the study boundary along with accurate tree height estimations.
How are data for a typical tree inventory usually gathered?
Christopher Post [CP]: For a typical inventory you would visit each individual tree, either using a GPS with a data collector or a piece of paper to record all the information about the tree. You could end up spending as little as three minutes or as much as 30 collecting the information you need, depending on the level of assessment. Our unmanned aerial vehicle (UAV) research met the requirements for a Level 1 tree risk assessment.
What are the biggest challenges with gathering data on the ground, by hand?
BR: Time is money, and that is the biggest shortcoming of this approach. Having enough personnel to accomplish the task is also a challenge.
Walking, observing, recording, and assimilating information into maps, takes a great deal of time. Using a GPS can reduce the amount of time spent, but even that method is imperfect and can actually magnify errors with placing trees, especially large ones.
CP: Even an expensive GPS may not work that well under a big tree canopy or an area with a lot of density. So there may still be problems with this approach and a fair amount of errors. One positive note is that GPS increases efficiency of data collection by restricting key coding errors if the work is being done by volunteers.
How is using a drone for urban tree inventories different?
BR: Modern UAV systems provide a number of advantages over traditional inventory methods. Time is perhaps the biggest difference; it’s just much faster to do it aerially. It requires fewer people than a “feet on the ground” inventory. UAVs are also low cost, and provide high resolution imagery, current information, short turnaround processing, repeatability and ease of use.
How much time saving are we talking about?
BR: Substantial. Using UAVs we saw a time savings of about 29 days over a traditional inventory. And that was just with two people: me and one other technician. Those represent huge savings per tree and across the whole area being assessed.
There are challenges with the drone approach, I imagine.
CP: Yes, one of those is that you cannot assess or see the trees that aren’t visible on an aerial photograph, like understory trees. You could potentially miss trees in the shadows of buildings. You wouldn’t see visible signs of damage like rot. With downward facing UAVs, which is what we used, it’s mostly about assessing overall health (as seen from above) and location.
BR: On the other hand, with the high resolution UAV imagery, you can see limb mortality from the air that you can’t see from underneath the tree. You also might be able to pick out a dead tree in a stand that you may not see if you just did a walk through. We found some of those that were threatening a building when I was doing my research. It was off the beaten path and not seen every day, so it was missed in the on-the-ground inventory, but big enough that had it failed it could have impacted a person or building.
How many flights did your UAVs do while conducting your research?
BR: We used a professional-grade, extremely light fixed-wing UAV, which is meant for longer flights. We flew 19 trips of 20 to 25 minutes each between July and October 2013.
On each flight, high resolution images were gathered. Just how detailed are they? And how are they different from aerial photography?
BR: Orthophotography is post processed imagery that has been orthorectified (tied to ground control locations) and stitched together to produce a seamless image across the study boundary. Positional accuracy was high (< 6”) using GPS ground control to position the images where they correspond to ground objects. Using true color UAV orthophotography, we gathered 3,466 color images with resolutions of 2.6-3.6 cm from a typical altitude of 90 m.
This is much higher resolution, and provides much greater detail, than traditional aerial imagery. If you look at a typical aerial photo and start zooming in, it becomes very fuzzy. With the images we captured, they don’t become fuzzy when you zoom in – instead, you see more detail.
When I present the images we got, I see people’s jaws drop, the light bulbs go off. They think about all the ways they might apply this.
CP: People love their trees – these beautiful images provide another way for them to see and identify with their urban forest. It’s a method of engagement. To acquire resources to maintain something, you need advocates.
You wrote in your thesis that “UAV applications are in their infancy.” How are UAVs currently being used?
BR: They’re limited in use and scope right now, and are mostly being used for short, specific assessment flights rather than for location and massive data gathering in urban areas. There are multiple types of sensors you can put on a UAV that allow you to capture all types of data and use them in different applications depending on what you what to know.
What are some of the most immediate uses for UAV that you see?
BR: We’re trying to find a way to make this useful for cities and towns everywhere. For example, some towns don’t have arborists. If you are looking for a way to reduce time and expenses, this can be a great solution for a level 1 assessment to determine species and location.
In my thesis I was able to predict good estimated heights from the aerial photography alone, but that’s something we’re researching more right now. We’re also looking into whether diameter and species can also be detected remotely.
The main thing for us is to figure out an efficient, easy way to develop a tree inventory and extract other related information from that remotely collected data.
Can anyone who wants to use a UAV to map their tree inventory do so?
CP: Right now, it’s not something anyone can do. We are anticipating that the FAA will shortly release rules for lightweight UAVs; those rules need to be in place before they become more commercially viable. It’s important to highlight that our primary concern is safety. Urban UAVs have huge potential but we need to make sure regulations are being implemented and adhered to. A single bad operator can ruin it for other people. Having said that, in a few years I fully expect use of UAVs to be a common part of tree inventory.
There are a lot of inexpensive UAVs, but you need a relatively high end commercial one for the applications we’re talking about. We used the SwingletCam model for our research. Including the UAV, the mount, and the camera, it is a little over 1 lb, which is quite incredible. Every ounce that they drop will open up more urban applications.
Are there trainings that urban forestry professionals can take to get more educated about using UAVs in their work?
BR: There is no step by step guide, at least not yet. It still requires some specialized knowledge, software, and training. Training to fly the vehicle itself is not a huge barrier as I see it, but arborist knowledge and image post-processing is a learning curve. It would take some practice before becoming proficient. There are some platforms already available for image processing, and we think more will be available with cloud based online processing available to those that do not want to do it themselves. I don’t think we’re close to UAVs being adopted in a widespread way, but specialized consultants will definitely be able to make inroads quickly.
CP: We expect the barrier to entry to continue to lower in the coming years. With the increasing miniaturization and lightness of new the models, we’ll be able to cover more area at a time with lighter UAVs.
Are UAVs a good solution for tree inventories in every city?
BR: Our focus is on smaller communities, not huge places like Atlanta or Los Angeles. That’s because right now you can cover only 100 to 150 acres per flight; it would take a long time to cover a big city using a single UAV at that rate. However, since inventories are often broken down into component neighborhoods, this may not be an obstacle. If you had an arborist in Atlanta focusing on one area at a time, you could fly individual sections very successfully.
How could this approach change how we do future tree inventories?
CP: A typical foot inventory may take place every 10 years or more. One of the biggest ways UAVs can make an impact is by keeping information up to date. With a UAV, you have repeatability within the same year (or any time period) so that inventories don’t need to be static snapshots for such long periods of time.
It’s such an advantage, and so valuable, to be able to have a visual historical record that tracks trees by location. UAVs offer us an economically feasible way to detect and track change to the urban forest over time.
How can UAV use for urban tree inventories help influence management decisions, policies, and budgets?
BR: So many ways! Many towns and cities have canopy cover standards; UAVs would allow you to fly an area and quickly come up with a canopy cover percentage to track whether you’re maintaining and adhering to that policy.
All the information that’s gathered can go to the person making management decisions about how to prioritize resources. This is particular relevant in arboriculture, where resources are limited and there is a lot of pressure to use them as efficiently as possible. UAV data can also highlight problem areas where you’re lacking resources and provide justification for additional budget, more crews, etc. Decision making at every level can be improved with this information.
Level 2 and Level 3 tree risk assessments would still require visiting trees in the field, but data gathered by UAV can help cities be smarter about focusing their efforts on the areas that need the most attention. Before starting this research, I didn’t realize how well we could pick up mortality in the trees using aerial views, even in dense settings. From that we could infer whether it was caused by disease or an insect, so it helped us prioritize where we need to send crews to do maintenance and work.
I wrote a bit about these benefits in my thesis; here’s a table summarizing those ideas:
Concluding thoughts for people who are in the process of planning for their next urban tree inventory?
BR: Urban forest management necessarily begins with a tree inventory. They are an essential component of developing a plan. But they only represent the conditions at the time of data collection and trees are growing, living things that change over time. Any effective approach to an urban forest industry needs to reflect that.
We can manage the forest from behind a desk to a degree, but this technology allows us to see things we can’t see using the traditional way, gives us information to be better managers, repeatability we didn’t have before, and allows us to archive all that information in a meaningful way. This long-term documentation can be continually tied into software like i-Tree to capture just how valuable those trees are – to demonstrate their environmental benefit to a community.
Thanks, Brian and Chris!