Donate | Join | Print Page | Sign In
Imagination is the Limit: Harnessing the Potential of Drones to Monitor Restoration

Contributed by Rakan A. Zahawi, Associate Researcher at the University of California, Santa Cruz. He will take over as Director of the Lyon Arboretum at the University of Hawaii starting in October 2017. 

Drones have been around for quite a while. Historically the namesake has been associated with rather more ominous tasks that were tied to US military activities in far flung parts of the world. But in recent years they have started to feature prominently in a number of new ways. Perhaps most notably as a new hobby gadget for adults and young people alike. The new generation of hobby drones is much smaller than military drones (most could fit inside a backpack) and, perhaps more significantly, much cheaper. For $1000-$1500 US, you can buy a basic drone that you can simply take out of a box and use. Most are equipped with a built-in camera. Owners can learn to fly a drone in a couple of hours and suddenly have access to incredible aerial footage of their immediate surroundings (both photographic and video). Most hobby drones can fly for about 15-20 minutes on one battery charge. Clearly this can result in some excellent scenic images and novel perspectives that take the selfie concept to a new dimension. But aside from a few nice photos (Figure 1), how can this new technology aid in restoration?


Figure 1. A drone’s point of view of one of our field sites in southern Costa Rica. The site is bounded on two sides by the gravel road. You can discern the three treatments including the more uniform plantation canopy on the left, the control (not planted) with a few sparse recruits in the center, and the island planting treatment on the right (Photo: Rakan  Zahawi). 

We decided to explore this idea further in our long-term restoration project in southern Costa Rica. My colleague Karen Holl (University of California, Santa Cruz) and I started the Islas Project in 2004. We established 13 replicated restoration sites in southern Costa Rica to compare whether planting trees in patches or islands results in a similar rate and composition of forest recovery to the more common practice of planting an entire area plantation-style, or to not planting at all (passive recovery). We collect a broad range of field data that we use to assess recovery parameters, some of which are more labor intensive than others. Indeed, monitoring, as many readers know, is one of the more expensive items in a restoration project budget – both monetarily and in terms of time. So, could there be an automated way to collect some data accurately, and thereby save on both time and money? We decided to explore this possibility using drones to measure structural variables of forest recovery in our plots and compare these data to manually collected measurements in the field. To do this we partnered with Erle Ellis and Jonathan Dandois from the University of Maryland who were working with drones and evaluating their ability to measure forest structural variables. They were eager to test the methodology in a more remote and challenging environment than that on offer in Maryland. 

In July 2013, we flew a drone over each of our restoration plots and equipped it with a simple point-and-shoot camera set to infinity, and programmed it to take a photograph every 2 seconds (Figure 2).

Figure 2: A drone conducting vegetation surveys for the Islas Project in Costa Rica (Photo: Rakan Zahawi).

 As you can imagine, a 15-minute flight captures thousands of photographic images – so what (besides looking at thousands of images of your plots) can you do with this information? We used a type of software called Structure from Motion (or SfM) to process the data. The software can take 2D images and transform them into 3D information – key to this process is a certain degree of overlap between each image taken to create an interconnected network of images that create a reference framework and essentially a mosaic of connected information. We used Ecosynth software to process data (there are now many software options that essentially do the same thing) and the result is a 3D-point cloud that represents the ‘structure’ of the surface that a drone flies over. From this you can estimate the height of a forest canopy, as well as determine how rough the canopy is (a measure of surface texture if you will), among other variables. Determining how tall a canopy is requires knowing where the ground surface is in this point cloud matrix – which can also be determined using the same technology. So, once we had processed the imagery we compared these results to data collected in the field and found that the two data sets were highly correlated. The results were published in the journal Biological Conservation and the article (which can be downloaded for free here) won the September 2015 Elsevier Atlas Award, which recognizes research that could significantly impact people's lives around the world.

The article won this award because this new monitoring technology can help facilitate data collection that is typically both labor- and time-intensive. What’s more, drones can allow researchers to scale up their monitoring programs to cover much larger areas and with the frequency that is needed. Drone monitoring also allows for different types of data collection. For example, we collected data on canopy roughness – a measure of how ‘bumpy’ a canopy is. These data can be collected using LiDAR or satellite imagery but it is far more expensive to do so. Thus, drones open up data collection options to smaller projects and practitioners who are implementing lower budget projects. The versatility of the technology also allows for monitoring of other variables besides height – for example leaf flush and flowering. As the technology improves, we expect that identification of tree species (and all of the possibilities that opens up in terms of data analysis) will also be viable via drone. Indeed, Wesley Rodrigues Silva gave a talk at SER2017 in Foz do Iguassu (S79.04) about a project that is using drones to map the recruitment of the early successional tree Cecropia hololeuca in a restoration project near São Paulo, Brazil. The species is highly recognizable from above due to the silver hue of its leaves. Using drones to map Cecropia hololeuca recruitment will be far more effective than manual surveys, can be done in a fraction of the time, and is more cost-effective than using other remote sensing methods. Finally, using drone technology creates the potential to monitor large-scale restoration efforts across broad areas. This can be highly valuable in remote parts of the world in order to assess whether projects are complying with landscape-scale forest restoration targets. The approach should streamline monitoring and allow evaluators to quickly discern problematic areas that are not in compliance with stated goals. 

There are countless other field-based projects where drones are being tested and used in novel ways – for example as seed dispersal agents in difficult-to-access landscapes around the world (e.g. Droneseed, a startup company based in Washington, USA). Indeed, it seems that the limiting factor in how to apply drones in ecology is our imagination. Drones are already revolutionizing the way we do field work, it will be fascinating to see what role they will play in the near and longer-term future as the technology becomes more widespread and versatile.

Dr. Zahawi was a co-recipient of the Theodore Sperry Award at SER2017 for his significant contributions to the science and/or practice of ecological restoration.

 Previous article  Next article