Massive potential for drones in forestry

• 

From data collection on the health of trees to monitoring plantation, road and river crossing conditions on the ground or doing pre-plant sprays, drones are the new ‘go-to’ technology that is busy transforming the way foresters go about their daily business. This Q and A with Simon Ackerman of the Forest Operations Research unit at the Department of Forest and Wood Science, Stellenbosch University, focuses on the potential for drones to provide radically new data driven solutions for forestry…

What is the buzz around drones?

SA: To some, drones are seen as a bit of a gimmick, or an annoyance with potential for contravening just about every clause in the POPI act. However, both research and industry are proving that drones truly represent an amazing new technology with the potential to produce radically new data driven solutions in land-use applications such as agriculture, forestry and conservation.

How can awareness of the apparent opportunities for using drones be improved?

SA: The Department of Science and Technology, through the Forest Sector Innovation Fund, awarded a project for the promotion and development of Precision Forestry Tools to improve the efficiency of forest operations in South Africa. Part of this funding was in support of a series of workshops in the forest sector. The Department of Forest and Wood Science at Stellenbosch University sees a need for the forest sector to take a more proactive role in ensuring a sustainable adoption and roll-out of drone-based sampling methods and services. This includes capacity building amongst our students, drone operators, and industry players alike. Awareness of this can be generated through articles, social media, and the road trip we embarked on. An increased use of drones is obviously going to benefit both research and practice.

• 

What were some of the main goals of this ‘campaign’?

SA: The main goals of the project, and the workshops, were to:
• Upskill the forestry industry in the application and use of tools to measure and manage our resources through the use of UAVs, including the advanced processing of these data. This is being done partly through postgraduate student projects and industry workshops
• Complement existing and develop new forms of data and data sets for use in our industry
• Create the links through academia and industry to skills both locally and through our cooperation partners abroad who have adopted drone based data in forest research to a far greater degree than we have.

How was this implemented practically?

SA: The Forest Operations group (FOR) arranged a series of workshops on ‘Drones in Forestry’ targeted at both established drone operators who needed to know more about the specifics of working in the forest sector, as well as forestry management and research staff who were interested in knowing more about potential applications, potential outputs, or just in exchanging experiences.

Why push for an increased use of drones in forestry?

SA: Drones are incredibly convenient tools for the collection of data in forestry. Plantation forests are typically between 3 000-10 000 ha, and have individual management units of 10-30 ha which need to be sampled for survival, health, growth and stocking densities more or less throughout their 6 to 25 year rotation. This is normally done on foot or through remote sensing, typically with LiDAR which has its benefits, but has a lower temporal and spatial resolution. Methods for analysing drone based data are rapidly evolving within and between research environments around the globe, and these developments provide a rich basis for research, and not least, research publications. We see lots of opportunities here, both for our industries and for ourselves, especially given that some of the species that we use have quite different attributes to those grown in Europe or North America where a lot of the research is currently being done.

What are the main drone payloads or sensor technologies of relevance?

SA: For a large part, drones carrying RGB sensors (i.e. normal cameras) are more than sufficient for the purpose of forest measurement. Through specialised software that creates stereopsis through a process known as structure-from-motion (SfM), a series of RGB images with sufficient overlap can be processed into a 3D point cloud. The point cloud is used in creating a 3D surface or digital surface model (DSM). Even consumer grade drones such as the DJI Mavic 2 with integrated camera easily meets or exceeds the needs for small area surveys in very high resolution.

A second commonly used payload is a multispectral or hyperspectral camera. A lot of people would know that these have been used for many years in agricultural settings, as the wavelengths captured in multispectral imagery can be used both in distinguishing between plant species (e.g. crop and weed), indicate the health of the crop (water stress, nutrient deficiency) or between live and dead or dying biomass, e.g. in evaluating effectiveness of a herbicide treatment.

Physical direct measurement through laser scanning (LiDAR, laser distance and ranging), using drone borne scanners, which was once prohibitively expensive for forestry applications, is becoming economically feasible and the applications are expanding accordingly. The benefit of LiDAR over photogrammetry based 3D models is that the LiDAR pulses are able to penetrate any gaps in the tree canopy and provide information from lower down on the stem or from the ground, which is seldom discernible in RGB data. Also, at the level of detail we can work with using drone data, it is quite essential to have a high resolution terrain model as well, and that can really only be measured with LiDAR.

Drones are also widely used in providing services, such as aerial herbicide or pesticide applications, something that is more well known in agriculture / viticulture. A couple of these have been well covered in the SA Forestry Magazine. Interestingly, they show a strong competitive advantage over tractor borne or manually applied chemicals when it comes to being able to apply the dosage precisely and in a timely way, obviously also providing access when trees get beyond a certain practically reachable height.

• 

How do drones fit in with other platforms used for remote and proximal sensing?

SA: Very well actually. Drone based data is mostly sampled in nadir, i.e. vertically, and is therefore 100% complementary to fixed-wing aircraft platforms (LiDAR and imagery) and satellite based data (imagery and radar). Each have an important role to play and will continue to do so. High resolution drone imagery can be used in calibrating the interpretation of imagery from the other platforms. Aircraft can cover large areas far more economically than drones can, but are not ideal for smaller areas or specific sites, while satellites offer data with a high temporal frequency (daily) but at a lower resolution. So, ideally, one will always be working with a portfolio of data from different platforms.

Can anyone fly a drone?

SA: Actually flying the drone is the least of the challenges. Usually a survey flight is pre-programmed on a tablet in the planning office, and flown without human intervention. One does however need to have the skills to step in if something unexpected happens, and that does happen more often than not. In South Africa, the use of drones in a commercial setting is strongly regulated by the civil aviation authorities (CAA). A commercial operator is defined as anyone that receives any sort of remuneration for the work, whether in the form of a bottle of wine or even co-authorship of a research paper. We were fortunate to have Robert Britz of DroneX join our series of workshops and give a very thorough run-through of the regulations, requirements and obligations, tailor-made for our industry.

What do you think about these regulations that are obviously aimed at the public and the protection of private property?

SA: It’s a bit of a paradox that here in our country, one can get a truck licence for R 3 000 and drive a 45 ton truck down a crowded main road at the end of an almost unlimited shift, while it costs ten times that (R 30 000) to get a licence to fly a 950 gram drone in a remote forest area, and one must also pass a medical exam. There’s some room for lobby work here. At FOR, we are planning to engage the industry in jointly applying to CAA for a couple of exceptions to the current regulations. We can also do more in regulating our own operators. Of course uncontrolled altitude is a threat to aviation, but this can be restricted in the software, and in cases where the fire bombers are working within the 100 m envelope, we as an industry should really take responsibility in developing our own protocols and regulations. If we are going to exploit this new highly valuable data source, we really need every forester to have a drone in their bakkie. They have to be able to fly a compartment at short notice to document e.g. fire damage, windfall, insect or pathogen attack, harvesting or thinning progress, amount of timber on the landing, or the condition of a road or stream crossing. There is no question that the benefits will far outweigh the costs. Until then, we need to train licensed operators on the peculiarities of forests and forestry, and pay them to occasionally come by and fly a survey flight.

• 

Is it possible to bring the analysis of drone data into the forestry curriculum?

SA: Our students will definitely need to know more about sequencing and planning data capturing campaigns, analysing the data and interpreting the results rather than actually flying a drone. They should of course not be deprived of the fun part of flying drones, but certainly it is the interpretation that is the crux. The forestry education at Stellenbosch does already include a good component of the analysis of remotely sensed data, and other departments offer courses on digital photogrammetry, image analysis, geomatics etc., so it is really more a question of deciding to enhance these elements in the curriculum.

What role will the Department of Forest and Wood Science be playing with regard to ‘Drones in Forestry’ in the next 5 or so years?

SA: There is obviously an important role to play in ensuring that our graduates are equipped to embrace the data coming from drone acquisitions, just as they are for other data sources today. This process can be short circuited somewhat through promoting postgraduate projects using drone data in all aspects of forestry already today, and we are doing that. The role of the department will hopefully be one of provisioning industry research by supplying young ‘experts’ into the value chain. We hope that in five years time, our graduates will be just as at home with using drones and drone data, as they are with a dbh calliper today!

*First published in SA Forestry Annual, 2021