Drones in Locust Control — putting theory into action
The desert locust (Schistocerca gregaria) has recently become a big problem in Kenya and the East Africa region in general. According to FAO, recent weather patterns in East Africa have created conditions that favour rapid locust reproduction. At present, there are at least 17 counties in Kenya (out of a total of 47) with confirmed desert locust infestations. An agriculture ministry survey showed Samburu, Turkana, Laikipia, Isiolo, Baringo and Elgeyo Marakwet were the hardest-hit by the first round of invasion earlier this year. Left unchecked, the numbers of crop-devouring insects there could grow over 500 times by the end of 2020. Such swarms — potentially containing hundreds of millions of individual desert locusts that can move 150 kilometres a day, are devastating rural livelihoods in their relentless drive to eat and reproduce. A desert locust devours its own weight in food per day which is about two grams.
A desert locust devours its own weight in food per day
With the very real and apparent threat to our food security, Astral Aerial, and CABI (Center for Agriculture and Biosciences International) teamed up to take on the desert locust invasion in Kenya head-on, or rather, on the head. Current estimates suggest that Kenya has procured over 216,000 litres of pesticides, with additional investments for aerial operations with aircraft and ground teams. We were of the view that drones could complement the current spraying means, as they are very targeted and can treat infestations that might otherwise be difficult to reach with vehicle-mounted or hand spraying methods. Drones could also target small swarms that may not be feasible to control by aircraft, populations that escape the aerial sprays, and roosting locusts in inhabited areas or agricultural lands.
The mission was clear cut! Astral Aerial was armed with two DJI Agras T16 crop-spraying drones, for targeted overhead spraying, CABI with the Green Muscle biopesticide that’s tough on locusts yet gentle on the environment, the elocust3m locust tracking platform application of the United Nations Food and Agriculture Organization (FAO), and a very committed friendly local scouting team that kept its ears to the ground, tracking locust movements deep into the night and practically watching them fly and determine which village they would land. The objective was simple. We needed to test best practices on using this new technology to spray desert locusts. However, as a complementary solution to current methods, we needed to establish a scientific basis for the approach, provide tactical assistance in combating the voracious locusts while keeping the environment safe, and get out there and test!
Armed with the tools and the goodwill, we were ready to cross enemy lines. Or were we?
First of all, it’s important to note that locusts favor roosting sites located AT THE END OF THE WORLD! The drive to the roosting sites was equally grueling as it was exciting, testing the limits of our 4x4 vehicles. Needless to say, we can only applaud the intricate tracking and communication structure that the local communities and county agricultural officers have developed to track locust movements in the county. The local communities report locust sightings to county officials spread out across the region. The officials then relay the information to the county commissioner of special programs (CC) who co-ordinates with response teams (us).
“No love lost for locusts at first sight”
Our first mission was more of a reconnaissance mission, where we wanted to find out modes of co-ordination, verify scouting and locust tracking effectiveness, assess drone spraying performance, and collect spray test samples of sprayed insects. We drove to Maralal - an area in Northwest Kenya - as advised by the county officer as it was a strategic location for easier tracking and reaching of roosting swarms.
As soon as he gave us the news, we were ready for him to point at the swarm location, and we would execute the attack. Simple, right? Wrong! The problem was, he pointed 12 km away from our “strategic” waiting site!
Challenges:
This was our first bump on the road so we did what anyone in our position would do. We simply got over it, looked for a place to spend the night and decided to wake up bright and early, ready to chase down the swarms. The following day, the swarms had moved to another area called Tepele area in Wamba North, (Samburu North), a 50 km distance from our location. This distance seems manageable on a tarmac road, but we are talking about a less than thrilling drive on a murram road, where if it rained we would need to cancel our trip. But, we followed the swarms there. We have to give it to the local community on tracking these swarms because, in retrospect, they are definitely not easy to find.
When we got there, we found that they were in an inaccessible valley that we couldn’t access by driving or walking. We deployed a survey drone to assess the roosting site, and strategized on how to carry out the spraying exercise. The swarm consisted of mostly immature locusts (pink in color), a few mature ones, and some were even mating. Once planned out, we launched the sprayer drone for spot spraying over the swarm and after that, waited for the pesticide to take effect. After some time, locust activity slowed down noticeably. Unfortunately, we could not assess the pesticide kill rate as locust movement was rapid, while the pesticide itself was slow acting.
We called it a day and decided to try again the following day, but this time spraying would start earlier while the temperature was still low to ensure the locusts’ inactivity. In general, as the ground warms up during the mid-morning, the locusts collect some heat that allows them to start their flight.
On the following day, at 0800 hrs, we arrived at Ndoyo Nasipa area, where the locusts had roosted the previous night. The roosting site was inaccessible by road, so we had to walk 5 km to get to them! Sounds easy, walking, right? But we had to carry all the equipment including the drone, batteries, and the pesticide! At this point we realized two things; the locusts were not going to make this easy for us, and we should all just go ahead and cancel any planned fitness activities we had for the duration of the project because they were clearly not needed anymore - although carrying the equipment for the distance could count for a good calories burner.
We arrived at the site and were all set to begin spraying, but could not. Apparently, an aircraft was scheduled to carry out spraying at the time on the same swarm. We had to wait until 1049 hrs to begin spraying and continued until 1200 hrs. Basking locusts were disturbed by the aircraft flight and started milling, ready to fly. Luckily, the temperature was too low for them to fly. The aircraft sprayer had covered the central swarm area. We targeted the edge of the swarms to supplement the aircraft coverage. We also targeted tree tops, shrubs and ground swarms at close proximity for increased effectiveness. We then waited for 2 hours to estimate mortality. Going forward we need to have clear co-ordination and open communication lines between response teams to avoid duplicating efforts.
With the reconnaissance run over we were ready to head back to Nairobi, having established our contact points among the local locust tracking teams, understood locust roosting behavior at a deeper level, proven our drone’s performance capabilities for the project, and, finally, ready to sign up for the next iron man triathlon because this mission was a military bootcamp in disguise.
Learnings:
Encountering the swarms for the first time was an eye-opener to the sheer scale of the problem.
- Coordinating with ground teams was vital in making these operations a success.
- Identifying flying sites and pushing the pilots’ capabilities to match the extreme terrain and difficult takeoff and landing locations proved to be of great importance and impacts the spraying time and coverage significantly.
- It’s practically impossible, and of no significant impact to spray once the locusts take-off and move from the roosting site.
“Once bitten……”
We headed to Marigat, Samburu county on 1st October to make another attempt on the control operation.
Objectives:
- Conducting operational experiments to find optimal spraying parameters such as flying height, spray rate and flight modes.
The specific swarm location was inaccessible because of security issues, so we had to spend the night in Nakuru and await further direction from our local scouts.
The following day, we were advised to follow the swarm to Rumuruti by the local community scouts, where we arrived at 1030 hrs. We found an immature swarm that covered roughly 400 acres over the flat topography with loamy soils. The swarm was about to fly out so we waited to follow it to the next roosting site. Spraying flying swarms is not only ineffective but also unsafe, as the hundreds of locusts can get jammed up in the drone’s fast-moving rotary blades, which may cause the drone to malfunction during flight.
The swarm moved 10 km away from a village called Survey to Nagum. It took us a long time to get there owing to poor road conditions. When we arrived, the swarms were moving to the next site. Nevertheless, we tried spraying the remaining swarm which now covered 100 ha. We covered 1.63 ha at a spray height of 12m and line spacing of 10m over an average flight time of 5.8 minutes. After the flight we collected locust samples for impact measurement and called it a day.
Next, we went to Kiwanja Ndege in Churo sublocation, where the locusts were roosting atop acacia trees and on the ground covering around 800 acres. We then mixed our pesticide solution and continued to spray a total of 4.96 ha, over a combined flight time of 55.45 minutes and 14 flights with different flight modes (M+, Manual, and Hybrid). We experimented with different spray heights and spray rates. After waiting out for the pesticide to take effect, we found that we had achieved a lower-than-expected kill rate. A bit disheartening, I admit, but it was evidently clear that we needed to work on the optimizations more. We packed up everything and headed back to Nairobi the following day with the goal of reviewing our results and performance to see what we could learn and modify on the next trip.
Learnings:
The heights above the target were determined by the vegetation, in this case the locusts were roosting on the top canopy of the tall trees.
- We found that flying 2.5m above the canopy ensured the tops of the trees were covered properly.
- The down draft from the props (the drone’s rotating blades) allowed for the pesticide to penetrate the canopy and reach the locusts that were roosting below.
- Controlling the spray drift was easier when flying lower to the target area and ensured there was less wastage of the liquid.
”A lot wiser on the third try”
The next trip felt like it was all perfectly scripted in comparison to what we had been through during the previous trips. The site in Maralal was easy to access, the topography flat and the wind speed low enough to enable efficient spray and spread of the pesticide.
We deployed a survey drone to identify the most suitable spraying site and settled on a sweet spot with an area of 4 ha. After calibrating the drone to the desired spray parameters, we launched the spray drone and carried out the spraying at a rate of 250 ml per minute over 4 spray nozzles which equates to spraying one liter a minute.
We continued to spray at different speeds, spray rates, and heights above the swarm to determine optimum effectiveness by measuring the swath width which was found to be 6 m, with 3 m on each side of the drone. We then collected the sprayed locusts using nets to determine the kill rate after a few hours.
We called it a day and awaited any new locust sighting updates for the following day.
We did not receive any news of locust sightings the next day, and headed back to Nairobi the day after. This goes to show how hard it can be to work with locusts, as their behavior is so unpredictable.
Challenges:
- Samburu County had the coordinates of only one mature locust swarm. The swarm was lost after the first day of spraying.
- The swarm was in an area that is known for insecurity therefore we could not access it.
Learnings:
Although we thought we were at the center of the swarm, it turned out that we were not. After packing up, we were overflown by what was clearly the epicenter of that swarm. Therefore, what we sprayed was only the outer edge. It was clear we had to do a better job of scouting the area and identifying where the majority of the swarm was. This is something we implemented in the subsequent trip.
“The Forth time’s a charm”
On our last mission, the locust swarm control mission led us to Baragoi. Baragoi is currently facing issues with insecurity owing to cattle rustling practices. The night before we arrived, a special branch police officer had been killed during an attack. Luckily, we had a police unit guarding our hotel at night. We had never had such an experience before and therefore this particular project tested our limits in more ways than we had anticipated.
The following morning we set out. We did a 1.5 hr drive to the swarm location. The closest we could get by driving was at a riverbed that was 2 KM from the roosting site. As per the usual protocol, our feet had to take us where the 4 x 4 couldn’t. Up a mountain! The swarms were so inaccessible that the local communities used smoke signals to indicate the location of the roosting sites. We scaled up the 2 KM hike with our drones, pesticide, generators, and extra equipment on hand. Our legs were shaking when we got there, but that did not matter because mentally, we were in military mode. Jocko Willink would approve.
The locusts were hoppers (younger locusts that hop and cannot fly yet) covering 1 ha /swarm in multiple swarms (groups) spread out across the area. There was no clear take-off or landing site and we had to clear shrubs for a makeshift take-off and landing site. With everything ready, we began sample collection, carried out the spraying exercise, and finally collected post control samples. This was arguably the easier part of the exercise. We had to complete the exercise quickly as it was about to rain and we had a 2km walk back without food or water. Luckily it started raining after we had already left.
On the second day, we had a 400 m hike to the roosting site, which was a walk in the park for us. We carried out the spraying after sample collection, then headed back to Nairobi.
Challenges:
- The hopper bands were 2km up a steep mountain, and the team had a tough time scaling the mountain. This involved carrying all the equipment; five 5kg batteries, 18.5kg drone, chemicals, and 60 liters of water.
- There was no clear take-off area, we had to clear bushes and shrubs for take-off and the pilot had to maneuver through tight spaces to make it above the canopy.
Learnings:
- Swath width experiments (done with Ms. Violet Ochieng, an Msc. Agricultural entomology student from University of Nairobi) were conducted at different heights. Although rudimentary, it clearly showed that lower altitudes to the target were more effective in terms of penetration and drift avoidance.
- We had the most impact spraying hoppers as they don’t move a lot and are aggregated in multiple small groups across large areas. In reality, we found that even though some of the hoppers are in thick bush and rocky areas, flying the drone near the target increased penetration and hence the consequent kill rate on the hoppers.
- This was a project of many variables, from different terrain to different locust ages and geo-political conditions. We had to keep an open mind and be ready for anything.
- For this operational environment, our optimum flight mode was assisted manual mode, where we set some flight parameters like spray height, drone speed, and line spacing while maintaining drone motion and shifting to the next flight line. This allowed us to focus on covering the target swarms as well as accurately test our SOP’s.
- E locust 3m was a helpful tool in indicating the swarm locations. The information was however incomplete in a few instances where the data was not updated in time due to lack of network coverage. Going forward it’s a good idea to use radio communication among the local scouts as it is not affected by poor coverage.
- Not all the scouts have smartphones to download the eLocust3m, also eLocust 3G is less common and only a few desert locusts officers have them.
- There is a coordinated system of information relay. Scouts also have the contacts of desert locust officers that enable them to disseminate information through calling or sending short messages on the presence of desert locusts.
- The locals are trained on how to identify the desert locust, how to report the information, and who to report to.
Recommendations
- More tests should be done to validate the optimum parameters such as height, swath width, track spacing, speed of the drone, and flow rate of the chemical.
- These tests will give the standard operating parameters that improve the efficiency of the spray drone.
- The tested parameters will be used to spray the chemical over the desert locust to achieve the best kill rate without contaminating untargeted areas and organisms.
- There is need to test the efficiency of several drones at a time
In conclusion, this was a successful first sprint that showed us overall that there is a need for targeted control of locust swarms. With the right spray height, flight line spacing, flight modes, and flight speed, the drone was able to achieve a swath width of 6m per drone. With the green muscle biopesticide, we require a deeper understanding of the concentrations required, within safe environmental limits, as well as a deeper penetration (which we believe can be achieved by implementing ULV technology across the drones’ spray nozzles). A drone fleet would be effective in covering large areas.
Going forward, we shall also need to develop SOPs (Standard Operating Procedures) among the different response teams (UAVs, aircraft, and ground spraying) to complement efforts and develop a synchronous working style that augments each parties’ efforts. The locust menace requires a team effort, and no one method is sufficient. A recent survey in Kenya shows that affected counties had an average of 77% of its crops severely damaged by locust swarms, with an overall 70% of the free-range grazing lands affected. We can make a difference to reduce these numbers with UAV technology.
We’re making serious headway with the control of the desert locust and will not stop until every last locust is eliminated in East Africa. We’ll also lose a few KGs while at it 😉.
