The Re-Forest-ER Consortium in Action
Space4Good and our partners, coming forward as the Re-Forest-ER consortium, have been actively engaged over the past weeks in the converging worlds of technology and forestry (rain-forestry to be specific :). We have had the privilege of travelling to the modern and green city of Singapore, where the XPRIZE Rainforest semi-finals were held at the Central Catchment Nature Reserve. It was an honor and pleasure to share time and space with Serge, a renowned ecologist from Liverpool John Moores University, the talented local drone pilots part of Halo Robotics (female majority) out of Indonesia, our pioneering partners NatureMetrics deploying their new environmental DNA sampling solution, and of course, our long trusted and esteemed Masarang Foundation / Arsari group with which we continue to expand and mature our journey upon regenerative agriculture/forestry.
Figure 1. Re-Forest-ER Team
Semi-Finals Round in Singapore
Thirteen teams were part of the Semi-Finals round. The teams were all competing for $ 10 million total in prize money available throughout the entire competition, for continued research and development around this project. Each team was assigned a plot in the Central Catchment Reserve which they had to survey and inspect using automated technologies and identify the maximum number of species in this plot.
9 Exhilarating Competition Days
Arrival (1st-2nd June): After our team landed in the Asian continent, we organized a hike the next day to have a dynamic meet-and-greet with the wider team and do a first human eye scan of the competition area.
Calibration Days (3-4th June): We then had the opportunity to test the equipment, (our competition gadgets), and show the panel of judges and evaluators we could fly drones as certified pilots. We proved the safety of newly developed technology, allowing us to use it in Singapore.
On this field day one, we found out we were assigned our field testing plot, which contained a water reserve, crucial for NatureMetrics solution deployment, as it involved collecting an eDNA sample from a water source. So far so good.
D day-24 hour data acquisition period (5-6th June): 8 drones check, extra batteries check. We had a successful setup for launch but faced a long morning’s wait as we patiently stood by for flight clearance by the Singapore Civil Aviation Authority
We've been waiting for this day for almost six months, and it finally came to reality. We named this day D-day, as you shall see, the events that unravelled thereafter were quite hectic but also quite characteristic of the surprises that fieldwork or “showtime” presents.
On the Space4Good side of things, we had to manoeuvre around both airspace and lightning warnings, we had to modify the altitudes we wanted to fly our drones in, and had some issues with overheating drones to name a few. Nonetheless, during the 24h time period, we successfully completed our goal of collecting multispectral, thermal and LiDAR data. We also deployed the mini drone with an AudioMoth to record birds, and we had an awesome M30T drone, only released last year, a high-tech and off-the-shelf solution for taking good-quality pictures of trees. As you will see next, we had issues reaching the back of the area, but at least we could survey part of the plot area.
We were all eager to test out one of the A-game solutions within the Re-Forest-ER team, the novel eDNA sampling by NatureMetrics also known as SPLASH, which had to be deployed in a water body. After several trials, we soon realized that getting to the needed water source could not be done in a safe manner given the competition conditions. The reservoir was almost a full kilometre down and as seen in Figure 2, the terrain profile was too hilly, presenting high elevation differences, causing a loss of both in signal and in line of sight of the drones. As such, we quickly had to abandon the idea of reaching the planned water source and pivot towards another solution.
Figure 2. Plot and elevation profile
Trial 2. We were enthusiastic to find a little stream running by our launch pad area. As such we engaged in meticulously getting a water sample from there with the Outreach Robotics/NatureMetrics solution. Moments later, long and behold, even with a documentary team on the watch, in the middle of the data collection, the NatureMetrics solution came down, freefalling 10 meters from thin air. Gasp. Luckily the sensor did not break, we still had the crucial part of the equipment live and running.
We <3 Tape. We had to come up with a solution to tie the eDNA sensor to the drone. Utilizing a mix of explorative science and innovation, our on-site hack resulted in a Macgyver-esque solution. We engineered some tape coils for collecting soil samples, but the wrong type of sticky tape meant the samples were in the end not used. Then we came up with a gauze hanging from the drone, which allowed us to successfully collect water samples from the nearby stream, and the Nature Metrics solution was in place to advance to the next steps.
Data Analysis (7-9th June): After moving our raw data to XPRIZE hard drives, we started 2 days of code-line crunching and tree species analysis.
Adopting a co-creative approach, we sent pictures of trees for the people in Indonesia to label, recorded as observations with taxonomy, coordinates and some other annotations. We deliberately did this, as we wanted the locals to help out on this process, adding to our tech contributions by sharing their on-the-ground knowledge. The trees were later visually identified by the botanists. Further, we had water samples processed by NatureMetrics in their portable lab overnight. We had audio where we identified birds and bats recordings. We had lidar and multispectral data to train our models, using the newly labelled data provided by the local experts, and subsequently run over the algorithm to make predictions for the entire survey area. Data was fully processed and a 100-page report was written.
Results and Outcomes
The results were turned in and showcased on June 9th at 7:00AM. Afterwards, we enjoyed a coffee moment with the wider team to reflect upon the enthralling experience we just lived!
After this unique opportunity, we are quite proud of the mechanisms in which our diverse team cooperated under this pressure cooker setting. Moreso, as has been evident in our other field experiences, incorporating local knowledge is one of the most evident value-adding aspects that can be highlighted from this process. It contributes to our consistent acknowledgement that without their expertise, conserving rainforests would be impossible.
What can we say about the rainforest we just scanned with a triple-stack of automated technologies? What we can say now is that the reserve where the competition was held is to be understood more as a secondary rainforest. This was immediately pointed out by the scientists in the team but also reflected in our data. Via our LiDAR point clouds, it was very evident that gaps in between the canopy were reflective of degraded areas, and experts pointed to the presence of invasive species. Nonetheless, secondary or not, there were still plenty of species to identify, and the patch of nature we just scanned serves its purpose as a forest to collect water and as a lung to the Singaporean city. More importantly, this opportunity represents a big leap for automated biodiversity detection advances!
A long holiday awaited at Tioman Island, in Malaysia, to enjoy a true primary rainforest.
But wait! What about the competition? What about the finals? Who won?
Results are up in July. We too are burning with curiosity. To be continued!
Would you like more information or are you interested in collaborating with Space4Good? Visit our website or contact us via hello@space4good.com.