Periodic Reporting for period 3 - FODEX (Tropical Forest Degradation Experiment)
Reporting period: 2021-01-01 to 2022-06-30
Tropical forests are greatly important stores of carbon and biodiversity, covering a huge land area (about 15% of the planet's land surface) but nonetheless having an outsized impact on the Earth System, regulating our climate and supporting livelihoods and populations everywhere. Given their importance and area, we know surprisingly little about how they are changing. We know, in broad terms, that people are deforesting an ever increasing proportion of tropical forests, and have evidence to suggest that they are damaging (degrading) a much larger area through selective logging, small-scale agriculture, and the extraction of resources, but we do not have a good idea of where these disturbances are happening or at what intensity. This is a tragedy, as it prevents the targeting of policies and resources to stop deforestation and forest degradation.
Further, we know that the tropical forest areas that have been left largely alone, or that are recovering from past disturbance, are doing a really important job in stabilising the climate. Over half the carbon we release into the atmosphere by burning fossil fuels and cutting down forests does not remain in the atmosphere, but is in fact taken up immediately by the Earth system. This 'missing sink' has helped us out greatly during the recent past, meaning that climate change has occurred much more slowly than it would have done if all that carbon had been allowed to remain in the atmosphere, warming the planet through the greenhouse effect. We do not know well where this 'missing sink' is located, but we believe it is a combination of capture in the oceans, and capture in the world's forests. Being able to map this sink is crucial, as it is really important for humanity to know where it is so we can strengthen protection on those forests that are protecting us most, and predict how this sink will change with time.
Over the past decades the NASA, the European Space Agency (ESA), and the European Union, among others, have invested in collecting more and more satellite data. These data are increasingly sophisticated in terms of their spatial resolution (the size of pixel sizes on the ground), the range of data being collected, and their temporal frequency (once we only 'saw' the planet every month or two with most higher resolution sensors, now it's often more than once a week). And, critically, the data are increasingly released in a free and open way, meaning scientists and companies can take advantage of hundreds of thousands or even millions of scenes, with only computing costs to worry about as data volumes increase. Combined with the big data revolution and massive developments in artificial intelligence and machine learning, this should mean that by now we have good systems of using these satellite data to monitor the forests of the tropics.
However, this revolution in satellite monitoring of forests has not really happened. Data on monitoring deforestation have definitely improved. But data on forest degradation, where only some trees are removed from an area, and regrowth, are still very poor. We believe that a major reason for this has been a lack of good 'ground truth' data on forest change. While there are scientific field plots where every tree is measured, enabling the calibration of static forest carbon maps, these plots are typically in protected areas and are very rarely disturbed or cleared. This makes it hard to calibrate products that are directly trying to measure this change.
Collecting suitable ground data to develop forest degradation and regrowth data is therefore the aim of this project, FODEX. We have set up permanent forest plots in two logging concessions, in contrasting forest types, one in Peru, and one in Gabon. In both of these we set up our plots and measured all the trees in 2019, including collecting laser scanner data from the ground so we had exact 3D models of every tree, and collecting LiDAR data from our drone so we could map millions of trees in the surrounding area. Then the plots were logged at different intensities. We have remeasured these plots in Gabon, and are still waiting to remeasure the Peru plots (our field campaign due in May 2020 has been delayed due to COVID). Ultimately, we will remeasure the plots again twice during their regrowth from disturbance, so each plot will have been measured four times, with substantial changes in biomass at each point.
We hope to use these data to develop and test new algorithms for mapping biomass change across the tropics. We will test many different methods, using different combinations of satellites, and come up with the best methods. We will put all our methods, code, data and output maps on our website, and work hard in the final year of the project to ensure they are well used by policy makers and NGOs, as well as other scientists. Ultimately, if successful, our project could jump start the conservation of this region through enabling project targeting and monitoring, and embarrassing some countries/regions while highlighting those that are doing well. It should also produce datasets that will improve our understanding as scientists of the global carbon cycle, helping us predict better what will happen in the future.
Further, we know that the tropical forest areas that have been left largely alone, or that are recovering from past disturbance, are doing a really important job in stabilising the climate. Over half the carbon we release into the atmosphere by burning fossil fuels and cutting down forests does not remain in the atmosphere, but is in fact taken up immediately by the Earth system. This 'missing sink' has helped us out greatly during the recent past, meaning that climate change has occurred much more slowly than it would have done if all that carbon had been allowed to remain in the atmosphere, warming the planet through the greenhouse effect. We do not know well where this 'missing sink' is located, but we believe it is a combination of capture in the oceans, and capture in the world's forests. Being able to map this sink is crucial, as it is really important for humanity to know where it is so we can strengthen protection on those forests that are protecting us most, and predict how this sink will change with time.
Over the past decades the NASA, the European Space Agency (ESA), and the European Union, among others, have invested in collecting more and more satellite data. These data are increasingly sophisticated in terms of their spatial resolution (the size of pixel sizes on the ground), the range of data being collected, and their temporal frequency (once we only 'saw' the planet every month or two with most higher resolution sensors, now it's often more than once a week). And, critically, the data are increasingly released in a free and open way, meaning scientists and companies can take advantage of hundreds of thousands or even millions of scenes, with only computing costs to worry about as data volumes increase. Combined with the big data revolution and massive developments in artificial intelligence and machine learning, this should mean that by now we have good systems of using these satellite data to monitor the forests of the tropics.
However, this revolution in satellite monitoring of forests has not really happened. Data on monitoring deforestation have definitely improved. But data on forest degradation, where only some trees are removed from an area, and regrowth, are still very poor. We believe that a major reason for this has been a lack of good 'ground truth' data on forest change. While there are scientific field plots where every tree is measured, enabling the calibration of static forest carbon maps, these plots are typically in protected areas and are very rarely disturbed or cleared. This makes it hard to calibrate products that are directly trying to measure this change.
Collecting suitable ground data to develop forest degradation and regrowth data is therefore the aim of this project, FODEX. We have set up permanent forest plots in two logging concessions, in contrasting forest types, one in Peru, and one in Gabon. In both of these we set up our plots and measured all the trees in 2019, including collecting laser scanner data from the ground so we had exact 3D models of every tree, and collecting LiDAR data from our drone so we could map millions of trees in the surrounding area. Then the plots were logged at different intensities. We have remeasured these plots in Gabon, and are still waiting to remeasure the Peru plots (our field campaign due in May 2020 has been delayed due to COVID). Ultimately, we will remeasure the plots again twice during their regrowth from disturbance, so each plot will have been measured four times, with substantial changes in biomass at each point.
We hope to use these data to develop and test new algorithms for mapping biomass change across the tropics. We will test many different methods, using different combinations of satellites, and come up with the best methods. We will put all our methods, code, data and output maps on our website, and work hard in the final year of the project to ensure they are well used by policy makers and NGOs, as well as other scientists. Ultimately, if successful, our project could jump start the conservation of this region through enabling project targeting and monitoring, and embarrassing some countries/regions while highlighting those that are doing well. It should also produce datasets that will improve our understanding as scientists of the global carbon cycle, helping us predict better what will happen in the future.
We have completed three field campaigns so far - two in Gabon (pre- and post logging), and one in Peru (pre-logging). Logging has happened in Peru, but thanks to the COVID shut down we have not been able to revisit. Nonetheless, we have lots of useful data to be working on - tens of thousands of tree diameter measurements, hundreds of terrestrial laser scanner scans (including for the first time globally, before and after logging scans of the same plots), and terabytes of LiDAR and image data from our drones. All staff are currently working on their own field or satellite datasets, with two papers near submission (led by Dr Iain McNicol based on tree recognition in LiDAR vs optical drone data, and Chiara Aquino on the ability of optical satellite sensor to map forest change at different resolutions, from 30 cm pixels up to 250 m). Further results are expected to flow from these initial campaigns, about degradation, over the coming 6-12 months, and then the focus will shift towards scaling up from our field sites to the whole of the tropics, and to looking at regrowth following this disturbance.
We have also had some significant communications successes. Posts on Twitter about the project have been viewed thousands of times, raising awareness about forest degradation and satellite/space data. We have organised meetings in Gabon and Peru with relevant scientists, policy makers and forestry officials, who are aware of and approve of what we are doing, and are looking forward to seeing concrete results. We have more formal workshops planned for next year, so we can do more knowledge communication, but in all our current field campaigns we have had local PhD students with us in the field who have been invaluable in helping collect data, but will also have gained knowledge and experience.
We have also had some significant communications successes. Posts on Twitter about the project have been viewed thousands of times, raising awareness about forest degradation and satellite/space data. We have organised meetings in Gabon and Peru with relevant scientists, policy makers and forestry officials, who are aware of and approve of what we are doing, and are looking forward to seeing concrete results. We have more formal workshops planned for next year, so we can do more knowledge communication, but in all our current field campaigns we have had local PhD students with us in the field who have been invaluable in helping collect data, but will also have gained knowledge and experience.
We have collected entirely novel data at this point: we are the first group to ever collect Terrestrial Laser Scanner and LiDAR data over forest before and after logging. While we have not yet had the chance to fully exploit these datasets, we are in the progress of making discoveries, and fully anticipate making big discoveries and advances in forest monitoring with these data as we proceed with the project.