Weighing trees with lasers: reducing uncertainty in tropical forest biomass and allometry

Measuring the volume and structure of a tree accurately allows us to calculate the total above-ground carbon (C) stored in the tree, a very important property. Trees remove CO2 from the atmosphere during photosynthesis and can store this C for decades or even centuries until the tree dies, when some of it is released back to the atmosphere through decomposition. Tropical forests store around half of all above-ground terrestrial C, but are at particular risk due to deforestation and degradation, as well as from changing rainfall and temperature patterns. Surprisingly, our knowledge of tropical forest C stocks is quite poor, and errors in these stocks are large and uncertain. This uncertainty feeds into estimates of CO2 emissions due to deforestation, degradation and land use change. We will address this major uncertainty in the terrestrial C cycle by deploying a new, NERC-funded terrestrial laser scanner (TLS) to scan 1000s of trees in tropical forests on three continents: Amazonia, the Congo Basin and SE Asia. The laser data will allow us to measure 3D tree volume and biomass non-destructively to within a few percent of the best current estimates, made by destructive harvesting and weighing. The current, large uncertainties arise because weighing a tree is extremely difficult: tropical trees may be over 50m tall, and weigh 100 tonnes or more. Harvesting also precludes revisiting trees over time to measure change. In practice, a small sample of trees that have been harvested and weighed are related to easy-to-measure parameters of diameter and height, using empirical 'allometric' (size-to-mass) relationships. These relationships are then used to translate diameter and height measurements made over wider areas into estimates of biomass. Allometry is also the only way to infer biomass at very large (pan-tropical) scales, from remote sensing measurements. Unfortunately, the sample of harvested trees underpinning global allometric relationships is geographically limited, and contains very few large trees. Current estimates of tropical forest C stocks from satellite and ground data, all based on these very limited allometry samples, diverge significantly in size and pattern, leading to heated debate as to why this should be.

We hope to settle this debate, given that our lidar-derived estimates of biomass are completely independent of allometry and unbiased in terms of tree size. We will 'weigh' more trees than are currently included in all global pan-tropical allometries and quantify uncertainty in the allometry models. We will also test assumptions made in allometric models regarding tree shape and wood density. Our measurements will also answer fundamental questions about geographical differences in structural characteristics across tropical forests. Our data will be vital for testing new estimates of biomass from remote sensing; the UK-led ESA BIOMASS RADAR and NASA GEDI laser missions will both estimate pan-tropical C stocks by relying on allometric relationships between forest height and biomass. Our work will feed into these two missions through long-standing collaborations with the lead scientists. More generally, the large number of tree measurements we will collect would be of great interest to researchers in tropical ecology, forestry, biodiversity, remote sensing and C mapping, among others.

A key aim of the project is to ensure the widest use of our results, by making our data and tools publicly available. We will work with partners to explore routes for commercial developments and input into government policy, particularly relating to forest management and C mapping and mitigation. Lastly, we will make our work accessible through a range of outreach activities, including developing links between a school in the Amazon and UK schools, to raise awareness of scientific, conservation and policy issues surrounding tropical forests.

We target four groups of users with interests in C stocks, biomass, tree structure and tropical forests more generally to maximise the impact of our work: users of data and tools; commercial sector; policymakers; wider public.

Users of data and tools includes forestry sector, conservation, land management, other researchers. We will make TLS data, derived tree structure and analysis tools publicly available. We will integrate the TLS-derived tree measurements into the database coordinated by co-I S. Lewis (www.forestplots.net), including data query and analysis tools. Timescale: 1 year for database integration. TLS data will be archived at NEODC/CEMS and also provided to the TLS International Interest Group (TLSIIG) via PP Strahler, for community use. All code used to process the TLS data will be put on github to enable other researchers to process and analyse them. Timescale: project duration, ongoing.

Policymakers: we will address issues of direct policy relevance: C cycle uncertainty, forest inventory and management policy. These issues are core to government response to climate change, particularly UN Reducing Emissions from Deforestation and Degradation (REDD+). This impact will be via PPs Herold, SAC, Forest Research, and ongoing collaborations with government agencies e.g. S. Lewis in Gabon (calibrating and validating national estimates of forest C stocks from airborne lidar for conservation and REDD+ readiness). We will work with PP Forest Research to test and improve existing UK forest inventory methods. Timescale: 1-2 years to show benefit of TLS methods; ongoing if demonstrated. Our data will add value to UK investment in the forthcoming ESA BIOMASS mission, through calibration and testing of retrievals (and via PP Quegan). Results will also feed into calibration and development for the NASA GEDI lidar mission (PP Dubayah). Timescale: 1-2 years for testing BIOMASS retrievals; 5+ years, for REDD+ positioning, and inclusion in BIOMASS, GEDI.

Commercial: we identify various opportunities for commercial exploitation of C stock mapping technology. Our allometry-free AGB estimates will enable PP Carbomap to assess their new airborne lidar-derived C stock estimates. Timescale: 1 year to provide 3D tree models; ongoing for testing new technology and retrievals. PP Planet Labs will work with us to test the feasibility of area-based C stock measurements using high spatial and temporal resolution satellite data: Timescale: 6 months. PP SAC will identify routes to policy and commercial data exploitation opportunities, including data hosting via the CEMS facility. Timescale: project duration. PP Skycap will explore UAV-based measurements of canopy height and cover, and how new low-cost lidar instruments might be used in this context. Timescale: project duration for development and then ongoing for commercial exploitation. PPs Herold and Schaaf will provide low-cost lidar instruments that we will test for biomass estimation. Such instruments may find much wider use in C stock mapping, eg in REDD+ activities, and we will exploit existing links with manufacturers and suppliers to develop these opportunities. Timescale: 1 year for demonstration; ongoing for exploitation.

Wider public: The project will provide a unique view of tropical forests that is particularly suited to engaging wider interest, due to its inherently visual nature. We will build on established links between a school in the Brazilian Amazon and UK schools to develop reciprocal work programmes and teaching materials to highlight environment, climate and biodiversity issues. PP Burkitt will develop web-based 3D visualisation of the TLS data. The PI and co-Is will expand existing outreach activities, such as the PI blog and involvement in TV (BBC, Reuters) and online teaching (ESA online) activities. Timescale: 1 year to set up school partnerships then self-sustaining; 1 year for visualisation; ongoing for other activities.