Post provided by Kim Calders, Glenn Newnham, Andrew Burt, Pasi Raumonen, Martin Herold, Darius Culvenor, Valerio Avitabile, Mathias Disney, and John Armston
To celebrate the 10th Anniversary of the launch of Methods in Ecology and Evolution, we are highlighting an article from each volume to feature in the Methods.blog. For Volume 6, we have selected ‘Nondestructive estimates of above-ground biomass using terrestrial laser scanning‘ by Calders et al. (2014).
In this post, the authors discuss the background and key concepts of the article, and changes in the field that have happened since the paper was published.
Terrestrial laser scanning (TLS) calculates 3D locations by measuring the speed of light between a transmitted laser pulse and its return. Firing hundreds of thousands of pulses per second, these instruments can represent the surroundings in detailed 3D, displaying them as virtual environments made up of high density points. The main applications of commercial instruments in the early 2000s were engineering or mining, but their application in natural forested environments was in its infancy. Forest ecosystems are structurally complex; clear reference points used to register multiple scans are rare and trees move due to wind creating artefacts in the data.
Researchers from the Commonwealth Scientific and Industrial Research Organisation (CSIRO) played a pioneering role in the development and uptake of terrestrial laser scanning in forest ecology and designed a unique instrument specifically targeting forest measurements (Jupp et al., 2009). They built the first full hemispherical full waveform laser scanner, naming it the Echidna Validation Instrument (EVI), after the Australian mammal the echidna because the hemispherical pattern of laser shots resembled the spines on the animal’s back. However, the EVI was both heavy and power-demanding, making it impractical to use in remote or challenging forest ecosystems. The EVI, and subsequent research instruments such as DWEL (Boston University) and SALCA (University of Salford) fulfilled important roles as research tools. They furthered the science of scanning natural environments, but were never put into commercial production.
Parallel to instrument development, initial methods processing terrestrial laser scanning data were focused on relatively simple structural metrics such as diameter, height, vertical profiles of forest density and analysis of gap fraction to understand light dynamics within the canopy. Reconstruction of full 3D architecture from point clouds only started to emerge towards the end of the first decade of 2000. At the same time, efficiently quantifying aboveground biomass (AGB) at plot scale became more important as a means of validating broad scale AGB mapping which underpinned the United Nations initiative on Reducing Emissions from Deforestation and Forest Degradation (UN-REDD) – a programme striving to reduce the destruction of forests and preserve the uptake of carbon by trees.
Around 2010, one of us (lead author Kim Calders) was a new PhD student that had access to a new high-precision TLS instrument from the Austrian company RIEGL and planned a number of research visits to Australia to build on the knowledge of the CSIRO EVI team. At that point, we quickly gathered new knowledge on how to scan larger areas more efficiently using the new equipment – however, we were still lacking robust algorithms for 3D modelling to calculate tree volumes and thus estimate biomass. More importantly, we realised that validation of AGB estimates using this new instrument was key for the success of programs like UN-REDD where efficient plot based data collection was critical for calibrating national to global AGB datasets.
At that time we were fortunate that destructive harvesting of trees was planned through the Comprehensive Carbon Assessment Program and Simon Murphy made sure that we could tag along to scan trees before harvesting. We collected the pre-harvest laser scanning data in 2012 in Rushworth Forest in south eastern Australia. Following a meeting in the summer of 2013 at University College London, we generated the first Quantitative Structural Model (QSM – i.e. the volumetric enclosure of the point cloud) from the data. It would take another year before the manuscript was finally submitted to Methods in Ecology and Evolution.
Nondestructive estimates through terrestrial laser scanning
Our paper “Nondestructive estimates of above-ground biomass using terrestrial laser scanning” brought together an international team of researchers to develop a way to harness laser technology for estimating and validating AGB of forests. This paper – which brings together ecologists, remote sensing scientists and mathematicians – is a great example of how international and interdisciplinary collaboration can be a catalyst for significant scientific progress in ecology and forestry.
As a result of the work, we were able to convert single tree point clouds to QSMs to estimate their volume and further converted to AGB using basic density values collected during harvesting. Our method allowed us to measure biomass with far more certainty than before. While traditional methods using allometric models yielded results that may have been off by as much as 37%, the LiDAR method developed in this paper delivers over 90% accuracy. This was the first true validation of this new alternative to destructive sampling and manual weighing of tree components to determine their biomass. A crucial discovery was that the error for estimating AGB using TLS was relatively constant with tree size, whereas we saw a significantly larger error in large trees when traditional allometric models were used. This is significant as large trees can hold a significant proportion of the total biomass within a forest, and are the most difficult to destructively harvest.
Five years later
Since the publication of this paper, our approach has been tested and validated in forests that are critical stores of worldwide carbon, including tropical forests in Asia, Africa and South America (De Tanago et al., 2017 and Takoudjou et al., 2017 – both published in MEE). There has been rapid growth of interest in the development of TLS for measuring tree and forest structure more generally (e.g. Disney, 2018 and Calders et al., 2020). Our MEE paper has now been cited close to 300 times, and we have extended our work to show how TLS can be used to measure large tropical trees, redwoods (Disney et al., in review), deciduous woodlands (Calders et al., 2018) and urban forests (Wilkes et al., 2018). A key part of our work since has been developing open tools to allow improved exploitation of TLS data, not only for woody structure but also information on leaves (see Burt et al., 2019; Boni Vicari et al., 2019; and Krishna Moorthy et al., 2020).
These developments in turn have led to other groups taking up these methods, increased international collaboration and a general rapid advancement of the use of TLS. This has been assisted by the continued development of affordable and commercially available laser scanners. Terrestrial laser scanning is now endorsed as a valid method for measuring AGB at a plot scale (typically up to 1 hectare) in the most recent Intergovernmental Panel on Climate Change (IPCC) national greenhouse gas reporting guidelines. Similarly, these methods are deemed increasingly important for the calibration and validation of broad-scale AGB mapping as an input to carbon cycle modelling. The emerging CEOS-AGB protocol (Committee for Earth Observing Satellites Land Product Validation Protocol for Aboveground Biomass) will include a chapter on the importance of terrestrial laser scanning to generate these reference measurements.
We envisage that with continued algorithm development – terrestrial laser scanning will become the benchmark for low cost ecological assessment. The broad scale adoption of these methods will allow us to answer challenging questions about how our environment is changing over time based on time-series of these data. The gradual expansion of these long term detailed ecological datasets will improve our knowledge of vegetation growth dynamics, ecosystem succession and disturbances and the impact humans have on our valuable natural environments.
A tribute to Mikko Kaasalainen
Sadly, 2020 has also marked the passing of one of our co-authors, Mikko Kaasalainen, Professor of Mathematics at Tampere University. Mikko was a true pioneer in 3D modelling of point clouds, and the methods he developed still form the foundation of our analysis today. Working in many different fields, his legacy goes far beyond his contributions to the lidar community and his research paved the way for new innovations in the years to come. His passing is a great loss to the community and Mikko will truly be missed – not only as a colleague and mentor, but also as a friend.
To find out more about terrestrial laser scanning, read the Methods in Ecology and Evolution article, Nondestructive estimates of above-ground biomass using terrestrial laser scanning
Find out our editors’ favourite papers on remote sensing in our
Volume 6 Editor’s Choice
Find out about the Methods in Ecology and Evolution articles selected to celebrate Volumes 1-5: