Beautiful Full-Waveform LiDAR of a Tropical Rainforest

Scanning forests with LiDAR is done to predict timber yield, estimate biomass, and monitor ecosystems. A detailed Digital Terrain Model (DTM) is needed to accurately compute forestry metrics such as tree height. For generating a DTM it is important to have enough “ground returns” that measure the elevation of the bare-earth for which the laser pulse needs to penetrate through the vegetation down to the forest floor and reflect back to the plane. This can be difficult in a dense tropical forest with tall trees and many layers of canopy as the light of the laser pulse has many opportunities to be reflected or absorbed before reaching the ground. Often none of the light energy will reach the forest floor or the returning ground echo will be too weak to be registered back at the plane.

The "airborne sensors" of  Asian Aerospace Services is equpiied with a RIEGL Q680i airborne LiDAR scanner.

The “airborne sensors” aircraft of Asian Aerospace Services is equipped with a RIEGL Q680i airborne LiDAR scanner.

Asian Aerospace Services Ltd. and rapidlasso GmbH flew a RIEGL LMS Q680i LiDAR system out of a Diamond DA42 “Airborne Sensors” aircraft above dense tropical rainforest in Thailand. We did a short test flight to determine whether the LiDAR would be able to penetrate the canopy of Khao Yai National Forest. The original objective was to demonstrate that the scanner is able to “see through” the canopy in order to produce data for the TREEMAPS project of WWF Thailand funded with EUR 1,730,205 through the International Climate Initiative of the German government. Unfortunately Germany had to pull the funding for TREEMAPS after political tinkering started to interfere with technical objectives, and the LiDAR produced by our “pro-bono flight” has become the only tangible outcome produced by TREEMAPS. We are now making this data available under a Creative Commons Public Domain license (for download see below).

Above Khao Yao National Forest scanning for the pro-bono test flight to investigate the canopy penetration of our LiDAR.

Above Khao Yao National Forest during a test scan to investigate the canopy penetration of the RIEGL Q680i LiDAR scanner.

Low clouds disrupted our original plans but we had one cloud-free stretch during a ferry flight between target locations. We were scanning for 90 seconds at 1200 meters above ground with a speed of 220 km/h. The Q680i was set to operate at a laser pulse rate of 80 kHz with the polygon mirror rotating 10 times per second. The effective scan rate was 240 / 360 * 80,000 = 53,333 shots per second, as each of the 4 mirror facets covers 60 degrees – summing up to 240 degrees – and there are no laser shots leaving the aircraft for the remaining 120 degrees of between facets.

The 53,333 shots per second are reflected over the length of 10 * 4 = 40 mirror facets so that a single scan line contains a sequence of 53,333 / 40 = 1333 shots spaced about 12.4 microseconds apart. After each scan line comes a short pause of 8.3 milliseconds during which the mirror rotates 30 degrees to the next facet. Flying at 220 km/h and 1200 meters above ground means that each second we scanned 62 meters of a 1430 meters wide swath. This resulted in a pulse spacing of 62 / 40 = 1.55 meters between subsequent scan lines and a pulse spacing of 1430 / 1333 = 1.07 meters along each scan line. The 90 second scan resulted in a strip length of about 5.5 km.

The PulseWaves export options available in RIEGL's RiPROCESS.

The PulseWaves export options available in RIEGL’s RiPROCESS.

For each shot the outgoing as well as the returning waveform are digitized and stored to RIEGL’s proprietary SDF format. We used RIEGL’s RiPROCESS software (version 1.5.8) to export the 4,770,152 shots as full waveform LiDAR to PulseWaves and to extract a total of 7,956,587 returns as LASzip-compressed discrete LiDAR to the LAZ format.

The DSM, DTM, and CHM of the 5.5 km long strip scanned above Khao Yai National Forest.

The DSM, DTM, and CHM of the 5.5 km long strip scanned above Khao Yai National Forest.

The Q680i produced sufficiently many ground returns to compute a “plausible” DTM (using lasground and las2dem of LAStools) despite the dense 45 meter tropical canopy during leaf-on season and high humidity. Typical terrain features such as streams and slopes are clearly visible in the hillshading. The corresponding Canopy Height Model (CHM) uses a false coloring that maps vegetation heights of 0 meters to blue and heights of 45 meters to red. There is less vegetation (darker blue) in areas that appear to be in the steeper parts of the DTM.

Only 5 % of pulses emitted produced a discrete return on the ground … this is rather low. It means that the generated DTM will be of much coarser resolution than the number of pulses shot per square meter would typically suggest. Of the 250,473 points classified as ground only 30,931 are single returns where the laser had an unobstructed view of the bare earth. More than half of the ground returns come from laser shots that had already produced two or more vegetation returns and 7,665 of them had their light energy weakened by grazing four or more layers before hitting the forest floor. Scrutinizing the full waveform LiDAR might lead to more ground information to improve the accuracy of the DTM.

We are happy to provide both – the discrete return as well as the full-waveform LiDAR – under a Creative Commons Public Domain license to researchers world-wide. You can download the data here:

You can decompress the pair of compressed PulseWaves files PLZ + WVZ into an pair of uncompressed PulseWaves files PLS + WVS with pulsezip.exe – a recent addition to PulseTools. We would be happy to hear back about your experiences with the data. You could join the PulseWaves user forum at http://pulsewaves.org and share your insights with the group. Below a visualization of a small part of the full waveforms …

Full Waveform LiDAR from Khao Yai National forest.

Full Waveform LiDAR from Khao Yai National forest.

Rapidlasso Teams Up with Carbomap

PRESS RELEASE (for immediate release)
December 4, 2014
rapidlasso GmbH, Gilching, Germany

Just in time for ELMF 2014 in Amsterdam, rapidlasso GmbH and Carbomap Ltd. have teamed up to further the development of tools that better exploit full-waveform LiDAR for the forestry and carbon market. This partnership brings together many years of expertise in processing discrete and full-waveform LiDAR with a wealth of experience in applying this technology within forestry and biomass applications.

The full-waveform tools are built around the PulseWaves format, an open LiDAR format that is the full-waveform sibling of the venerable LAS format, reaffirming a joint committment to support the use of open data formats within the LiDAR industry. Software will be developed both as stand-alone tools as well as for use within the IDL framework.

The use of LiDAR in the forest and carbon industries is expanding rapidly. The new partnership between rapidlasso and Carbomap targets the development of solutions for this growing market. Together the two companies can offer a wider range of tools for vegetation analysis that better exploit the additional information captured by a modern full-waveform scanner, including the newest multispectral instruments.

Antoine Cottin (left), CTO of Carbomap, and Martin Isenburg (right), CEO of rapidlasso, discuss technologie details about their new partnership.

Antoine Cottin (left), CTO of Carbomap, and Martin Isenburg (right), CEO of rapidlasso, discuss technology details about their new partnership.

About rapidlasso GmbH:
Technology start-up rapidlasso GmbH specializes in efficient LiDAR processing tools that are widely known for their high productivity. They combine robust algorithms with efficient I/O and clever memory management to achieve high throughput for data sets containing billions of points. The company’s flagship product – the LAStools software suite – has deep market penetration and is heavily used in industry, government agencies, research labs, and educational institutions. Visit http://rapidlasso.com for more information.

About Carbomap Ltd:
Carbomap is an environmental survey company spun out of the University of Edinburgh. The company takes forward over five years of world-class research in the development of a Multispectral Canopy LiDAR, a revolutionary, patent-pending laser scanning instrument designed to fill a gap in airborne forest survey requirements. The founders are international renown for remote sensing methodologies, satellite radar mapping, forest structure mapping, carbon sequestration and airborne survey. Carbomap is currently the only company with tools for analysing multispectral LiDAR for forest applications. Visit http://carbomap.com for more information.

Worldwide LiDAR of Rainforest Biomass for REDD+

We at rapidlasso have long been big fans of the biomass and biodiversity work done by Greg Asner’s group and their Carnegie Airborne Observatory. They were, in fact, the earliest “power-users” of the BLAST extension of LAStools and helped finding all the bugs when rasterizing billions of LiDAR points collected during large-scale surveys in the Amazon into Digital Elevation Models (DEMs) with blast2dem. Below a video fly-through of some of the LiDAR they collected.

A few days ago, Greg Asner together with his colleagues Joseph Mascaro, Stuart Davies, Alex Dehgan, and Sassan Saatchi published a thought-provoking article called “These are the days of lasers in the jungle” which is essentially a “call for action” to map the world’s tropical forests with a fleet of airplanes outfitted with advanced LiDAR to rapidly and to accurately assess global forest carbon stocks.

Why would anyone want to do this? In order to properly quantify actual emissions savings for REDD+ (Reduced Emissions from Deforestation and Degradation). REDD+ is a tropical forest carbon accounting program of the United Nations Framework Convention on Climate Change that aims to compensate tropical countries for reducing carbon emissions from deforestation and forest degradation that account for roughly 10 percent of global greenhouse gas emissions. The key to implement such a program is the ability to accurately and affordably estimate the actual emissions savings and a worldwide LiDAR inventory of tropical forests will accomplish just that argues the paper. This “call for action” has since been picked up by Mongabay – a site that examines emerging trends in climate, technology, and finance on conservation and development.

Interesting is the price tag that they estimate, which is a fraction of the cost of a typical Earth observation satellite mission, They claim: “Our ambitious plan can be accomplished for far less than what we have already spent on avoided deforestation. Aircraft leasing, data collection and processing costs for 30 days of flying can reasonably be limited to USD 500,000 Using this monthly sampling unit, collecting at an average of 100,000 hectares per day, a fleet of ten aircraft could do the job in four years at USD 250 million, or just 5% of pledged REDD + funding.”

The authors state “The time has come for a brute-force effort to directly assess the carbon stock for all of the world’s tropical forests by 2020” because “airborne LiDAR is uniquely suited for this role because it can be collected, standardized, reported and verified in a simple manner by both a landholder and any third party”. Should such a campaign turn out to be a viable option to implement REDD+ we hope full waveform LiDAR – not just discrete returns – will be collected.

Waveform digitizers have become popular because they can capture the reflection of the emitted laser pulse with much more detail than a discrete return system. The intensity of the signal returning to the plane is digitized up to one billion times per second, giving a vertical resolution of one digitized amplitude each 15 centimeters. As this can capture the interactions between each laser pulse and the vegetation with much greater detail, it would seem that having full waveform instead of discrete returns will prove especially useful for biomass studies. A future blog post will talk about our own experiences of scanning tropical rainforest to produce full waveform LiDAR in PulseWaves format.

Another approach with a similar objective is being taken by NASA with their future GEDI space LiDAR. The Global Ecosystem Dynamics Investigation (GEDI) instrument will be the first to systematically probe the depths of the forests from space to reveal their 3D structure, as depicted in the artist’s concept below, and provide crucial information about the impact that trees have on the amount of carbon in the atmosphere. “One of the most poorly quantified components of the carbon cycle is the net balance between forest disturbance and regrowth”, said Ralph Dubayah, the GEDI principal investigator at the University of Maryland. “GEDI will help scientists fill in this missing piece by revealing the vertical structure of the forest, which is information we really can’t get with sufficient accuracy any other way”. The instrument will be built at NASA’s Goddard Space Flight Center in Greenbelt, Maryland. Not sure about the price tag.

nasa_gedi

Image Credit: NASA’s Goddard Space Flight Center

Until either of these projects become reality, “enjoy” this video by Carnegie Airborne Observatory of a 3D flyover that shows rapidly expanding palm oil plantations in the Peruvian Amazon rainforest that are contributing to deforestation.

LASmoons: Anu Swatantran

Anu Swatantran (recipient of three LASmoons)
Department of Geographical Sciences
University of Maryland, College Park, USA

Background:
Single photon LiDAR (SPL) is a promising new technology for measuring terrain and vegetation characteristics over large spatial scales. SPL is different from conventional LiDAR instruments because it operates in the visible wavelength (532 nm) and requires only one detected photon per ranging measurement. This allows for rapid data collection over large areas with high point densities but also includes background solar noise within and above canopy. There is a lot of interest in using this technology for vegetation monitoring but its accuracies and overall potential remain largely unexplored.

Noise-free SPL data draped with multispectral imagery (NAIP). The point cloud shows individual trees.

Noise-free SPL data draped with multispectral imagery (NAIP). The point cloud shows individual trees.

Goal:
The goal of this project is to evaluate an experimental SPL dataset for mapping canopy structure and terrain over an entire county of Maryland, USA. Attributes from SPL will be compared with discrete return LiDAR and other datasets to assess their accuracies.
This study will improve our understanding of SPL and its capabilities for mapping forest structure and biomass over large areas. It will also inform future space borne missions [ICESAT-2] that are using this technology.

Data:
+ Single photon LiDAR was collected over Garrett County in Maryland using the moderate-altitude high-resolution quantum lidar system (HRQLS).
+ A total area of 1700 sq. km. was covered in 12 hours of flight time with 50% overlap between adjacent flight lines.
+ Point densities range from 8 – 13 /sq. m. and area higher in areas of overlap.

LAStools processing:
1)
apply filters for removing background noise using batch scripting [lasnoise]
2) classify the noise filtered data into ground and canopy [lasground, lasheight]
3) generate a high resolution county wide DTM [las2dem, blast2dem]
4) generate a (pit-free?) Canopy Height Model (CHM) [las2dem, lasgrid]
5) generate metrics at different resolutions from the normalized point cloud [lascanopy].