LASmoons: Elia Palop-Navarro

Elia Palop-Navarro (recipient of three LASmoons)
Research Unit in Biodiversity (UO-PA-CSIC)
University of Oviedo, SPAIN.

Background:
Old-growth forests play an important role in biodiversity conservation. However, long history of human transformation of the landscape has led to the existence of few such forests nowadays. Its structure, characterized by multiple tree species and ages, old trees and abundant deadwood, is particularly sensible to management practices (Paillet et al. 2015) and requires long time to recover from disturbance (Burrascano et al. 2013). Within protected areas we would expect higher proportions of old-growth forests since these areas are in principle managed to ensure conservation of natural ecosystems and processes. Nevertheless, most protected areas in the EU sustained use and exploitation in the past, or even still do.

lasmoons_elia_palopnavarro_0

Part of the study area. Dotted area corresponds to forest surface under protection.

Goal:
Through the application of a model developed in the study area, using public LiDAR and forest inventory data (Palop-Navarro et al. 2016), we’d like to know how much of the forest in a network of mountain protected areas retains structural attributes compatible with old-growth forests. The LiDAR processing tasks which LAStools will be used for involve a total of 614,808 plots in which we have to derive height metrics, such as mean or median canopy height and its variability.

Vegetation profile colored by height in a LiDAR sample of the study area.

Vegetation profile colored by height in a LiDAR sample of the study area.

Data:
+ Public LiDAR data that can be downloaded here with mean pulse density 0.5 points per square meter. This data has up to 5 returns and is already classified into ground, low, mid or high vegetation, building, noise or overlapped.
+ The area covers forested areas within protected areas in Cantabrian Mountains, occupying 1,207 km2.

LAStools processing:
1) quality checking of the data as described in several videos and blog posts [lasinfo, lasvalidate, lasoverlap, lasgrid, las2dem]
2) use existing ground classification (if quality suffices) to normalize the elevations of to heights above ground using tile-based processing with on-the-fly buffers of 50 meters to avoid edge artifacts [lasheight]
3) compute height-based forestry metrics (e.g. ‘-avg’, ‘-std’, and ‘-p 50’) for each plot in the study area [lascanopy]

References:
Burrascano, S., Keeton, W.S., Sabatini, F.M., Blasi, C. 2013. Commonality and variability in the structural attributes of moist temperate old-growth forests: a global review. Forest Ecology and Management 291:458-479.
Paillet, Y., Pernot, C., Boulanger, V., Debaive, N., Fuhr, M., Gilg, O., Gosselin, F. 2015. Quantifying the recovery of old-growth attributes in forest reserves: A first reference for France. Forest Ecology and Management 346:51-64.
Palop-Navarro, E., Bañuelos, M.J., Quevedo, M. 2016. Combinando datos lidar e inventario forestal para identificar estados avanzados de desarrollo en bosques caducifolios. Ecosistemas 25(3):35-42.

LASmoons: Alen Berta

Alen Berta (recipient of three LASmoons)
Department of Terrestrial Ecosystems and Landscape, Faculty of Forestry
University of Zagreb and Oikon Ltd Institute for Applied Ecology, CROATIA

Background:
After becoming the EU member state, Croatia is obliged to fulfill the obligation risen from the Kyoto protocol: National Inventory Report (NIR) of the Green House Gasses according to UNFCCC. One of the most important things during the creation of the NIR is to know how many forested areas there are and their wood stock and increment. This is needed to calculate the size of the existing carbon pool and its potential for sequestration. Since in Croatia, according to legislative, it is not mandatory to calculate the wood stock and yield of the degraded forest areas (shrubbery and thickets) during the creation of the usual forest management plans, this data is missing. So far, only a rough approximation of the wood stock and increment is used during the creation of NIR. However, these areas are expanding every year due to depopulation of the rural areas and the cessation of traditional farming.

very diverse stand structure of degraded forest areas (shrubbery and thickets)

Goal:
This study will focus on two things: (1) Developing regression models for biomass volume estimation in continental shrubberies and thickets based on airborne LiDAR data. To correlate LiDAR data with biomass volume, over 70 field plots with a radius of 12 meters have been established in more than 550 ha of the hilly and lowland shrubberies in Central Croatia and all trees and shrubberies above 1 cm Diameter at Breast Height (DBH) were recorded with information about tree species, DBH and height. Precise locations of the field plots are measured with survey GNNS and biomass is calculated with parameters from literature. For regression modeling, various statistics from the point clouds matching the field plots will be used (i.e. height percentiles, standard deviation, skewness, kurtosis, …). 2) Testing the developed models for different laser pulse densities to find out if there is a significant deviation from results if the LiDAR point cloud is thinner. This will be helpful for planning of the later scanning for the change detection (increment or degradation).

Data:
+
641 square km of discrete returns LiDAR data around the City of Zagreb, the capitol of Croatia (but since it is highly populated area, only the outskirts of the area will be used)
+ raw geo-referenced LAS files with up to 3 returns and an average last return point density of 1 pts/m².

LAStools processing:
1)
extract area of interest [lasclip or las2las]
2) create differently dense versions (for goal no. 2) [lasthin]
3) remove isolated noise points [lasnoise]
4) classify point clouds into ground and non-ground [lasground]
5) create a Digital Terrain Model (DTM) [las2dem]
6) compute height of points above the ground [lasheight]
7) classify point clouds into vegetation and other [lasclassify]
8) normalize height of the vegetation points [lasheight]
9) extract the areas of the field plots [lasclip]
10) compute various metrics for each plot [lascanopy]
11) convert LAZ to TXT for regression modeling in R [las2txt]

Generating Spike-Free Digital Surface Models from LiDAR

A Digital Surface Model (DSM) represents the elevation of the landscape including all vegetation and man-made objects. An easy way to generate a DSM raster from LiDAR is to use the highest elevation value from all points falling into each grid cell. However, this “binning” approach only works when then the resolution of the LiDAR is higher than the resolution of the raster. Only then sufficiently many LiDAR points fall into each raster cell to prevent “empty pixels” and “data pits” from forming. For example, given LiDAR with an average pulse spacing of 0.5 meters one can easily generate a 2.5 meter DSM raster with simple “binning”. But to generate a 0.5 meter DSM raster we need to use an “interpolation” method.

Returns of four fightlines on two trees.

Laser pulses and discrete returns of four fightlines.

For the past twenty or so years, GIS textbooks and LiDAR tutorials have recommened to use only the first returns to construct the interpolating surface for DSM generation. The intuition is that the first return is the highest return for an airborne survey where the laser beams come (more or less) from above. Hence, an interpolating surface of all first returns is constructed – usually based on a 2D Delaunay triangulation – and the resulting Triangular Irregular Network (TIN) is rasterized onto a grid at a user-specified resolution to create the DSM raster. The same way a Canopy Height Model (CHM) is generated except that elevations are height-normalized either before or after the rasterization step. However, using a first-return interpolation for DSM/CHM generation has two critical drawbacks:

(1) Using only first returns means not all LiDAR information is used and some detail is missing. This is particularly the case for off-nadir scan angles in traditional airborne surveys. It becomes more pronounced with new scanning systems such as UAV or hand-held LiDAR where laser beams no longer come “from above”. Furthermore, in the event of clouds or high noise the first returns are often removed and the remaining returns are not renumbered. Hence, any laser shot whose first return reflects from a cloud or a bird does not contribute its highest landscape hit to the DSM or CHM.

(2) Using all first returns practically guarantees the formation of needle-shaped triangles in vegetated areas and along building roofs that appear as spikes in the TIN. This is because at off-nadir scan angles first returns are often generated far below other first returns as shown in the illustration above. The resulting spikes turn into “data pits” in the corresponding raster that not only look ugly but impact the utility of the DSM or CHM in subsequent analysis, for example, in forestry applications when attempting to extract individual trees.

In the following we present results and command-line examples for the new “spike-free” algorithm by (Khosravipour et. al, 2015, 2016) that is implemented (as a slow prototype) in the current LAStools release. This completely novel method for DSM generation triangulates all relevant LiDAR returns using Contrained Delaunay algorithm. This constructs a “spike-free” TIN that is in turn rasterized into “pit-free” DSM or CHM. This work is both a generalization and an improvement of our previous result of pit-free CHM generation.

We now compare our “spike-free” DSM to a “first-return” DSM on the two small urban data sets “france.laz” and “zurich.laz” distributed with LAStools. Using lasinfo with options ‘-last_only’ and ‘-cd’ we determine that the average pulse spacing is around 0.33 meter for “france.laz” and 0.15 meter for “zurich.laz”. We decide to create a hillshaded 0.25 meter DSM for “france.laz” and a 0.15 meter DSM for “zurich.laz” with the command-lines shown below.

las2dem -i ..\data\france.laz ^
        -keep_first ^
        -step 0.25 ^
        -hillshade ^
        -o france_fr.png
las2dem -i ..\data\france.laz ^
        -spike_free 0.9 ^
        -step 0.25 ^
        -hillshade ^
        -o france_sf.png
las2dem -i ..\data\zurich.laz ^
           -keep_first ^
           -step 0.15 ^
           -hillshade ^
           -o zurich_fr.png
las2dem -i ..\data\zurich.laz ^
        -spike_free 0.5 ^
        -step 0.15 ^
        -hillshade ^
        -o zurich_sf.png

The differences between a first-return DSM and a spike-free DSM are most drastic along building roofs and in vegetated areas. To inspect in more detail the differences between a first-return and our spike-free TIN we use lasview that allows to iteratively visualize the construction process of a spike-free TIN.

lasview -i ..\data\france.laz -spike_free 0.9

Pressing <f> and <t> constructs the first-return TIN. Pressing <SHIFT> + <t> destroys the first-return TIN. Pressing <SHIFT> + <y> constructs the spike-free TIN. Pressing <y> once destroys the spike-free TIN. Pressing <y> many times iteratively constructs the spike-free TIN.

One crucial piece of information is still missing. What value should you use as the freeze constraint of the spike-free algorithm that we set to 0.9 for “france.laz” and to 0.5 for “zurich.laz” as the argument to the command-line option ‘-spike_free’. The optimal value is related to the expected edge-length and we found the 99th percentile of a histogram of edge lengths of the last-return TIN to be useful. Or simpler … try a value that is about three times the average pulse spacing.

References:
Khosravipour, A., Skidmore, A.K., Isenburg, M. and Wang, T.J. (2015) Development of an algorithm to generate pit-free Digital Surface Models from LiDAR, Proceedings of SilviLaser 2015, pp. 247-249, September 2015.
Khosravipour, A., Skidmore, A.K., Isenburg, M (2016) Generating spike-free Digital Surface Models using raw LiDAR point clouds: a new approach for forestry applications, (journal manuscript under review).

LASmoons: Andreas Konring and Susanne Bjerg Petersen

Andreas Konring and Susanne Bjerg Petersen (recipients of three LASmoons)
Department of Environmental Engineering
Technical University of Denmark, Lyngby, DENMARK

Background:
Copenhagen has in the recent years experienced severe floodings due to cloudbursts which has increased the focus of climate adaption and the implementation of green infrastructure. The use of sustainable urban drainage system (SUDS) solutions to divert stormwater from the existing drainage system will be a central measure to increase the climate resilience while greenifying the city and Copenhagen municipality is investing 700 million euros in SUDS projects alone. Additionally, the city has decided to plant 100.000 new trees in the next 10 years as another measure to enhance natural amenities but also because of air cleansing and cooling effects. However, it has not been investigated what effect the current canopy cover has on the rainwater retention due to increased evaporation and soil infiltration and if planting more trees could help improve the pluvial flooding issues.

Example of a pit-free CHM in an urban environment.

Example of a pit-free CHM in an urban environment.

Goal:
This study aims to estimate the current number of trees and extract tree metrics such as volume, canopy cover and densities with the use of the national LIDAR dataset and NIR ortophotos from summer and spring. These canopy metrics will be used to inform a simple tree model which will be implemented in a 2-D overland flow model to assess the effect of trees on flood mitigation. The created CHM could also be used in further analysis of the urban heat island effect.

Data:
+
100 square kilometers of the Danish national LiDAR dataset collected in November 2014 covering the municipality of Copenhagen.
+  density of 4 – 5 last-returns per square meter
+  classified into surface (1), ground (2), vegetation (3,4,5), buildings (6), noise (7) and water (9).

LAStools processing:
1)
create square tiles with buffer to avoid edge artifacts [lastile]
2) generate DTMs and DSMs with only buildings and terrain [las2dem]
3).normalize height, remove outliers and keep classes 2, 5 and 6 [lasheight]
4) create rasters with forest metrics [lascanopy]
5) calculate the pit-free Canopy Height Model (CHM) proposed by Khosravipour et al. (2014) [lasthin, las2dem, lasgrid]

Reference:
Copenhagen Municipality, 2011. Copenhagen Climate Adaption Plan.
Geodatastyrelsen, 2014. Danmarks højdemodel, DHM/Punktsky – Dataversion 2.0 januar 2015. Product specification.
Khosravipour, A., Skidmore, A.K., Isenburg, M., Wang, T.J., Hussin, Y.A., 2014. Generating pit-free Canopy Height Models from Airborne LiDAR. PE&RS = Photogrammetric Engineering and Remote Sensing 80, 863-872.

Rapidlasso receives “Green Asia Award” at ACRS 2015

PRESS RELEASE (for immediate release)
November 16, 2015
rapidlasso GmbH, Gilching, Germany

At the Asian Conference on Remote Sensing 2015 (ACRS 2015) held in Manila, rapidlasso GmbH was honored with the “Green Asia Award” by the Chinese Society of Photogrammetry and Remote Sensing (CSPRS). This award is given to a paper that directs Asia towards a greener future using remote sensing technology. This year’s award commends rapidlasso GmbH on advancing the area of LiDAR processing through their PulseWaves effort. PulseWaves is a vendor-neutral full waveform LiDAR data exchange format and API that simplifies access to full waveform data and allows researchers to focus on algorithms and share results. In the future this technology may prove valuable to improve biomass estimates for carbon credit programs such as the TREEMAPS project of WWF.

Prof. Kohei Cho and Prof. Peter T. Y. Shih present the award

Prof. Kohei Cho and Prof. Peter T. Y. Shih present the Green Asia Award

The society communicated to Dr. Martin Isenburg, CEO of rapidlasso GmbH, that this award was also meant to honor his many years of teaching and capacity building across the Asian region. Since the beginning of 2013 rapidlasso GmbH has conducted well over 50 seminars, training events, and hands-on workshops at universities, research institutes, and government agencies in Thailand, Malaysia, Myanmar, Vietnam, Indonesia, Singapore, Taiwan, Japan, and the Philippines. The on-going LiDAR teaching efforts of rapidlasso GmbH in Asia and elsewhere can be followed via their event page.

Green Asia Award for CEO of rapidlasso GmbH

Green Asia Award given to the CEO of rapidlasso GmbH

The award certificate that was presented to Dr. Martin Isenburg by Prof Kohei Cho and Prof Peter Shih during the closing ceremony of ACRS 2015 came with a cash reward of USD 300. The award money was donated to the ISPRS summer school that followed the ACRS conference to top off the pre-existing “green sponsorship” by rapidlasso GmbH that was already supporting a “green catering” of summer school lunches and dinners to avoid single-use cups, plastic cutlery and styrofoam containers. The additional award money was used for hosting the main summer school dinner at a sustainable family-run restaurant serving “happy chickens” and “happy pigs” raised organically on a local farm.

during the closing ceremony of ACRS 2015

Award Ceremony held during Closing of ACRS 2015

About rapidlasso GmbH:
Technology powerhouse 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.

Two ASPRS awards for “pit-free” CHM algorithm

PRESS RELEASE (for immediate release)
July 29, 2015
rapidlasso GmbH, Gilching, Germany

The paper “Generating Pit-free Canopy Height Models from Airborne LiDAR” co-authored by rapidlasso GmbH and published in the September 2014 issue of PE&RS (the journal of the ASPRS) was awarded twice at the IGTF 2015 – ASPRS Annual Conference in Tampa, Florida last May. The paper took home the John I. Davidson President’s Award for Practical Papers (2nd Place) as well as the Talbert Abrams Award (2nd Honorable Mention).

The John I. Davidson President’s Award for Practical Papers (2nd Place).

The “pit-free” CHM paper wins the John I. Davidson President’s Award for Practical Papers (2nd Place) and the Talbert Abrams Award (Second Honorable Mention).

The “pit-free” CHM paper is joint work with Anahita Khosravipour, Andrew K. Skidmore, Tiejun Wang, and Yousif A. Hussin of ITC and University of Twente. It describes a technique that can create raster Canopy Height Models (CHMs) without the so called “pits” that tend to hamper subsequent extraction of individual tree attributes such as number, location, height, and crown diameter. The paper uses data measured in the field by ITC researchers to show that “pit-free” CHMs significantly lower the commission and omission errors in single tree detection.

Side-by-side comparison of a "standard" CHM and a "pit-free" CHM.

Visual side-by-side comparison of a “standard” versus a “pit-free” CHM.

The “pit-free” CHM algorithm can easily be implemented with LAStools either by modifying an available batch script or by executing the LAStools Pipelines distributed with the toolboxes for ArcGIS and QGIS. A detailed blog article that compares various different methods for creating CHMs is available via the Web pages of rapidlasso GmbH.

We at rapidlasso GmbH like to especially congratulate the main author, Ms. Anahita Khosravipour, who managed to get two awards with her very first academic publication. Those who like our “pit-free” CHM algorithm will probably also love the new technique that our team will introduce later this year at SilviLaser 2015 in France.

About rapidlasso GmbH:
Technology powerhouse 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.

Rasterizing Perfect Canopy Height Models from LiDAR

In literature you sometimes read “we generated a Canopy Height Models (CHM) and then did this and that” without the process that was used to create the CHM being described in detail. One approach computes the CHM as a difference between DSM and DTM: create a DTM from the ground returns and a DSM from the first returns and subtract the two rasters. Also here is still a question left to be answered: how exactly are the DTM and the DSM generated. A different approach computes the CHM directly from height-normalized LiDAR points. And again there are many ways of doing so and we want to look at the possibilities in more detail.

the 100 by 100 meter sample plot 'drawno.laz'

the 100 by 100 meter sample plot ‘drawno.laz’

In the following we demonstrate different alternatives for CHM generation on a 100 by 100 meter sample LiDAR tile ‘drawno.laz‘ from a forest near Drawno in Poland (that you can download here), slowly converging towards the CHM generation method that we recommend using. We start with ground classifying the LiDAR using lasground:

lasground -i drawno.laz ^
          -wilderness ^
          -o ground.laz

Then we height-normalize the LiDAR using lasheight. As we know that there are no trees higher than 28 meters in this plot we drop all LiDAR points that are higher than 30 meters that may be bird hits or other noise.

lasheight -i ground.laz ^
          -drop_above 30 ^
          -replace_z ^
          -o normalized.laz

As the sample plot has an average pulse spacing of around 0.3 meters we decide to use a step size of 0.33333 meters to create a 300 by 300 pixel raster. We produce a false color visualization instead of a height raster for our CHMs so we can include the results here. The simplest method uses lasgrid with option ‘-highest’ that uses for each pixel the highest z coordinate among all LiDAR returns falling into the corresponding 0.33333 by 0.33333 meter area.

lasgrid -i normalized.laz ^
        -step 0.33333 ^
        -highest ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_grd.png
Gridding the highest point that falls into each 0.33333 meter by 0.33333 meter cell.

Gridding the highest point that falls into each 0.33333 meter by 0.33333 meter cell.

The resulting CHM (shown at a 200 % zoom) is full of empty pixels and so called “pits” that will hamper subsequent analysis for single tree detection, height and crown diameter computation, and the like. A simple improvement can be obtained by replacing each LiDAR return with a small disk. After all, the laser beam has – depending on the flying height – a diameter of 10 to 50 centimeter and approximating this with a single point of zero area seems overly conservative. In lasgrid there is the option ‘-subcircle 0.1’, which replaces each return by a circle with a radius of 10 centimeter or a diameter of 20 centimeter.

lasgrid -i normalized.laz ^
        -subcircle 0.1 ^
        -step 0.33333 ^
        -highest ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_grd_d20.png
Gridding the highest z value after turning each point into a circle with 20 cm diameter.

Gridding the highest z value after turning each point into a circle with 20 cm diameter.

The resulting CHM (shown again at a 200 % zoom) is much improved but there are still empty pixels and “pits”. We could simply widen the circles further with ‘-subcircle 0.15’, ‘-subcircle 0.2’, or ‘-subcircle 0.25’. As you can see below this produces increasingly smooth CHMs with widening tree crowns. But this “splats” the LiDAR returns into circles that are growing larger and larger than the laser beam diameter and thus have less and less in common with reality.

Gridding the highest returns will often leave empty pixels in the data even when “splatting” the points. Another popular approach avoids this by interpolating all first returns with a triangulated irregular network (TIN) and then rasterizing it onto a grid to create the CHM. This can be implemented with las2dem as shown below:

las2dem -i normalized.laz ^
        -first_only ^
        -step 0.33333 ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_tin.png
Rasterizing the TIN that interpolates all first returns onto a  0.33333 meter grid.

Rasterizing the TIN that interpolates all first returns onto a 0.33333 meter grid.

The result has no more empty pixels but is full of pits because many laser pulses manage to deeply penetrate the canopy before producing the first return. When combining multiple flight lines some laser pulses may have an unobstructed view of the ground under the canopy without hitting any branches. These “pits” and how to avoid them is discussed in great length in the September 2014 edition of the ASPRS PE&RS journal by a paper of Khosravipour et al. We build upon these ideas in the following. But first we combine the ‘-highest’ gridding with TIN interpolation with a two step approach: (1) keep only one highest return per grid cell with lasthin (2) interpolate all these highest returns with las2dem.

lasthin -i normalized.laz ^
        -step 0.33333 ^
        -highest ^
        -o temp.laz
las2dem -i temp.laz ^
        -step 0.33333 ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_tin_his.png
Rasterizing the TIN that interpolates only the highest points falling into each  0.33333 meter by 0.33333 meter grid cell.

Rasterizing the TIN that interpolates only the highest points falling into each 0.33333 meter by 0.33333 meter grid cell.

Next we integrate the idea of “splatting” the points into circles with a diameter of 20 centimeter to account for the laser beam diameter by adding option ‘-subcircle 0.1’ to lasthin.

lasthin -i normalized.laz ^
        -subcircle 0.1 ^
        -step 0.33333 ^
        -highest ^
        -o temp.laz
las2dem -i temp.laz ^
        -step 0.33333 ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_tin_his_d20.png
Rasterizing the TIN that interpolates only the highest points of a 0.33333 meter grid  after first splatting points into circles with 20 cm in diameter.

Rasterizing the TIN that interpolates only the highest points of a 0.33333 meter grid after first splatting points into circles with 20 cm in diameter.

The results are much nicer but there are still pits. However, one may argue at this points that thinning with a step size of 0.33333 is too agressive and removes too many points for subsequent interpolation and rasterization with the exact same step size. So we double the resolution of the temporary point cloud by thinning with half the step size of 0.16667.

lasthin -i normalized.laz ^
        -subcircle 0.1 ^
        -step 0.16667 ^
        -highest ^
        -o temp.laz
las2dem -i temp.laz ^
        -step 0.33333 ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_tin_hhs_d20.png
Rasterizing the TIN that interpolates only the highest points of a 0.16667 meter grid  after first splatting points into circles with 20 cm in diameter onto a raster with step size 0.33333.

Rasterizing the TIN that interpolates only the highest points of a 0.16667 meter grid after first splatting points into circles with 20 cm in diameter onto a raster with step size 0.33333.

We now have more detail but also many more pits. Furthermore the interpolation of highest returns makes a big error across areas were we do not have any LiDAR returns that are flanked by canopy returns on both sides. This happens in the area circled where wrong canopy is created because the TIN is interpolating canopy returns across a small wet area without any returns, We show now how the pit-free method of Khosravipour et al. can be used to generate the perfect CHM:

rmdir tmp_chm /s /q
mkdir tmp_chm
las2dem -i normalized.laz ^
        -drop_z_above 0.1 ^
        -step 0.33333 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_ground.bil
lasthin -i normalized.laz ^
        -subcircle 0.1 ^
        -step 0.16667 ^
        -highest ^
        -o temp.laz
las2dem -i temp.laz ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_00.bil
las2dem -i temp.laz ^
        -drop_z_below 2 ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_02.bil
las2dem -i temp.laz ^
        -drop_z_below 5 ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_05.bil
las2dem -i temp.laz ^
        -drop_z_below 10 ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_10.bil
las2dem -i temp.laz ^
        -drop_z_below 15 ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_15.bil
las2dem -i temp.laz ^
        -drop_z_below 20 ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_20.bil
las2dem -i temp.laz ^
        -drop_z_below 25 ^
        -step 0.33333 -kill 1.0 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o tmp_chm/chm_25.bil
lasgrid -i tmp_chm/chm*.bil -merged ^
        -step 0.33333 ^
        -highest ^
        -false -set_min_max 0 25 ^
        -ll 278200 602200 -ncols 300 -nrows 300 ^
        -o chm_pit_free_d20.png
rmdir tmp_chm /s /q
Running the pit-free algorithm on the highest LiDAR returns in a 0.16667 meter grid (after splatting them to circles 20 cm in diameter) and producing a 0.33333 meter raster CHM.

Running the pit-free algorithm on the highest LiDAR returns in a 0.16667 meter grid (after splatting them to circles 20 cm in diameter) and producing a 0.33333 meter raster CHM.

With such a perfectly pit-free output we can now be even more conservative and lower the diameter of the circles that we replace each LiDAR return with from 20 cm to 10 cm by replacing ‘-subcircle 0.1’ with ‘-subcircle 0.05’ in the above script.

Running the pit-free algorithm on the highest LiDAR returns in a 0.16667 meter grid (after splatting them to circles only 10 cm in diameter) and producing a 0.33333 meter raster CHM.

Running the pit-free algorithm on the highest LiDAR returns in a 0.16667 meter grid (after splatting them to circles only 10 cm in diameter) and producing a 0.33333 meter raster CHM.

Compared to the original pit-free algorithm published by Khosravipour et al. there are two minor differences: (1) instead of using all first returns as input we use one highest returns per grid cell after splatting all returns as small circles (instead of area-less points) onto a grid that has twice the output resolution. (2) we also have a ‘-kill 1.0’ threshold to also generate a partial CHM from all points for ‘chm_00.bil’ and add a new ‘chm_ground.bil’ to fill the potential holes. This prevents that higher up canopy returns are wrongly connected across water bodies where there are no LiDAR returns at all.

Reference:
Khosravipour, A., Skidmore, A.K., Isenburg, M., Wang, T.J., Hussin, Y.A., 2014. Generating pit-free Canopy Height Models from Airborne LiDAR. PE&RS = Photogrammetric Engineering and Remote Sensing 80, 863-872.