[GRASS-user] v.rast.stats and Lidar data

GRASS GRASS user
1

Dear mail list

I am using Grass 6.2.2 on Ubuntu 7.10

I have some LiDAR data of a forested area and I am trying to calculate tree heights / canopy heights based first and second return points (pre-filtered ground and non-ground). I have two questions that I hope some of you may be able to help me with. Firstly I have windowed / clipped my data to a small 'test' region with about 150 000 non-ground and 100 000 ground points.

1. I have interpolated the ground points using idw in grass and have taken that as my ground surface. I would now like to assign an extra column to my non-ground vector file and populate it with the raster values of the DSM to calculate actual point height above ground (points are currently in meters above sea-level). Traditionally, one would just create a digital surface model and a canopy model and subtract the two to get tree canopy height. I would like to do the subtraction in the database and then interpolate the result. I am trying to do this with v.rast.stats and it is taking a really long time. How can I speed this up? Reading the help page I also see that the vector data are rasterised and values are assigend based on categories. I dont want categories, I am just interested in the raster value directly below each point? Is v.rast.stats the correct application to use?

2. While the above mentioned data set is fairly small my largest data set is around 700MB with

wc -l = 20602625 AreaThreeNonGround.xyz

If I import using v.in.ascii without a db and topology will I still be able to manipulate the vector data as described above and will I be able to use R scripts and functions on the data set?

Apologies for the long mail, I hope it is not too confusing.

Many thanks and looking forward to your comments and suggestions.
Regards,
Wesley

Wesley Roberts MSc.
Researcher: Forest Assessment (Remote Sensing & GIS)
Forestry and Forest Products Research Centre
CSIR
Tel: +27 (31) 242-2353
Fax: +27 (31) 261-1216
http://ffp.csir.co.za/

"To know the road ahead, ask those coming back."
- Chinese proverb

--
This message is subject to the CSIR's copyright terms and conditions, e-mail legal notice, and implemented Open Document Format (ODF) standard.
The full disclaimer details can be found at http://www.csir.co.za/disclaimer.html.

This message has been scanned for viruses and dangerous content by MailScanner,
and is believed to be clean. MailScanner thanks Transtec Computers for their support.

2

Hi Wesley,

I have some LiDAR data of a forested area and I am trying
to calculate tree heights / canopy heights based first and
second return points (pre-filtered ground and non-ground).
I have two questions that I hope some of you may be able to
help me with. Firstly I have windowed / clipped my data to a
small 'test' region with about 150 000 non-ground
and 100 000 ground points.

1. I have interpolated the ground points using idw in grass
and have taken that as my ground surface. I would now like
to assign an extra column to my non-ground vector file and
populate it with the raster values of the DSM to calculate
actual point height above ground (points are currently in
meters above sea-level). Traditionally, one would just
create a digital surface model and a canopy model and
subtract the two to get tree canopy height. I would like to
do the subtraction in the database and then interpolate the
result. I am trying to do this with v.rast.stats and it is
taking a really long time. How can I speed this up? Reading
the help page I also see that the vector data are rasterised
and values are assigend based on categories. I dont want
categories, I am just interested in the raster value
directly below each point? Is v.rast.stats the correct
application to use?

No, r.in.xyz is. :wink: Well at least it is much less trouble to work with.

(or if you want to stick with your current method, use v.what.rast instead of v.rast.stats to pick off single points at the data point; and v.surf.rst might make a nicer surface than IDW, at the risk of overshoots)

2. While the above mentioned data set is fairly small my
largest data set is around 700MB with

wc -l = 20602625 AreaThreeNonGround.xyz

If I import using v.in.ascii without a db and topology will
I still be able to manipulate the vector data as described
above

without a DB you can not populate a DB..
v.surf.* have been modified to work even without topology, just for this case.

r.in.xyz should handle 700MB input|20m points without complaint.

and will I be able to use R scripts and functions on
the data set?

don't know.

Apologies for the long mail, I hope it is not too
confusing.

Many thanks and looking forward to your comments and
suggestions.

hopefully r.in.xyz solves your problems, it is updated with more statistics and features in GRASS 6.3.0 if you can get your hands on that.

Hamish

3

Hi Wesley,

as Hamish points out r.in.xyz is the way to go. I have been working with large lidar data files over forest for a couple of months now and find the intersection between the ground and the vegetation to create the Digital Canopy Height Model - quite time consuming. Here is my basic workflow

read in the ground using v.in.ascii with topology turned off

v.in.ascii -tzbr input=grd.txt output=grd_pts x=2 y=3 z=4 fs=" "

create the ground surface using v.surf.rast (this is useful when you have irregularly spaced strikes under dense canopies)

v.surf.rst grd_pts elev=grd_dtm npmin=120 segmax=25 layer=0

create a flat vegetation file with non-ground strikes intersecting the DTM (this can be quite time consuming)

I use awk to calculate the difference between the canopy elevation and ground elevation = canopy height

cat AreaThreeNonGround.xyz | r.what in=grd_dtm | awk ‘print $1, $2, $3, $4, $5, $3-$5)’ | veg.txt

#I then use r.in.xyz to calculate statistics of binned data from the flat vegetation file
g.region res=10
#calculate the mean vegetation height greater than 0.5m for a 10m cell
r.in.xyz input=veg.txt output=mean_10m method=mean type=FCELL fs=‘|’ x=1 y=2 z=6 zrange=0.5,200 percent=100
#calculate the 90th vegetation height percentile
r.in.xyz input=veg.txt output=mean_p90_10m method=percentile pth=90 type=FCELL fs=‘|’ x=1 y=2 z=6 zrange=0.5,200 percent=100
…
…

I think the r.in.xyz has most of the statistics that you would like to use.

reading in the point cloud into R can be quite memory intensive

but it is fun to do

cheers

Andy

On Thu, Jun 19, 2008 at 3:44 AM, Hamish <hamish_b@yahoo.com> wrote:

Hi Wesley,

I have some LiDAR data of a forested area and I am trying
to calculate tree heights / canopy heights based first and
second return points (pre-filtered ground and non-ground).
I have two questions that I hope some of you may be able to
help me with. Firstly I have windowed / clipped my data to a
small ‘test’ region with about 150 000 non-ground
and 100 000 ground points.

  1. I have interpolated the ground points using idw in grass
    and have taken that as my ground surface. I would now like
    to assign an extra column to my non-ground vector file and
    populate it with the raster values of the DSM to calculate
    actual point height above ground (points are currently in
    meters above sea-level). Traditionally, one would just
    create a digital surface model and a canopy model and
    subtract the two to get tree canopy height. I would like to
    do the subtraction in the database and then interpolate the
    result. I am trying to do this with v.rast.stats and it is
    taking a really long time. How can I speed this up? Reading
    the help page I also see that the vector data are rasterised
    and values are assigend based on categories. I dont want
    categories, I am just interested in the raster value
    directly below each point? Is v.rast.stats the correct
    application to use?

No, r.in.xyz is. :wink: Well at least it is much less trouble to work with.

(or if you want to stick with your current method, use v.what.rast instead of v.rast.stats to pick off single points at the data point; and v.surf.rst might make a nicer surface than IDW, at the risk of overshoots)

  1. While the above mentioned data set is fairly small my
    largest data set is around 700MB with

wc -l = 20602625 AreaThreeNonGround.xyz

If I import using v.in.ascii without a db and topology will
I still be able to manipulate the vector data as described
above

without a DB you can not populate a DB…
v.surf.* have been modified to work even without topology, just for this case.

r.in.xyz should handle 700MB input|20m points without complaint.

and will I be able to use R scripts and functions on
the data set?

don’t know.

Apologies for the long mail, I hope it is not too
confusing.

Many thanks and looking forward to your comments and
suggestions.

hopefully r.in.xyz solves your problems, it is updated with more statistics and features in GRASS 6.3.0 if you can get your hands on that.

Hamish

grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user