Paul-
You and I may be the only ones interested...I posted several weeks back when I read Allen's post on the LIDAR that NPS and another agency had, of the Congaree.
Of course when I heard that there was already imagery for the Congaree, I thought, what do we need to collect to be able to connect with it.

Accurate GPS locations.  

That's all.  So I'm hoping that the Congaree folks have "synched" their GPS's!  At least have them all on the same Datum/Projection.

My interest has taken me to the Web, where I've found public domain software called FUSION, that works with the LIDAR data.

You can accurately measure tree heights to .5 to 2 feet, depending on data density.  I'm going out next month with some Inventory folks, who will visiting a site in advance of their upcoming Forest Inventory, which will involve LIDAR.  High density LIDAR data can collect up to 4 "hits" per square meter...pretty data rich remote sensing!

One of the guys that I'll be meeting with co-wrote the FUSION software, and is pretty up on this stuff.

How was your interest piqued?
-Don

Don, Paul,

I certainly can see the applications of the LIDAR data to the work ENTS people are doing.  For example if you had the topographic surface elevation data for a site and the LIDAR showed the top of the trees, then it would be feasible with the right software to simply subtract the two and get a tope of canopy height.  One of the things we are all aware of is that trees in valleys tend to be taller than their counterparts on the hilltops resulting in the surface of the tree tops reflecting a more subdued version of the general topography of the surface.  Dale has observed that along these river corridors the trees near the base of the slope at the side of the valley wall tend to be taller than those more in the flood play and on the slopes above.  These relationships that can be observed would stand out in a topo of canopy height generated by this process and the validity of these assumptions could be determined and documented.  Certainly it would give a broader overview things like canopy height topography relationships, and canopy height and relationships to many different environmental parameters.  What relationships are there that might jump out at you once the data is manipulated, compiled, and displayed?

It is something worth pursuing, but for me at the moment I doubt I have the capacity to manage these data sets effectively on my laptop, and even if I could I don't have the programs needed to do so, and don't have the topographic and LIDAR datasets  If big chunks of the LIDAR is available online, that is great, but I still would need the topo data sets that would be compatible with the process, and the data manipulation programs.    I don't have the bucks for a good GIS package or even a minimal one.  (If you have sources of freebies let me know.)  

I think many ENTS are interested in the LIDAR data, but just don't have the ability to make use of the data without being at a university.

Ed

Ed-
It might be that the NPS dataset could be accessed by an ENTS volunteer willing to spend time at the NPS office that has Congaree LIDAR imagery.  Often times they'll have competent computer facilities.
Gary Beluzo has a potential solution, with GIS computer competency.
I guess my point is, maybe not all ducks are yet in a row, but enough of them are that it may be worth looking into.  Your comment re topo surface elevation data being subtracted from the LIDAR data is correct...it's simple mathematics at one level, but with enormous repetitions such as computers are uniquely suited for.

The bottom line?  If we can access ACCURATE tall tree location data in the Congaree, it can be later used as an axillary data layer, and co-registered to LIDAR.  It may be that ground truthed data as accurate as ENTS collects might be valuable to NPS folks...something could be worked out, perhaps?
GPSs like Will's new Garmin are capable of being averaged, downloaded and post-processed into a potentially valuable dataset.
-Don

Don,

I've been following LIDAR for a long time and waiting for it's practical application by the general public.  The time is close to arriving with fast home computers, downloadable datasets, and amazingly, some free, public domain processing and visualization software.  Once all the pieces can be put together into a reasonable process, then we take advantage of this technology.  I've been following LIDAR since I first learned that it was being used to measure topography, tree height, leaf area indices,etc. probably years ago.  I found the information that I sent yesterday while researching old USGS aerial photos.  

On the USGS EarthExplorer web site, you can search for all possible downloadable or orderable USGS data on a specific point on the earth based upon graphic map selection or coordinate entry.  I was looking at their USGS and Army Map Service archive which goes back to 1939? for aerial photography.  They have georeferenced photo mosaic indices that can be browsed to get medium resolution downloadable roll and frame numbers to browse and download for old frames that aren't searchable yet.  I was browsing the more recent NAPP imagery and the older NAPP and Army mosaics, then browsed the preview thumbnails and low res imagery, then downloaded medium res frame scans.  While "poking" around on the USGS web site imagery product page, http://www.usgs.gov/pubprod/aerial.html, I noticed that they had a special link for LIDAR data that I have not seen before, so I investigated...

Anyway, I thought that some topographic data was inherent in raw LIDAR data.  Forested areas can be post-processed to "remove" the trees from data to produce "bare earth" topographic data without using secondary sources that aren't as accurate or at least relative to the local tree height and density data.  The raw data can also be used to produce LAI data, so it is obviously giving much more data than just the height to the first hard object.  I believe that this was inferred by articles posted here earlier about LIDAR tracking of invasive tree species in Hawaii?  and for locating potential ivory-billed woodpecker habitat.

You guys might want to talk to BVP about it at Congaree.  I believe that his University is one with a remote sensing lab that is actively involved with LIDAR and natural resources/forestry applications.

Non-Lidar FYI:  Additionally, recent high (2005 & 2008) and med. res. (2004&2006) leaf-on National Ag. imagery is available at
http://datagateway.nrcs.usda.gov/
or from local sites of members of the USGS AmericaView.org program, such as http://www.wisconsinview.org/

Regards,

Paul J.

Don Bertolette writes (Feb 19, 2009)

Paul-
I'm sure I'm oversimplifying it, but LIDAR has two wave types, one P and one L.  P penetrates the forest canopy and L is 'slowed' by the crowns.  The difference is the tree heights. So just using the P wave, one could create a LIDAR equivalent of a DEM (Digital Elevation Model-a 3D base map).

With High Density L waves, there are 4 pie-plate sized "hits" per square meter, with Low Density L waves, you only get 1 "hit" per square meter, but is much cheaper.

You mentioned U. of Washington folks, one of the two people I'll be joining, perhaps next month for a trip out into the field, is one of them...Hans Andersen is often one of three folks that co-author many LIDAR papers (for example, Andersen,
    H.-E., S.E. Reutebuch, and R.J. McGaughey. 2006, Forest measurement and
    monitoring using high-resolution airborne LIDAR. Society of American
    Foresters (SAF) South Puget
    Sound Chapter meeting, March 16, 2006, Fife, WA)

Back when I using FARSITE to model fire area growth, one of the hardest inputs to get was crown bulk density...what better way then to use LIDAR to obtain that! I think it would be an exciting time for young folks with the 'knack' to be in this field.

From my initial understanding (and I'm just getting my feet wet in LIDAR), one can do pretty well with a moderately competent home computer, as long as the datasets are relatively small. LIDAR is very data-rich, so large area investigations would move you into the workstation arena pretty quick.

Ed-
One of the anecdotal accounts on LIDAR I wanted to mention earlier came from a conversation I had with a LIDAR savvy friend...he said for the FIRST time in more than a decade, he was actually SEEING an individual tree from remotely sensed (specifically, LIDAR) data, and the prospects were exciting.

-Don

-Don

Continued at:  http://groups.google.com/group/entstrees/browse_thread/thread/c215704378436355?hl=en

In the past I played with Landsat HDF files but had a difficult time
classifying the data.  The resolution also wasn't useful for my/our
purposes, so I gave up on it.  I looked at AVHRR, SIR-C and SAR and couldn't
find practical applications of that data to ENTS functions.

On the other hand, the LiDAR is great.  The Fusion demo data is higher
resolution than the USGS online data, and gives a good picture of it's
potential usefulness.  From what I have seen so far, the USGS data has good
enough resolution to find and measure tall trees in a large landscape.  We
don't know how badly any of them are mismeasured badly yet, but it at least
allows us to focus our searches a little better.  Here's what I've done so
far:

Downloaded Fusion software and sample data from the "Projects" "FUSION/LDV"
frame:
http://forsys.cfr.washington.edu/
Downloaded Fusion tutorial pdf and data files from:
http://www.fs.fed.us/eng/rsac/fusion/
Get familiar with software.

Go to USGS LiDAR web site, use HTML/Java viewer to see where data is
available.
http://lidar.cr.usgs.gov/LIDAR_Viewer/
Download the .kml or .kmz file to better locate exact desired LiDAR tile
data LAS file in GoogleEarth or ESRI ArcGis Explorer:
At least for North Carolina, they are georeferenced to four tiles per DOQQ,
which is 16 tiles per 7.5 minute topographic quadrangle.
North Carolina has total coverage, Lousiana has partial coverage at USGS and
more coverage at a LSU web site,  other locations are spotty coverage with
additional free data available from the state university GIS web sites, or
free or for pay data from state, county, or local governments.

Once I had a LAS format LiDAR data tile from the USGS, I needed an image
file for the same tile.  DOQQ or digital orthophoto quarter quadrangles
based upon 1/4 of a 7.5 minute topographic quadrangle, are available for
most of the U.S. from the USGS and state GIS departments.  There are recent
free medium and high resolution ortho imagery files available from:
http://datagateway.nrcs.usda.gov/GatewayHome.html
and also from member states sites of the AmericaView program.  Wisconsin's
is:
http://www.americaview.org
I got my North Carolina file from

I got mine from North Carolina's GIS site hoping that it would be in the
same projection as their LiDAR file.  They were, but the only difference was
one was in feet and the other was in meters:
http://www.cgia.state.nc.us/DataResources/tabid/55/Default.aspx
I tried their online viewer, but I can't get it to work properly on my
computer at home or work:
http://204.211.239.202/viewer/
So, I manually found out which quarter quad (DOQQ) image I needed from their
ftp download site:
http://www.nconemap.net/Default.aspx?tabid=286
Here, I got a 1998 DOQQ CIR MrSID file and the metadata info.
First, I browsed the DOQQ names on the pdf index reference file:
ftp://204.211.239.203/outgoing/raster/doqq/cir_1998/
Then downloaded santck1 data from the directories down the folder tree.

Fusion can't read MrSID files, so I used free time-limited LizardTech
GeoExpress View software to convert it to a properly georeferenced TIFF with
a world file that I could edit.
http://www.lizardtech.com/download/dl_options.php?page=viewers
I couldn't find any other free software that would give me the conversion in
the same resolution and without messing up the georeferencing.  Their
plug-in version reduces the image resolution to the screen resolution.

The .tfw file is not in the correct units for Fusion.  I had to convert it
from meters to feet:
Original santck1.sdw world file data, (x-scaling, y-scaling of x data,
y-scaling, x-scaling of y-data, x offset, y offset):
1.000000000000000
0.000000000000000
0.000000000000000
-1.000000000000000
155061.204046555270907
191808.684446442872286

Conversion tables scale 1 meter = 3.2808 feet.  By scaling the values from
DOQQ metadata, the ratio of the LAS to image data can be properly scaled.
A little error on a large data set will create a large error at the end of
the x and y data.
From the metadata files, I determined the following conversion factor:
LiDAR LAS Projection false easting = 2000000.002617 feet
DOQQ Projection false easting = 609601.22 meters
2000000.002617 /609601.22 =3.2808333333338801388881734849546
I know, there are too many digits beyond the significant ones, but I just
cut and pasted out of the Windows calculator.  I multiplied all the entries
in the world file by this amount to scale the data to units equivalent to
the LiDAR LAS data file.

The resulting properly scaled santck1.tfw world file is:
3.2808333333338801388881734849546
0.0
0.0
-3.2808333333338801388881734849546
508729.96694282487296176202528535
629292.32555480953887910689675521

Both the LAS and TIF(TFW is automatically imported when the TIF is read) are
input data files for Fusion.  This allows viewing of 3D surface data with
color coded heights or color by image data for the data set with little user
experience.

What I'm doing now is using the groundfilter and gridsurfacecreate command
line tools to generate bare earth data to allow individual tree measuring.
I need to optimize the filter coefficients to get better results than the
default values allow.  So, far the default values provide a surface with
significant outliers.  They are above ground outliers, so they only make
some trees appear shorter, not taller.  Once that work is done for the USGS
datasets, the commands can be entered in a batch file to allow anyone to use
the data from USGS regardless of location.  I'll post another message with
screen shots showing individual tree measuring and 3D landscapes once I
optimize the ground filtering for some more reliable data.

Paul J.

I found that Topcon Total Station uses a conversion factor of 1 meter = 3.280833333333333 survey feet.  So, with more reasonable significant digits, I updated the TFW or TIFF world file described in Part 1 as follows:

3.280833333333333
0.0
0.0
-3.280833333333333
508729.966942740032947
629292.325554704592889

Then, with the LiDAR LAS file was converted to a ground points file with the following command.

c:\fusion\groundfilter /gparam:-2 /wparam:3 /iterations:10 c:\lidar\wrightcreek\groundpoints.lda 9 c:\lidar\wrightcreek\NC_Phase3_2005_35083c8c1.las

The surface file then was converted to a gridded surface for the software to use with the following command:

c:\fusion\gridsurfacecreate c:\lidar\wrightcreek\surface.dtm 3 F F 2 0 2 2 c:\lidar\wrightcreek\groundpoints.lda

It leaves a very few traces of outliers but has the truest lay of the land without distorting it excessively to make a pretty, perfect visualization.  State provided maps on the NC DOT web site are less than half the resolution (20 foot cells vs. 9 foot cells used here.)

First, I selected a portion of the data for what in the image file looks like an old growth section of Wright Creek:

Then, I subtracted the bare earth surface height from the tree canopy heights to flatten the image and show only tree heights.  Since we are looking for big trees, I set the software to color by height and by class and set it to color trees up to 130 in blue, in the 130's green, 140' in yellow, 150's in orange, and above 160 in red.  I set it to exclude trees under 150 feet tall.  These pictures are larger to show the content better:

Switched it to overhead mode, enabled the measurement marker and set it to snap measurement marker to highest points.  Then, I enabled the image plate underneath the data and measured the tree heights.   By hitting "h" and then enter, the height is updated and logged.  After logging all the desired points to check out, I saved the "measurement line" to a .csv file which in the form of x,y,height.  

Measuring can be done in overhead or 3D modes.  Colors can be smoothly gradiated or colored by ranges.  You can move the measurement marker tool over the color coded data view or over the image as shown here.  This is a tall emergent tree with a large crown.  It may be on a small rise or hump, or it may actually be a big tree.  Ground truthing will bear it out...

Regards,

Paul