top of page
Search
  • Writer's pictureVashaun Henderson

Topography in densely Vegetated Regions | LiDAR versus Photogrammetry | What is the best way to map forests

Updated: May 17


Balloon Creek in the Seymour River Area surrounded by dense BC forest
Looking north from a dirt access road at Balloon Creek in the Seymour River area

Compared to 2011, drone based mapping methods like LiDAR and Photogrammetry are becoming more affordable for not only large companies and large budgets but, also small businesses and even private land owners. With this new level of affordable access to these sophisticated mapping processes, the big questions our clients have to ask becomes not can I afford to do this but, which service is right for my situation. Further to that end, LiDAR costs more, so when do I need it, and when can I go with the more budget friendly photogrammetry option? One of the biggest deciding factors to that question is what are your site conditions and, more specifically, what are your vegetation conditions. Before we go down the rabbit hole, lets zoom out a little.


What is LiDAR


drone lidar getting ready for launch
Rekon DJI M300 drone with LiDAR senor sitting on the launch table waiting for power up

The word LiDAR is an odd acronym of sorts, it stands for "Light Detection And Ranging". It is a term used to describe a process of remote sensing that uses laser light to range a target surface or feature, inertial sensors to define an orientation, and a GNSS (global navigation satellite system) to provide global positioning. When the laser in the LiDAR system sends out one laser beam from one of it's emitters we call that a "pulse". Most modern systems produce hundreds of thousands of pulses per second. When a pulse hits an object it bounces back toward the LiDAR module and is read by a sensor, we call this a "return". A typical drone lidar mapped area will be covered by enough pulses to produce a point cloud with 400 points per square meter! That is equal to 1 point every 2 inches, and, that is the low end! Mapping missions can be planned in difficult areas to contain more than 1000 points per square meter, we generally reserve such density for difficult to penetrate canopies or scans that require very high detail.


What is Photogrammetry


3D point cloud from drone photogrammetry of dense forest in the Vancouver area
The above screen shot is from a photogrammetry processing software showing the 3D points covering trees. Each purple pyramid represents the position of an image and the direction the camera faced during capture.

Photogrammetry is a process through which photos taken of a target area in an arranged pattern are used to compute an accurate 3-dimensional model of the target. This process uses similar key points it identifies in adjacent images to triangulate these key point locations. This triangulation process is done many thousands, millions or even billions of times to produce a point cloud.

DJI Zenmuse P1 with 35mm Lens
DJI Zenmuse P1 with 35mm Lens

The main piece of equipment required for photogrammetry is the camera. This process can be performed on images from almost any camera but, there are factors that contribute to the accuracy of the 3-dimensional model made. The biggest factors are the resolution of the camera's sensor, the amount of overlap of adjacent images, the amount of distortion inherent in the camera lens, the ability to geotag the images with accurate geographic positional data, the current lighting conditions onsite and the presence of ground control targets in the mapping area. There are many more factors but, that we can save for when we dive deeper into the mechanics of photogrammetry, for now let's continue on the topic of acquiring topographic data in dense vegetation.


The challenges of mapping dense forests


Canada has some of the most beautiful landscapes in the world but, I suppose if your landmass comprises nearly 10% of the entire world's landmass you may have a pretty good chance of having a few nice looking parts sprinkled in here and there. With those kinds of numbers we are also likely to find that some areas are very densely covered in vegetation.


A screen shot of the in-flight heads up display from the DJI M300 drone of a dense forest LiDAR mapping mission in the Vancouver area
In flight view of the Lower Seymour Conservation Reserve Drone LiDAR mission

On a recent visit to the Lower Seymour Conservation Reserve our mapping team was faced with a big challenge. Trees in this area have grown together in places very tightly and have less gaps in the canopy for light, lasers or images to get through and capture detail on the ground below. The big challenges here are acquiring enough laser points or images in the target area through the tight canopy to produce an accurate and representative model of the topography below the canopy.


How does the vegetation penetration differ between LiDAR and Photogrammetry


Drone LiDAR point cloud cross-section view of dense forest
0.5m crossection of lidar data in a dense forest

Drone photogrammetry point cloud cross-section view of dense forest
0.5m cross-section of points made using photogrammetry in a dense forest

Here is where the rubber meets the road and the major differences become more apparent between LiDAR and Photogrammetry. Above in the two images there are cross-sections representing LiDAR (in white) and Photogrammetry (in red). In the LiDAR cross-section we can see that in some tight tree cluster areas the thickness of the points on the ground gets less, but some still make it through. When we look at the photogrammetry in the same locations of those tight clusters, instead of some reduced pentation we get none. This is the major difference between LiDAR and Photogrammetry in terms of mapping topography below a dense vegetation canopy.


Where LiDAR excels


Why does this happen? There are a few contributing factors, for the first one, LiDAR is an active method of measurement. This means the sensor is actively sending out energy into the environment in the form of a tightly columnated laser pulses. It does this so many hundreds of thousands of times per second that only a small fraction need to make it through the canopy to produce adequate coverage on the ground surface.


Where Photogrammetry trips up


Photogrammetry is a passive form of measurement, the act of taking a photo requires the presence of light, in this case, provided by the sun. It also requires a high percentage of overlapping data. Each image requires a minimum overlap with it's neighbors, most mapping mission require 75% in typical mapping environments and in the case of dense vegetation 90% to 95% overlap is required into order to collect enough data for most post processing software's to be able to build an accurate model. The mission above with the cross-section shown in red is an example of 90% front and side overlap and even with the enormous amount of overlap we still see that sections of the ground are simply occluded by the dense canopy. This is mainly due to the fact that even though we surely have one or maybe two photos that get a peak through the canopy in these dense areas the few pixels that represent that data are not shared by enough of the photos adjacent and so the software does not have enough data to build 3-dimensional points in that area.


Where does Photogrammetry excel


Now knowing why photogrammetry has a hard time in vegetation you can probably guess that in clear open terrain like the sage brush hills in Kamloops Photogrammetry would do well to provide ground topography. Development projects that have already stripped the vegetation from their site, gravel pits, urban mapping areas and city centers are all environments with less canopy to block the ability of photogrammetry to provide useful topographic data.


Drone LiDAR and Photogrammetry generated 3D model of Penticton hillside with a mix of low and high vegetation
Low density vegetation in the lower right of this image is a good example of an area where photogrammetry does well . The upper left is also a good example of an area not ideal for photogrammetry.

Photogrammetry equipment today, the variety capable of producing high levels of accuracy, cost about 1/3 of what an equally accurate LiDAR system would. In terms of setup it is also more simple to install, initiate a mission and setup the ground control. Photogrammetry does have something LiDAR does not, photos! The very photos taken and used to produce topographic data (or not) can be used to build a Ortho rectified image of the mapping area. An Ortho rectified image is an image that lacks parallax or, more simply, has a perfectly downward looking view of the area mapped at any point in the image. These images, if properly processed with ground control, are scalable and are usually used to augment the topographic data in design and planning activities.


The Best Way to Map Forests


To take all of the above and sum it up into a few sentences. LiDAR is an active measurement system that performs well in the challenging environment that dense forests present. Photogrammetry is a passive measuring method and does not do well at providing topographic data in dense forests but it does provide ortho-rectified-imagery. Together LiDAR and Photogrammetry provide different components that are used to inform design, management and planning.


So, what is the best way to map forests? From what we have learned above it depends, the better question to add on is what do I need when we are mapping this forest. If what I need is a good topographic model of the ground in a dense forest then LiDAR is the best option. If all I need is an ortho image and only the top profile of the trees are needed then photogrammetry is fine. If we need both a good ground topography and an ortho image of site then both LiDAR and Photogrammetry is needed.


With all of that to keep in mind, it can be a little overwhelming, and, that is why our team at Rekon will always be there to help you choose the right service or services for your project needs. Reach out today for a free consultation.










20 views0 comments

Recent Posts

See All

Comments


bottom of page