I'm using the L2 to scan vegetation. I need a high point density to return accurate vegetation parameters. I'm using a 50% overlap, which I figure will return as close to an even point density as possible (I know due to the geometry that a completely even point density is not possible). Even so, the flight lines are very visible in some of the products due to the uneven point density. I've been experimenting with the "by distance" point reductioon in Terra to try to even out the density, but I'm getting strips with about 600points/m2 next to strips of about 900points/m2.

Does anyone have any practical advice for achieving even point density? I'm trying voxelisation with small voxels but I'm worried about distorting some of the finer structures (which are very beatuiful in the L2 data, btw).

Lidar L2 and L1 are "low-end" equipment. You have largely overestimated its capabilities for your project. Especially with Terra.

Start by reducing the linear speed to the minimum possible, it could help.

LV_Forestry Posted at 8-15 01:43
Lidar L2 and L1 are "low-end" equipment. You have largely overestimated its capabilities for your project. Especially with Terra.

Start by reducing the linear speed to the minimum possible, it could help.

So with higher-end equipment it would adjust the scanning pattern to achieve a consistent point density?

glenrs Posted at 8-15 01:53
So with higher-end equipment it would adjust the scanning pattern to achieve a consistent point density?

I recommend you to go have consultation with Lidar manufacturer such as Yellowscan regarding vegetation survey.

The density is function of the surface reflected, angle, intensity/color, material...
Therefore believing that you will get uniform density on the raw point cloud is only hypothetical, and most likely not going to happen.

I'd recommend to not use DJI Terra for post-processing. Use another software specialised for that : Blue Marble, Terrascan, ...

Yellowscan is indeed much better but there is no free lunch...it costs on software licenses more than a new L2 each year. As always, ask yourself : do you really need that and are you willing to pay for it. Or good is good enough :-)

I have previously used a Hesai-Pandar XT-32 based system. The L2 seems to compare well, though I haven't done any formal tests. I've been impressed with the detailed trunks visible underneath tree canopies.

I only do processing to las in Terra, the rest in CloudCompare and LASTools.

I am well aware that local point density will vary. My problem is striping in VCI and other products. This is a desnity raster. Dark bands around 600 points/m2, bright bands around 900.


I have been trying lasvoxel and lasthin to even out the density, but with limited effectiveness since I'm trying to keep the point density as high as possible. It would be easy if I wanted to reduce to 10points/m2, but that would destroy much of the information in the data.

I figure someone must have encountered problems like this before, but I'm failing to find anyone who can address it directly.

AR_AirPrecision Posted at 8-15 05:05
I'd recommend to not use DJI Terra for post-processing. Use another software specialised for that : Blue Marble, Terrascan, ...

Yellowscan is indeed much better but there is no free lunch...it costs on software licenses more than a new L2 each year. As always, ask yourself : do you really need that and are you willing to pay for it. Or good is good enough :-)

Agreed. For the last system I purchased I investigated YellowScan but balked at the cost, given there are other Velodyne based systems with similar characteristics for much less cost.