We had the wonderful opportunity to test the L2 at El Mirador-Calakmul Karst Basin in Guatemala.  This was an effort in collaboration with DJI Enterprise, Drone Plus +, SkyCam and DEL - Desarrollo, Evaluación y Logística.  No down time warming IMU and way more points per second with 5 rebounds.  And so, we are now looking forward to scan over 550 new square kilometers of dense jungle with L2s and M 350s; lower altitude and slower scanning speeds m/s allowing much more density of points/m2 and Grtn.  This is an effort to support FARES and Dr. Richard Hansen who has dedicated his life to protect and promote El Mirador-Calakmul Karst Basin in Guatemala.

The DJI Terra for L2 has proven way more ground points with powerful range and elevated precision exceeding comparable results with RIEGLs LMS-Q1560 and LMS-Q780 used between 2015 through 2018 in the El Mirador-Calakmul Karst Basin.  The RIEGLs scanned at nominal altitudes of 550 m and 650 m obtaining at most 48.39 points/m2 (density including vegetation) and 2.95 Grtn/m2 with swath withs of 513-741 m and overlaps of 55-70 percent with 15-20  pulses/m2.

El Mirador-Calakmul Karst Basin LIDAR Analyses

We will surely work a webinar with DJI Enterprise to share technical results.

Do you have a dataset to share? Before process in Terra.

LV_Forestry Posted at 10-10 08:03
Do you have a dataset to share? Before process in Terra.

Hello. We will soon share a sample data set when we come out with the webinar.  L2 was just launched today at InterGeo in Berlin, and we were just authorized to publish our El Mirador experience.

Here is a preliminary LiDAR analyses for our last mission into the El Mirador-Calakmul Karst Basin testing the DJI Zenmuse L2 LiDAR Camera compared to previous missions with RIEGLs and DJI Zenmuse L1 LiDAR Camera.

LiDAR Analyses Report

I knew we had already communicated on the forum but I didn't know where. I just found out, were you the one who wanted to put the DRTK2 on a 30m mast?

No disrespect but the data you present which highlights L2 values is totally biased. What emerges is that this sensor generates more points than its competitors for the same flight parameters. And that's it. Too many points (over-sampled cloud) equals more noise, more interference... This is why the sensor is not locked at 240kHz...

With a LiDAR there are essential things, note that it is not because your RAW cloud point will output x million points that these x millions will be quality trusted points. Noise need to be removed among other essential task.


In the document you mention that the processing was done with Terra. This software does not do post-processing. It only merges the data from the different sensors and creates a cloud of points in a format readable by other software. It can roughly do ground point classification, but after testing it is not at all good. So I have a lot of doubt about the veracity of your rasters obtained with this set.

Figure 6 of the document leaves me perplexed. Either it's a marketing team that knows nothing about it that made the drawing, or well, I don't know exactly!

Figure 7 there is something troubling, 5 returns is this really functional or a commercial argument?

Figure 12, the difference is certainly not due to the sensor but to post-processing. Send me the two datasets, I am convinced that I obtain the same result with both.

In conclusion you describe "How smoother is the point cloud". So I don't want to destroy your work, but I still alert you to the fact that the "smooth point cloud" function of DJI Terra is good to throw in the trash.

In the image below you can see the true curvature of the ground in red. And in blue what Terra decided it would be. Suffice to say that if you are looking for vestiges of infrastructure under the vegetation, you are certainly missing a lot.


I don't know who at DJI will present the L2 during the webinar, but if it's the guy, I think I recognize in the video you posted, then we have to hope that his approach will be less commercial and more technical than during his previous performances made for other products.

If you have to prepare the webinar with them, please keep in mind that what interests us as surveyors is to know stuff like :
How many cm the point cloud shifts in real conditions, with DRTK2, with NTRIP , in PPK...?
What is the necessary calibration frequency if there is one?
What Terra output formats are possible? (geoid conversion)
What camera settings are possible, white balance, focus...?
Is it recommended to do a particular pattern to calibrate the IMU before launching the mission?
What level of air humidity affects the measurements?
Up to what latitude is the assembly capable of operating?
...

We don't need to ear much about the "Christmas decoration" features. Pin point, Overview on RC Plus...

Thank you for your feedback. We will work on a white paper towards a webinar on the experience and results.  Your comments have us working on a response, amplification and extension of this first draft.

Wow, interesting to see!