One of the common questions we are asked is how many pulses per square meter (pulses/m2) should be prescribed in a LiDAR survey to result in a “good” digital terrain model (DTM / DEM). The question then becomes what density of vegetation is covering the terrain. This awareness about the relationship between LiDAR pulses/m2 and quantity of ground returns, used to create a digital terrain model when under plantation forest canopy. Therefore it is of interest to know the ratio of ground returns to first returns (RGR), the returns which bounce off the vegetation canopy, and in so identifying the minimum pulse intensity required to develop a sufficient DTM.

This article covers analysis of 3 datasets of LiDAR, two of them being Pinus radiata Plantation and the third a Douglas fir Plantation. We have analysed the data to gain a better understanding on this subject to help guide clients which are considering LiDAR collection over these types of forest. Pinus radiata Forest One was collected at four pulses/m2, Pinus radiata Forest Two data at two pulses/m2 and Douglas fir forest data at five pulses/m2.   Overlap specification for the flights was 50%.   All 3 flights were done using the sample LiDAR scanner and aerial survey provider.

Method of Analysis

FUSION software was used to process the LiDAR ‘LAS’ files and, in particular, the CloudMetrics function, which extracts ground return values and canopy density values.

The study method below was followed for these forests:

  1. Create DTM, cell size of 1 by 1 meter, using FUSION command line GridSurfaceCreate.
  2. Create two CloudMetrics tables, per plot, using FUSION command. One of the tables has the metrics from the unfiltered data files and the second one has the metrics from the bare earth (ground return) data only. Two cut-off specifications were used; a 0.5m height cut-off to separate off the ground return from the vegetation data, and a 0m and 60m cut-off filter to remove point data aberrations.
  3. Generate the ratio of ground returns (RGR): Ground (bare earth) count divided by Canopy Density (first returns) count.
  4. Create relationships between forest parameters and RGR (ratio ground returns).
Results

To help understand how the ratio of ground returns to first returns (RGR) changes with typical forest metrics, six graphics are shown below for each of the Pinus radiata forests used in the study: RGR vs. Forest parameters (Basal Area, Top Height, Stocking, Total Recoverable Volume, Total Stem Volume and Age). For the Douglas fir forest RGR vs. four forest parameters; Average DBH, Average Height, Stocking and Age.

Pinus radiata Forest One

033014_2135_whatisagood1

Pinus radiata Forest Two

033014_2135_whatisagood2

 

Douglas fir Forest

033014_2135_whatisagood3

Discussion

It should be noted that:

  • the Pinus radiata Forest One area covered by the data included a wider range of age classes and stocking, BA, etc therefore show stronger relationships with RGR than for Pinus radiata Forest Two. In addition, it was difficult to do a comparison with stand ages younger than 10 years old and those older than 10 years. This was due to high variation in low canopy closure in young highly stocked stands due to the effect of periodic thinning.
  • It can be seen in the Pinus radiata Forest One graphs that Ratio of Ground return (RGR) decrease with increasing age, total recoverable volume, top height and basal area. RGR relationship with stocking is dependent on canopy closure level in the stand.

To answer the original question of what the minimum pulse per square meter is required to create a good DTM model, the comparison is best between forest age classes greater than 10 years old. We can see in the following graphs that there is a higher RGR in the Pinus radiata Forest One (four pulses per meter square) than in Pinus radiata Forest Two (two pulses per meter square) due to pulse intensity in the LiDAR data when collected. The Pinus radiata graphics for Forest One give us more information than Forest Two, so if we look at these and the Douglas fir graphics we can see a similar relationship between Basal Area and Average DBH and Top height and Average Height. However, the stocking shows a different behaviour for Douglas fir and this could be because this species grows slower than the Pinus radiata and canopy closure therefore later.

033014_2135_whatisagood4033014_2135_whatisagood5033014_2135_whatisagood6

RGR for these three populations with ages greater than 10 years old.033014_2135_whatisagood7

033014_2135_whatisagood8

 

Conclusion

We can see in the box whisker plot that the Pinus radiata Forest One has a higher RGR number so more pulses arrive to the ground during the LiDAR data flight. Even more evident is even with 5 pulses/m2 the Douglas fir forest has a much lower RGR mean than the Pinus radiata Forest One with four pulses/m2 square showing stocking and canopy closure has the strongest influence on RGR.

Finally, we can check our graphic results if we apply the Ratio Ground Returns formula where GR is ground returns and PD is pulse density. This formula can extract the quantity of the returns which arrive to the ground.

033014_2135_whatisagood9

Pinus radiata Forest One flown at 4 pulses/m2 has the best quantity of ground returns (0.35376), so if we decide to generate a DTM for this forest, the DTM quality would be at least 4 times better than the other radiata pine forest flown at 2 pulses/m2. The difficulty in getting good ground results is obvious in Douglas fir due to the highly stocking and canopy closure of this species as grown in plantations in New Zealand. This highlights that there is not just a single answer to this question and careful planning and analysis of the forest types and LiDAR scanner technology is needed.

If you would like to better understand this for your forest type and flight planning for capture of LiDAR, please feel free to contact us.