LiDAR mapping and Artificial Intelligence for forest inventory automation

Forest LiDAR mapping visualization

Forests provide us with a large number of ecosystem services.

Knowing in detail the current state os these forests and the possible changes they may undergo is of vital interest, both for managers, researchers and forest administration.

To this end, in 2020, the Operational Group for innovation in sustainable forest management go monte digital, successfully completed the innovation project funded by the National Rural Development Program, relating to support and rural development through the European Agricultural Fund for Rural Development.

The Monte Digital team originates from the collaboration of three entities: University of Castilla-La Mancha (UCLM), Dielmo 3D and Naturtec.

Table of contents

Understanding the forest inventory

What is a forest inventory?

A forest inventory is the main tool for describing the structure of the forest(understood as the set of trees in the forest) and quantifying its renewable resources.

Traditionally, the forest mass condition has been determined by foot count  inventory or by statistical sampling, using plots of a certain size (classic inventories).

Classic forest inventory

Main drawbacks of traditional forest inventory

Whether by foot count or statistical sampling, these methods require a large amount of time and human resources to be carried out correctly. The excessive cost of these traditional inventories, in terms of time and money, often makes them impossible to complete, thus hindering the correct (sustainable) management of the forest mass. 

Technological solutions

In the last decade, we have been witnessing a small revolution in the tools available for the measurement of forest resources, due to greater technological development. Such is the case of LiDAR (Light Detection and Ranging, or Laser Imaging Detection and Ranging) technology.

LiDAR mapping as a forest inventory tool

In this scenario, new technologies are key when looking to optimize forest inventories, both in time and cost, for which the main objetvice is the usage of the natural resources the forest provides.

lidar forest animation

Aerial LiDAR vs LiDAR mobile

So far, only AIRborne LiDAR has been used as an attempt to improve the estimates made by classical inventories. However, the results in the quantification of forest resources have not been very efficient.

Because of this, our operative group has used the same laser technology, but in this case, instead of applying it from the air, applying it directly from the ground (LiDAR mobile).

In the case of AIRborne LiDAR, once the points are classified as vegetation, we then try to predict the forest mass variables based on statistics and point cloud variables (percentiles). However, AIRborne LIDAR does not satisfy many requirements because of the inability to observe tree trunks.

This is where LIDAR mobile comes into play, which, due to its characteristics, can observe what is interesting to forestal technicians, among others, the trunk of the tree.


In order to automatically identify trees and obtain dendro and dasometric parameters, we have developed an automated software that executes a sequential process in three phases:

  1. To handle large volumes of data, such as those generated by the MTL, pre-processing of data is required. The input data is divided into blocks with an overlap between them, so that there are no discrepancies in block changes. A series of automatic products are then generated, such as soil classification, digital terrain and surface models, and point cloud normalization.
  2. Tree detection using artificial intelligence. For this purpose, an image object detection model has been trained to detect rings on raster images in the LiDAR point cloud at different heights. The combination of all the different detections at different heights gives us three-dimensional information of the position of the trees and allows us to create output vector layers with the position of these, as well as to measure their geometric properties.
  3. Finally, for each identified tree, the different dendrometric parameters are generated and by addition, the dasometric parameters at plot and canton level.
Aerial lidar vs terrestrial

Results obtained

  • Validate the reliability of the inventory by means of mobile terrestrial lidar (MTL) at sample plot scale, using standard commercial software to process the point cloud.

  • Generate algorithms for the calculation of dendrometric variables based on the point cloud, integrated in a user-friendly software that can be easily used to quantify these variables in an automated way (artificial intelligence technology).

  • Application and validation of the algorithms at a larger scale than the plot (canton).

Forestry LiDAR software

The final product of this project, undoubtedly, has been the development of an innovative methodology aimed at the digitization of the forest, through the inventory of forest masses accurately, quickly and at low cost, allowing monitoring and reliable decision making, in sustainable forest management, using the Mobile Terrestrial Lidar Laser (MTL) technology.

This product has also included the development of the specific AID Forest software.

What are the benefits of this project for future applications?

With this project, Dielmo opens the doors to any company that requires a powerful and efficient forestry application. Being able to carry out tasks such as:

  • Automate forest inventory
  • Increase forest stand analysis capacity
  • Save time
  • Perform real measurements of 100% of the terrain
  • Bring new technological solutions to forestry services

Share this:

Related posts

Copyright © 2020 Dielmo 3DAviso legalPolítica de privacidadPolítica de cookies