In this entry, I will discuss another application of LiDAR technology for estimating timber stocks in a forested area, focusing on an extreme case (the wood volume estimation of eucalyptus plots) that many professionals in the sector have had to deal with at some point.
INTRODUCTION
Terrestrial LiDAR technology, also known as mobile or personal terrestrial laser scanning (TLS), is gaining popularity in recent years. It offers an alternative approach to traditional forest inventory mensurations by gauging distances to multiple points on surrounding object surfaces, thus being able to map a forest from its core. LiDAR generates three-dimensional point clouds and enables rapid spatial representation of forested areas, capturing the profiles of individual trees. However, these devices come with a high cost, often exceeding €30,000, making them inaccessible to many potential users. Their weight poses another challenge, hindering mobility in certain areas and increasing transportation costs. The need for specialized software programs also limits the user base of terrestrial LiDAR, requiring specific training. To achieve broader acceptance as a valid alternative to conventional inventory methods, there is an urgent need to find affordable, simple, and efficient hardware and software solutions.
Since 2020, Apple Inc. has been incorporating a LiDAR sensor into some iPhone and iPad models. Compared to other scanning devices, Apple mobile phones equipped with LiDAR are available at significantly more affordable prices (starting from €729) and are lighter (187–684 g). These phones have proven to be suitable for capturing 3D point clouds in forest environments. However, to extract information about individual trees from the obtained point clouds (e.g., variables like trunk diameter), further analysis on an external device using various programs is required, due to the current lack of specific Apple applications.
This is where ForestScanner, a free application developed by the Japanese company MAPRY Co., comes into play, enabling LiDAR-based forest mensurations using an iPhone or an iPad. ForestScanner estimates trunk diameters and spatial coordinates of each tree in real-time, while users scan the trees, allowing them to view, verify, and share results in the field. These features facilitate rapid forest inventories and improve accessibility to LiDAR technologies for individuals without prior experience.
MATERIAL AND METHODS
The ForestScanner app is available for free on the App Store, and it works on iPhones or iPads with LiDAR sensors. This specific experience has been tested on an iPhone 15 PRO. As of the publication date of this post, there are no Android devices with built-in LiDAR technology.
«The chosen location to test the application and device has been one of the toughest, for this type of measurements, that can be found in my surroundings: a plantation of Eucalyptus globulus. Given the fragmented nature of land ownership in Galicia, these plots are usually small (around 1,500 m2 in this case), which means that the lack of precision of the location device of the mobile phone can lead to significant errors. The mild climate and clear canopy of the eucalyptus forests leads to the proliferation of understory vegetation, reducing mobility and operability in the field. The small size plot also prevents statistical sampling, making a tree-by-tree measuring mandatory. Figure 1 shows a panoramic photo of the study plot.
The operating mode is as follows: with a maximum scanning distance of 5 meters, as the user moves the device with ForestScanner through the forest stand, it generates a 3D point cloud of the surrounding surfaces around the phone’s target. While in use, the point cloud and 3D triangle meshes appear on the screen in real-time, allowing users to visually recognize the surfaces being scanned (Figure 2; Video 1). The point cloud is colored with RGB information collected by the device’s own camera.
During scanning, ForestScanner tracks the relative coordinates of the device from the starting point of the inventory, based on the Inertial Measurement Unit (IMU) of the phone. The absolute location (i.e., geographic coordinates) of the starting point is determined using the built-in GNSS receiver. This can lead to significant errors, as will be seen later on.
For each tree, we position the crosshair symbol (in Figure 3, the small magenta cross) at the desired height where the diameter is to be measured (e.g., diameter at breast height, or BHD). Then, we tap the tree icon to record the trunk, along with its corresponding diameter. The numerical value of the diameter will be displayed on the screen attached to the tree (Figure 3). The recommended lower threshold for the diameters to be measured is 5 to 10 cm, considering the limitations of the iPhone’s LiDAR sensor in detecting objects of a few centimeters in size.
The user can thus register several trees in a single pass, moving from one tree to another while scanning. Alternatively, the user can touch the «Pause» button, move to the next tree, and then touch the «Play» button to register it and continue with the inventory. During the pause, the LiDAR sensor stops scanning while the IMU continues to track the device’s location. Using the pause function helps reduce the size of the point cloud file.
The program also allows the user to manually record the species to which each tree belongs by tapping the «Tree species» button and entering it. This can be done using the on-screen keyboard or the microphone of the iPhone/iPad. Once the «Stop» icon is pressed, the inventory is finalized, and it is no longer possible to enter data into the same database. The acquired point cloud will be displayed on the screen (Figures 4a and 4b).
RESULTS
The first thing to highlight is the handiness of the entire process, as it will always be more comfortable to walk through a forest stand holding only a mobile phone than carrying a tree caliper, a notebook, a marker (chalk or spray), etc. That said, forest mensuration using ForestScanner is not without its problems. I will now list the ones that, in my opinion, are most relevant.
First is the dangerous tendency of the program to freeze. And I say dangerous because, once the application hangs and stops working, all data recorded up to that point, from the beginning of the recording, is lost. It happened on several occasions, and while it’s true that the program is quite stable, the few times it happened to me were quite frustrating because I had to start over.
Another issue lies in the amount of data that the program is capable of handling. It’s not about the number of trees, but rather the size of the database being generated. After entering nearly fifty trees, a warning appeared on the screen: the point cloud was too large. Therefore, the program was going to stop recording data. Although, obviously, you can open another batch of measurements to gather them all into a single file back in the office, doing so is inconvenient and can also may lead to registration problems and errors.
In relation to the previous paragraph, the problem of calculating the diameter of a trunk appears. Using a single pass the program calculates the diameter based on what it records from that perspective. However, it is not uncommon in eucalyptus, as well as in other species, for the trunk section to be anything but circular, so that from one side we might record a section of 25 cm, for example, and turning 90 degrees from the previous position we register a diameter of 30 cm. That’s why, when measuring with calipers, we are forced to record the average of the two diameters measured perpendicular to each other, or diameters in cross-section. This problem, using ForestScanner, is solved by recording along the perimeter of the trunk, without the need to completely surround it. Once operated in this way, the results are not very different from those obtained with calipers, as can be seen in Table 1.
Table 1. Measurement of the same trees with calipers (taking the average of the two cross-sectional diameters) and with ForestScanner..
| Forest caliper (cm) | ForestScanner (cm) | Species |
| 35,2 | 34,3 | Eucalyptus |
| 34,6 | 32,1 | Eucalyptus |
| 13,2 | 12,9 | Eucalyptus |
| 23,4 | 21,4 | Eucalyptus |
| 17,3 | 16,7 | Eucalyptus |
| 46,4 | 45,1 | Eucalyptus |
| 19,7 | 21,5 | Eucalyptus |
| 29,1 | 33,5 | Eucalyptus |
| 15,9 | 15,5 | Eucalyptus |
| 22,1 | 22,4 | Pinus |
| 8,9 | 10,1 | Birch |
| 11,0 | 12,3 | Oak |
However, this recording method significantly increases the size of the point cloud and, along with that, comes the instability of the program.
Another problem is the assignment of geographic coordinates to the trees, although the application is not to blame for this matter. The precision of the internal GNSS (Global Navigation Satellite System) in mobile devices is very limited, as they calculate positions autonomously without using differential methods that allow for corrections. The accuracy is optimally around 2 meters under clear conditions (which are rarely found in forest environments). This can pose a problem when determining which trees are included in our inventory and which ones are not. In my case, as shown in Figure 5, the trees at the southern end of the plot appear somewhat distant from it. I am certain that the trees on that side were accurately located because the plot bordered to another recently cleared, but what about adjacent plots with trees of similar age to ours? The task of verifying whether a tree is included in our mensuration or not may lead to additional verifications that slow down and increase the cost of the work.
In Figure 5, it can be observed how, in the first round of measurements, the trees were correctly framed within the limits of the plot. The second pass, however, shows a discrepancy that leads to doubts about which trees should be counted for the inventory. In a small assessment like this one, such a gross uncertainty can lead to errors and very disparate results.
The iPhone doesn’t have good geolocation apps (free of charge, in any case), so it’s not possible to monitor movement through the forest and know firsthand and in real-time whether we are straying from our target plot and recording trees that are not included in our inventory plan.
Although it is not be the main purpose of the application, the point cloud it generates is not comparable to that of conventional laser scanner devices. Signal return intensity values are not recorded, and it is not possible (as far as I have tested) to determine heights, even indirectly through post-processing of the point cloud (Figure 6).
CONCLUSIONS
Despite everything discussed so far, I cannot help but be amazed by this application. The potential it holds is enormous; the time and effort savings it can provide for forest inventories is considerable. Perhaps, in its current state, it may not be of interest from a professional perspective for the most demanding tasks, but if LiDAR technology becomes widespread in mobile phones, improvements can be expected, and even the creation of other applications that may well address the problems mentioned earlier.
The context in which ForestScanner has been tested is one of the most demanding in terms of physical effort and execution time. In other scenarios, such as statistical sampling by plots or inventories in more accessible stands, everything seems to indicate that the program could perfectly fulfill its purpose. The difference in equipment cost, compared to other LiDAR measurement setups, or even more conventional inventory methods, is enormous. It is enough to indicate that an electronic caliper, which would serve a similar function, costs around €2,000.
The evolution of this technology, applied to dasometry, is very promising. If you have a LiDAR-equipped iPhone, you should definitely give it a try.
As a line of work for future publications, the combination of data collected through ForestScanner with the aerial LiDAR point clouds of the Spanish Plan Nacional de Ortofotografía Aérea (PNOA) is proposed, to determine the timber stocks of a given surface.
FURTHER READING
Article where the creators of the application present and analyze it: here.
AKNOWLEDGEMENTS
To Carmen A., for her corrections.
atlas_enxeñeria 