Advancing forest inventory: a comparative study of low-cost MLS lidar device with professional laser scanners

Abstract. In the context of forest inventory, there is a growing need for 3D data to produce detailed geometric information. While terrestrial laser scanning (TLS) is traditionally used for this purpose , several factors have prompted the exploration of alternative solutions, such as handheld mobile laser scanners (MLS). One key limitation of TLS is its static data acquisition, which makes it less suited for the complex and heterogeneous nature of forest environments. A primary challenge with TLS in forestry is the occlusion effect, where parts of trees (such as stems, branches, or leaves) may not be captured due to obstacles between the scanner and the target. Additionally, TLS is known for long acquisition times, which, while yielding high-quality data, may exceed the requirements for standard forest inventory tasks. The cost associated with TLS is also significant; although feasible for small forest patches, scaling these methods to larger areas would demand substantial resources. Similarly, while handheld MLS devices offer more flexibility in data acquisition and the possibility to cover a wider area in the same acquisition time, professional versions are still relatively costly, adding to the need for more affordable alternatives. This underlines the demand for a low-cost, efficient method for 3D data acquisition in forest inventories. In this study, forest structural variables obtained with a low-cost MLS (LC-MLS; Mandeye) were compared with two professional MLS devices (GeoSlam Horizon and GreenValley LiGrip H120) and a professional TLS (Trimble X7). With the open-source software 3DFin, we processed the point cloud data from all the devices, enabling the extraction of diameters at breast height (DBH) and total tree heights (TH). The LC-MLS device shows a positive bias in DBH measurements (1.62 cm), indicating it tends to overestimate compared to the TLS reference. Despite this, it demonstrates competitive quality relative to the two other MLS systems. In terms of TH, the LC-MLS has a negative bias of −2.16 m, suggesting it underestimates tree height. When compared to other professional MLS devices, the LC-MLS exhibits a higher RMSE% in TH measurements (12.97%), indicating less accuracy in tree height estimation.