Next Article in Journal
Mapping Spatial Drivers of Predicted Active Fires Kernel Density with Geographically Weighted Regression in Mexico
Next Article in Special Issue
Effects of Highway Construction on Landscape Patterns, Ecosystem Service Value, Habitat Connectivity and Their Associations in Zhejiang, China
Previous Article in Journal
Changes in Potentially Suitable Habitats and Priority Conservation Zones of Prunus sibirica L. in China Under Climate Change
Previous Article in Special Issue
Internet of Things (IoT)-Based Applications in Smart Forestry: A Conceptual and Technological Analysis
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 17
Issue 2
10.3390/f17020265
forests-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

ALS and SfM Field Data Survey as a Basis of Forest Road Design

by
Ivica Papa
,
Luka Hodak
*,
Maja Popović
,
Andreja Đuka
,
Tibor Pentek
and
Mihael Lovrinčević
*
Department of Forest Engineering, Faculty of Forestry and Wood Technology, University of Zagreb, Svetošimunska Cesta 23, 10000 Zagreb, Croatia
*
Authors to whom correspondence should be addressed.
Forests 2026, 17(2), 265; https://doi.org/10.3390/f17020265
Submission received: 15 December 2025 / Revised: 12 February 2026 / Accepted: 14 February 2026 / Published: 16 February 2026
(This article belongs to the Special Issue Advances in Remote Sensing for Forest Resource Monitoring and Management)
Browse Figures
Versions Notes

Abstract

Field data of high accuracy and precision is the basis for creating the high-quality design of a forest road. In this study, three survey methods for collecting field data were tested: ALS UAV, LiDAR data of the Republic of Croatia, collected by airplane, and UAV SfM. A total of three detailed forest road projects were created based on the collected data. The designed forest roads had the same horizontal and vertical development, thus eliminating the human factor from the design process. Four important forest road parameters were tested: earthwork cut and fill volume, cross-terrain slope, and carriageway value. No significant statistical difference was found for any of the tested parameters between designs. The design based on ALS data had a total number of earthworks of 1026.03 m3, the amount was 1449.56 m3 for SfM design, and the number of earthworks for the State Geodetic Administration LiDAR data was 889.02 m3. The calculated amount of cut volume was significantly affected by the error of the carriageway value for the State Geodetic Administration LiDAR data-based design. The results indicate the possibility of using all used methods on terrain with a moderate slope, but there is a need for further testing on different terrain slope classes.

1. Introduction

The need for further increasing forest road density varies, not only between countries, but also between smaller areas [1]. Countries with generally high road accessibility have an increasing challenge of high cost of maintenance and do not need to build new forest roads [1,2]. On the other hand, many countries with low road density still have to plan, design, and build forest roads in order to fulfill all forest management tasks [1,2,3,4,5]. To establish an optimal forest road network, it is necessary to comprehensively assess the proposed routes from economic and ecological standpoints through parameters such as construction and maintenance costs, the impact on the costs of logging and wood transport, social benefits, and the risk of negative environmental impact [6,7]. Proper planning of the optimal forest road network can result in a construction, reconstruction, and maintenance cost reduction of 47% [8]. According to Rayan [9], the optimal forest road network establishment consists of four work phases: planning, design, construction with supervision, and maintenance. Field data surveys, as a crucial part of forest road design, can be conducted using classic (theodolites and leveling instruments) and modern (total stations and GNSS devices) surveying methods [10], while Lovrinčević et al. [11] also mention the experimental field surveying method (use of photogrammetry and LiDAR sensors), which is increasingly being used in forestry. As the labor shortage is a general problem in Europe [12,13], and the design process itself is demanding financially and in terms of human labor, it is necessary to develop a sufficiently accurate, precise, and fast method of field data surveying.
Recently, the forestry profession has recognized the importance of using unmanned aerial vehicles (UAVs) in the remote sensing of forested areas [14,15,16,17,18]. One of the reasons for the increased use of unmanned aerial vehicles can be seen not only in the rationalization of survey time [19], but also in facing the challenge of labor shortages [12,13]. UAVs are a more cost-effective and faster system compared to traditional aerial technologies [20]. Data recorded by laser scanning (LiDAR sensors) are used in forestry to estimate tree height, diameter, and volume and to determine the assortment of the structure [21,22,23]. On the other hand, structure from motion (SfM) is emerging as a cheaper option, with the central tasks being photo activation, referencing, measurement, and interpretation [24]. As mentioned, SfM is a cheaper alternative for forest structure assessment, but it is less capable than LiDAR systems for measuring ground elevations [25,26], as SfM is limited to the reconstruction of surfaces visible in aerial photos, providing ground information only where large vegetation gaps exist [27]. When it comes to the precision and accuracy of different UAVSfM surveys, they can be affected by various parameters, such as flight altitude [28], photo overlay [29], and correction methods [30,31,32] for SfM, while Klass-Witt and Emeis [33] list five factors that can affect errors during LiDAR surveying: orographic complexity, terrain roughness and vegetation, atmospheric stability, measurement height, and half-cone opening angle.
The root mean square error (RMSE) for the SfM survey method differs in different forest conditions, from 2.3 to 123 cm horizontally and 3 to 79.4 cm vertically [29,31,32,34,35,36,37,38]. Airborne LiDAR scanning (ALS) with UAVs achieves a spatial accuracy of 4.9 cm to 12.1 cm horizontally and 1.7 cm to 4.1 cm vertically and is highly dependent on the point cloud density and the sensors themselves [11,39,40,41]. On the other hand, the spatial accuracy of data collected by airplane ALS ranges between 12 and 50 cm [42,43,44].
An important factor that can significantly affect construction costs during the optimal forest road network establishment is the earthwork volume needed for subgrade construction [45]. The earthwork amount differences between different survey methods are addressed in Refs. [46,47,48,49], but there is a need for further research on this topic.
Although the sensors used in this study were tested for accuracy and precision in forest conditions, it is evident that their field applicability in the process of forest road design (as well as maintenance) has not been sufficiently investigated. The diversity of forest conditions and forest road design methods makes it imperative to test these sensors in different fields and forest conditions before operative use. In this study, the sensors were tested in deciduous forests on terrain with moderate slopes. Data collection was performed using an unmanned aerial vehicle equipped with an RGB camera, an unmanned aerial vehicle equipped with a LiDAR sensor, and LiDAR data from the State Geodetic Administration (SGA) collected through aerial photography using an aircraft over the entire territory of the Republic of Croatia. The data collected using the above methods served as the basis for the creation of the detailed designs of a forest road, on the basis of which the number of earthworks, the carriageway value, and cross slopes were calculated and compared.

2. Materials and Methods

2.1. Study Area

This research was conducted in the Forest Training and Research Center Zagreb (FTRC), management unit “Dotrščina”, section 15a (Figure 1). The mentioned management unit is managed by the Faculty of Forestry and Wood Technology, University of Zagreb. The area of the researched section (15a) is 14.58 ha. The phytocenosis in the researched section is Epimedio-Carpinetum betuli (Horvat 1938) Borhidi 1963. The tree species that dominate the researched area are Fagus sylvatica L., Quercus petraea L., and Carpinus betulus L. The altitude of the researched section ranges between 180 and 220 m above sea level with varying slopes mostly between 8% and 25%. The average slope of the research area is 21.9%.

2.2. Data Collection and Processing

Field data collection (survey) as part of the forest road design phase was carried out using experimental surveying methods. First, a zero line was planned on a 1:5000 scale contour map. The maximum used longitudinal slopes of the zero line did not exceed 10%, which is in accordance with the technical requirements for forest roads in the Republic of Croatia [50,51]. Surveying was conducted using two different UAVs equipped with different sensors, one equipped with an RGB camera (DJI, Shenzhen, China) (UAVSfM), while the other was equipped with a LiDAR sensor (DJI, Shenzhen, China) (ALSUAV). Data from the airplane ALS survey was provided by the State Geodetic Administration (ALSDGU) upon personal request. The axis polygon (horizontal alignment of forest roads), which was fitted to the zero line in the field, was recorded with a GNSS device (Stonex S900A, Stonex, Paderno Dugnano, Italy) in RTK mode with the CROatian Positioning System (CROPOS) correction database [52]. Data collection was carried out by two people.

2.2.1. Field Data Survey Using the DJI Mavic 3 Enterprise

A DJI Mavic 3 Enterprise UAV (DJI, Shenzhen, China) equipped with a high-resolution camera was used to collect SfM data of the research area. The technical specifications of the used UAV are shown in Table 1. The mission of the flight operation was created in the DJI Pilot 2 application. The survey was conducted outside of the vegetation period. The flight height was 80 m with a front and side overlap of 80%, with the terrain follow function turned on. The recording time was 23:33 min (Figure A1A). The option of a camera mechanical shutter was used, which avoided blurring during the UAV movement and enabled faster surveying [53,54]. A total of 722 photos were taken for an area of 26 ha. Post-processing kinematic (PPK) processing of the captured photos was performed in Emlid Studio (v 1.7) and DJI Terra (v 4.2.13). For the PPK, a ground point was recorded with the Emlid Reach 2+ device (Emlid, China) (1.50 h of recording) with the CROPOS GNSS station (Trimble Alloy) used as stationary base station. The distance to said station was 8 km. Photogrammetric analysis was conducted in Pix4D mapper software (v 4.8.4). Automatic point cloud classification was used. The time required for office (PPK, point classification and digital terrain model (DTM) creation) data processing was 360 min (Figure 2). Ground points of the tested methods after point cloud classification are shown in Figure A2. The pixel size of the created DTM was 10.75 cm.

2.2.2. Field Data Survey Using the DJI Matrice 350 RTK with DJI Zenmuse L2 Sensor

A DJI Matrice 350 RTK (DJI, Shenzhen, China) drone equipped with a DJI Zenmuse L2 sensor was used for the ALSUAV field survey. The technical specifications of the drone and sensors used are shown in Table 1. The flight operation was created in the DJI Pilot 2 application. The survey was carried out outside the vegetation period. The flight height was 100 m with the terrain follow function turned on. The side overlap of the LiDAR scan was 50%. The density of the point cloud was 273 points/m2. The flight time was 18:30 min (Figure A1B). The total recorded area was 24 ha. The post-processing kinematic method of the point cloud was performed in the Emlid Studio (v 1.7) and DJI Terra (v 4.2.13) programs based on ground points recorded with the Emlid Reach 2+ device (1.50 h of recording) and the GNSS station of the CROPOS system. Point cloud classification and DTM creation were conducted in LiDAR360 software (v 7). Automatic gound point cloud classification resulted in 10 points/m2. The time required for office data processing (PPK, point classification, and DTM creation) was 130 min (Figure 2). A UAV with a LiDAR sensor has better coverage, or greater penetration, compared to photogrammetric surveying, but with slightly lower altitude accuracy [55,56]. During the survey, there were still leaves on the trees. For this reason, the DJI matrix 350 with an L2 sensor was used as the reference surveying method.

2.2.3. State Geodetic Administration LiDAR Data of the Republic of Croatia

The SGA of the Republic of Croatia, for the purposes of the project “Multisensor Aerial Survey of the Republic of Croatia for the purposes of assessing disaster risk reduction”, carried out aerial surveys of the entire territory of the Republic of Croatia in 2022. The data from the ALS survey was the basis for creating and updating the DTM and the digital surface model (DSM). During the collection of LiDAR data, certain environmental conditions were respected that are important from a forestry point of view: snow-free conditions, leaf off (surveying during the dormant period of vegetation), and no flooded survey areas. The minimum required point density for areas covered by forest (deciduous and coniferous) was 4 points/m2 with an altitude accuracy of ±0.1 m and a positional accuracy of ±0.2 m (Figure 2) [57].
The ground points of tested methods are shown in Figure A2.

2.3. Detailed Forest Road Design and Analysis

Based on the collected and provided data, the detailed forest road designs were created in the software RoadEng (v. 10.0.664.0). The horizontal and vertical development of the designed forest road routes were identical for each data set as well as the normal cross-section parameters (Figure 3). In total, 19 circular horizontal curves were designed and rounded with radii in the range of 20 m to 6000 m. The total length of the designed forest roads was 854.60 m (08 + 54.60 hm). The vertical alignment was defined by nine vertical alignment breaks (IPV) rounded by parabolas whose parabolic value (K) ranged from 8 to 600. The forest road wearing course width (without widening) was 3.50 m, with 0.50-wide shoulders. The height difference overcome between the starting and ending points of the forest road was 19.99 m. The detailed forest road projects were developed in accordance with the technical conditions for forest roads in the Republic of Croatia [50,51].
From completed detailed forest road designs, the following parameter values were read and tested (Figure 3):
  • Earthwork volumes (fill and cut).
  • Carriageway value (wearing course height (in centerline) reduced by the ground height).
  • Cross-terrain slope.
The values of the tested parameters were read at the points (cross-sections) of interest. These places were defined at each horizontal curve beginning (BC), curve end (EC), in the middle of the curve (MC), and 10 m after EC position on sections where the distance between the EC of the previous curve and the BC of the next curve was greater than 15 m. In total, 84 of these cross-sections of interest were used.
A database was created in MS Excel. Based on the differences between ALSUAV (reference method) and other field data collection methods, RMSE for the tested forest road parameters was calculated using the formula:
R M S E v d = ∑ v d 2 n
where
  • vd is the value difference, and
  • n is the number of observations.
Statistica (v.14.0.0.15) software was used for further statistical analysis. Data normality was tested by Kolomogorov–Smirnov and Lilliefors tests. To determine the differences between tested parameters, one-way analysis of variance (ANOVA) followed by Tukey’s HSD post hoc test were used. Correlation and regression analyses were conducted to determine the relationship and connection between the measurement parameters.

3. Results

The obtained results are presented in four parts by the tested parameters and the interaction between the errors of the calculated earthwork volumes and errors of the carriageway values.

3.1. Earthwork Volumes

3.1.1. Cut Volume

Detailed forest road designs based on ALSUAV surveying, the reference method in this research, resulted in 1511.99 m3 of cut volume. A lower total cut volume was calculated for the ALSDGU design (1470.55 m3) while a detailed forest road design based on UAVSfM surveying overestimated the cut volume by 313.48 m3 (1825.47 m3). The RMSE of the excavation volume for designs based on ALSDGU and UAVSfM are 4.03 m3 and 5.82 m3, respectively (Table 2). Designs based on UAVSfM, on most sections of the forest road route, overestimated the calculated cut volumes, while the design based on ALSDGU underestimated it. Although the ALSDGU design on most cross-sections underestimated the cut volumes, the largest differences compared to the ALSUAV were recorded between cross-sections 69 and 70 (+10.27 m3) and between cross-sections 66 and 67 (+9.33 m3), where the ALSDGU design resulted in a higher cut volume. Cross-terrain slopes for the mentioned cross-sections were 14.09% and 18.41%, respectively. On the other hand, the largest calculated error of cut volumes of the UAVSfM design were recorded between sections 57 and 58 in the amount of −13.21 m3, and between cross-sections 1 and 2 in the amount of 12.39 m3. This and other major errors in calculated cut volumes of the tested method designs were not recorded on the same parts of the designed route. The ALSDGU-based design showed higher cut volumes and higher errors from cross-section 62 to the route end (cross-section 84), while the UAVSfM design had higher cut volumes and higher errors between cross-sections 57 and 77. The largest recorded difference between the ALSDGU and UAVSfM designs was determined between cross-sections 57 and 58, where the difference in cut volume was 18.34 m3.
ANOVA did not reveal any statistically significant difference in the excavation volumes between the investigated methods (Figure 4). A large data scatter was observed.

3.1.2. Fill Volume

Compared to the ALSUAV design (485.96 m3), the ALSDGU design calculated 19.67% more of fill volume (581.54 m3), while the UAVSfM design underestimated the fill volume by 22.65% (375.89 m3). The recorded RMSE values of the fill volume were lower than the RMSE values of the cut volume: 2.04 m3 for the ALSDGU design and 2.92 m3 for the UAVSfM design (Table 3). In contrast to the cut volume, the ALSDGU-based design overestimated the calculated fill volumes, while the UAVSfM-based design underestimated them. The ALSDGU design recorded lower fill volumes compared to the ALSUAV design from cross-sections 62 to 84, where, for cut volume, ALSDGU recorded an overestimation. Between single cross-sections, the largest error was recorded between cross-sections 61 to 62 using the ALSDGU design, in the amount of +5.62 m3. Between the ALSUAV and UAVSfM designs, the largest differences were found between cross-sections 49 and 50 (−3.22 m3) and between cross-sections 61 and 62 (+10.01 m3). The highest error between the ALSDGU and UAVSfM designs was recorded between cross-sections 17 and 18, where the difference amounted to 12.98 m3.
As with cut volume, ANOVA did not find any statistically significant difference in the embankment volume between designs based on the investigated field data collection methods (Figure 5). The observed trends also suggest a large scatter in the data.
The UAVSfM design calculated the highest value of accumulated earthwork volume along the length of the project alignment (1449.56 m3), which was 41.28% more than the ALSUAV design (1026.03 m3). A smaller difference was recorded for ALSDGU (−13.35%, 889.02 m3). From cross-sections 1 to 61, the mass haul difference between ALSUAV and ALSDGU increased, and after which the difference decreased. The difference between ALSUAV and UAVSfM increased uniformly from the first cross-section (Figure 6).

3.2. Cross-Terrain Slope

The average cross slope between the methods ranged between 15.62% and 15.80% (Table 4). The ALSUAV design recorded the highest cross-terrain slope for a single cross-section (66), while the ALSDGU design achieved the lowest maximum (20). In cross-section 20, the ALSDGU design recorded 22.76% and UAVSfM recorded 22.55% cross-terrain slopes. The biggest difference was found at cross-section 66, where the cross-terrain slope of the ALSUAV design was 16%, while the ALSDGU and UAVSfM designs calculated an identical cross-terrain slope of 21%. UAVSfM and ALSDGU, on most cross-sections, overestimated cross-terrain slopes, and ANOVA did not show statistically significant differences for cross-terrain slopes between the tested methods (Figure 7).

3.3. Carriageway Value

Compared to the previously analyzed parameters, the carriageway value showed greater oscillations. The values ranged from −0.71 m (profile 2) to 1.23 m (profile 61) for the ALSUAV design, −0.68 m (profile 2) to 1.19 m (profile 61) for the design based on ALSDGU data, and from −0.66 m (profile 2) to 1.01 m (profile 61) for the UAVSfM design. For both designs, based on ALSDGU and UAVSfM, the highest error was carriageway value overestimation, as seen on cross-section 49 for the ALSDGU design (+0.21 m), and on cross-section 83 for UAVSfM (+0.41 m). The ALSDGU-based design overestimated carriageway value on the first 61 cross-sections, while UAVSfM underestimated carriageway value on 81% of the analyzed cross-sections without a clear pattern. This was reflected in the carriageway value mean values, which were higher for the ALSDGU design, that is, lower for the UAVSfM design (Table 5). Although larger deviations were found, especially for the UAVSfM design, the conducted ANOVA did not find statistically significant differences between the methods for carriageway values (Figure 8).
A moderate and negative correlation was determined between cut error and cross-terrain slope error for the ALSDGU-based design (r = 0.42) (Figure 9A). It was found that the differences in cross-terrain slope accounted for 18.1% of the variability in the estimated differences in excavation volume (R2 = 0.17). For the UAVSfM design, a moderate correlation was determined between the cut error and cross-terrain slope error (r = 0.25), with a coefficient of determination (R2) of 0.06 (p = 0.00) (Figure 9B).
The difference in the calculated fill volumes showed a moderate positive correlation with the differences in cross-terrain slopes (r = 0.54) (Figure 10A) for ALSDGU. A regression analysis determined that the model explains 29.03% of the variability (R2 = 0.28). A weak and negative correlation was determined between fill volume and cross-terrain slope errors for the UAVSfM design (r = 0.13) (Figure 10B). The coefficient of determination was 0.7.
A high and negative correlation was determined between cut error and carriageway value error for the ALSDGU-based design (r = 0.79) (Figure 11A). The results of the regression analysis confirmed a significant relationship between the difference in cut volume and the carriageway value difference. It was found that the differences in working elevation explained 62.5% of the variability in the estimated differences in excavation volume (R2 = 0.63). For the UAVSfM design, a moderate and negative correlation was determined between the cut error and carriageway value error (r = 0.48), with a coefficient of determination (R2) of 0.23 (Figure 11B).
The difference in the calculated fill volumes showed a moderate positive correlation with the difference in carriageway value (r = 0.62; p < 0.001) (Figure 12A) for ALSDGU. A regression analysis determined that the model explains 38.4% of the variability (R2 = 0.38). A weak and positive correlation was determined between fill volume and carriageway value errors for the UAVSfM design (r = 0.30) (Figure 12B). The coefficient of determination was 0.08.

4. Discussion

As certain countries still need to increase forest road accessibility [1,2], and as the field data survey process with a classic surveying method (equipment) is time-consuming [58], it is necessary to test survey equipment that will enable fast and accurate data collection. The conducted research tested three ways of collecting field data, based on which the process of designing a forest road was carried out in the form of airborne LiDAR scanning by UAV (ALSUAV) and by airplane (ALSDGU), as well as structure from motion by UAV (UAVSfM). All detailed forest road designs had the same horizontal and vertical alignment and road parameters, thus eliminating the designer’s influence on the research results.
The ALSDGU-based forest road design had the tendency to overestimate fill volumes on the first 62 tested cross-sections (3/4 of the forest road length). After the 62nd cross-section, the mentioned design overestimated cut volumes. In the 62nd cross-section, the road gradient changed from positive to negative. In total, more fill volume and less cut volume were calculated for the ALSDGU-based design compared to ALSUAV. This is in accordance with other studies, where airplane ALS underestimated cut volumes and overestimated fill volumes compared to the reference survey methods [47]. Since the volume of earthwork is directly dependent on carriageway value [59] (by increasing the carriageway value, the road rises higher above the ground), and—for the ALSDGU design, higher average carriageway values were calculated compared to ALSUAV (Figure 8)—this phenomenon of higher calculated fill volumes of the ALSDGU design should not be surprising. One of the possible reasons for higher carriageway value and fill volumes is the lower Z-values of ALSDGU DTM, as shown in the study by Lovrinčević et al. [11]. This phenomenon of lower Z-values for ALS airplane LiDAR data was also determined by Hruza et al. [60] in their forest road wearing course damage study, where they tested the Czech Republic national DTM based on airplane ALS data.
The UAVSfM-based design overestimated cut volumes and underestimated fill volumes. In the available literature [20,61,62], it is not possible to find a trend related to the overestimation or underestimation of the volume of earthworks. SfM is not a direct measurement but a computational technique, and multiple factors affect the final measurement result when compared to LiDAR measurements [63]. This may be one of the reasons why it is not possible to determine the trend in the occurrence of a surplus/shortage of earthwork volumes. The average carriageway value was the lowest for the UAVSfM design (Figure 8), which certainly contributed to the overestimation of the excavation volume of this method. One possible reason for this is the higher Z-values of the UAVSfM DTM caused by the nature of the automatic classification of the point cloud or perhaps even the leaf layer that was on the ground during the survey. In some places, the automatic classification omitted points that should have been classified as ground. This resulted in larger holes in the point cloud used for DTM creation (Figure A2). Part of the error caused by this oversight of automatic classification could be addressed by the manual classification of the point cloud, or by additional control of the automatic classification. On the other hand, if the error is caused by leaf litter, manual classification would not make sense. In such a case, it is necessary to find a correction method. This will be addressed in future research.
The moderate correlation between cross-terrain slope and earthwork errors for ALSDTM confirms that higher slopes can cause LiDAR DTM errors [64,65,66], but further research is necessary to draw a firm conclusion. On the other hand, only a weak correlation was determined between the cross-terrain slope and earthwork errors of the UAVSfM-based design.
The negative correlation between the carriageway value error and the cut volume error, that is, the positive correlation between the carriageway value error and fill volume error, can be explained by the nature of the carriageway value calculation: higher carriageway values (and higher errors) represent a higher height of the road in relation to the ground (or to the reference design). Based on this, it can also be assumed that with increasing cut/fill volumes, higher errors may also occur, especially for the ALSDGU-based design.
Finally, although no time study was performed, the time spent on data acquisition during the flight missions was similar for the ALSUAV and SfM surveys. On the other hand, the data processing time is much higher for SfM measurements. Of course, the time required for data processing and DTM creation strongly depends on the performance of the computer on which it is performed. The SfM survey requires lower survey costs compared to the ALSUAV surveys, especially compared to the airplane ALS survey if the data is not available at the national/regional level. Whichever measurement is used, it is necessary to train the personnel, both for recording and for data processing.

5. Conclusions

Based on the obtained results, the authors draw the following conclusions:
  • The ALSDGU and UAVSfM survey methods provide comparable and sufficiently accurate field data necessary for forest road design on terrain with moderate cross-terrain slopes and simpler reliefs.
  • Airplane ALS survey data, when used for forest road design, has a tendency to underestimate cut volume and overestimate fill volume.
  • The UAVSfM-based design, in this study, had a tendency to overestimate cut volume and underestimate fill volume. As there is no clear trend of this phenomenon in other studies, it cannot be claimed that this is a constant phenomenon.
  • The carriageway value error strongly affects the calculated cut volume and moderately affects the fill volume of designs based on ALSDGU data, while for UAVSfM-based design, there is a moderate and weak correlation between carriageway value error and cut and fill volume error.
  • The cross-terrain slope is moderately related to the earthworks error of the ALSDGU-based design and weakly related to the volume error of the UAVSfM design.
  • It should be noted that the results obtained are in a forest with a moderate slope and a medium-developed understory. Further testing in different field conditions is necessary to draw strong conclusions about the usability of the tested conditions.
  • It is necessary to eliminate the human factor as a source of error in testing different measurement systems for forest road design, such as horizontal or vertical road development.

Author Contributions

Conceptualization, I.P. and L.H.; methodology, L.H. and M.L.; software, M.L.; validation, T.P. and A.Đ.; formal analysis, M.P.; investigation, L.H.; resources, A.Đ.; data curation, M.L.; writing—original draft preparation, L.H. and M.L.; writing—review and editing, I.P.; visualization, L.H. and M.L.; supervision, I.P.; project administration, T.P.; funding acquisition, A.Đ. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the Croatian Science Foundation under project number HRZZ-UIP-2019-04-7766.

Data Availability Statement

The data is contained within this article.

Acknowledgments

The authors would like to thank the University of Zagreb Faculty of Forestry and Wood Technology for their support during this research.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Forests 17 00265 g0a1
Figure A1. Flight route: (A) UAVSfM, (B) ALSUAV.
Figure A1. Flight route: (A) UAVSfM, (B) ALSUAV.
Forests 17 00265 g0a1
Forests 17 00265 g0a2
Figure A2. Ground points: (A) ALSUAV, (B) ALSDGU, (C) UAVSfM.
Figure A2. Ground points: (A) ALSUAV, (B) ALSDGU, (C) UAVSfM.
Forests 17 00265 g0a2

References

  1. Đuka, A.; Papa, I.; Proto, A.R.; Latterini, F.; Mederski, P.S.; Borz, S.A.; Iordache, E.; Sokolović, D.; Karšik, A.; Stojnić, D. Forest Accessibility and Road Network Density: A Global Overview with a Special Focus on Europe. Curr. For. Rep. 2025, 11, 26. [Google Scholar] [CrossRef]
  2. Lyons, C.K.; Borz, S.A.; Harvey, C.; Ramantswana, M.; Sakai, H.; Visser, R. Forest roads: Regional perspectives from around the world. Int. J. For. Eng. 2022, 34, 190–203. [Google Scholar] [CrossRef]
  3. Hodić, I.; Jurušić, Z. Analysis of Primary Opennes of Forest Managed by Hrvatske Šume Ltd. as Basis for Disigning of Future Policy Forest Roads Construction. Šumarski List 2011, 135, 487–499. Available online: https://hrcak.srce.hr/clanak/111444 (accessed on 10 November 2025).
  4. Petković, V.; Potočnik, I. Planning forest road network in natural forest areas: A case study in northern Bosnia and Herzegovina. Crojfe J. For. Eng. 2018, 39, 45–56. Available online: https://hrcak.srce.hr/193550 (accessed on 8 November 2025).
  5. Hapa, M.; Dinca, L.; Crisan, V. Forests along roads-a case of stability, resilience and biodiversity. Present Environ. Sustain. Dev. 2023, 17, 257–273. [Google Scholar] [CrossRef]
  6. Hasegawa, H.; Sujaswara, A.A.; Kanemoto, T.; Tsubota, K. Possibilities of Using UAV for Estimating Earthwork Volumes during Process of Repairing a Small-Scale Forest Road, Case Study from Kyoto Prefecture, Japan. Forests 2023, 14, 677. [Google Scholar] [CrossRef]
  7. Dražić, S.; Danilović, M.; Ristić, R.; Stojnić, D.; Antonić, S. Evaluation of morphometric terrain parameters and their influence on determining optimal density of primary forest road network. Crojfe J. For. Eng. 2023, 44, 301–312. [Google Scholar] [CrossRef]
  8. Simões, D.; Cavalcante, F.S.; Lima, R.C.A.; Rocha, Q.S.; Pereira, G.; Miyajima, R.H. Optimal Forest Road Density as Decision-Making Factor in Wood Extraction. Forests 2022, 13, 1703. [Google Scholar] [CrossRef]
  9. Ryan, T.; Phillips, H.; Ramsay, J.; Dempsey, J. Forest Road Manual: Guidelines for the Design, Construction and Management of Forest Roads, 1st ed.; COFORD: Dublin, Ireland, 2004; Available online: https://www.unirc.it/documentazione/materiale_didattico/598_2007_39_832.pdf (accessed on 15 November 2025).
  10. Papa, I.; Pentek, T.; Janeš, D.; Šerić, T.; Vusić, D.; Đuka, A. Usporedba podataka prikupljenih različitim metodama terenske izmjere pri rekonstrukciji šumske ceste. Nova Meh. Šum. 2017, 38, 1–14. Available online: https://hrcak.srce.hr/192251 (accessed on 10 November 2025).
  11. Lovrinčević, M.; Papa, I.; Janeš, D.; Hodak, L.; Pentek, T.; Đuka, A. New Possibilities of Field Data Survey in Forest Road Design. Sensors 2025, 25, 4192. [Google Scholar] [CrossRef]
  12. Šporčić, M.; Landekić, M.; Šušnjar, M.; Pandur, Z.; Bačić, M.; Mijoč, D. Shortage of labour force in forestry of Bosnia and Herzegovina–forestry experts’ opinions on recruiting and retaining forestry workers. Crojfe J. For. Eng. 2024, 45, 183–198. [Google Scholar] [CrossRef]
  13. Handel, M.J. Labor shortages: What is the problem. Intereconomics 2024, 59, 136–142. [Google Scholar] [CrossRef]
  14. Tang, L.; Shao, G. Drone remote sensing for forestry research and practices. J. For. Res. 2015, 26, 791–797. [Google Scholar] [CrossRef]
  15. Guimarães, N.; Pádua, L.; Marques, P.; Silva, N.; Peres, E.; Sousa, J.J. Forestry Remote Sensing from Unmanned Aerial Vehicles: A Review Focusing on the Data, Processing and Potentialities. Remote Sens. 2020, 12, 1046. [Google Scholar] [CrossRef]
  16. Dainelli, R.; Toscano, P.; Di Gennaro, S.F.; Matese, A. Recent Advances in Unmanned Aerial Vehicles Forest Remote Sensing—A Systematic Review. Part II: Research Applications. Forests 2021, 12, 397. [Google Scholar] [CrossRef]
  17. Açıl, A.; Aydın, A.; Eker, R.; Duyar, A. Use of UAV data and HEC-RAS model for dimensioning of hydraulic structures on forest roads. Crojfe J. For. Eng. 2023, 44, 171–188. [Google Scholar] [CrossRef]
  18. Yrttimaa, T.; Matsuzaki, S.; Kankare, V.; Junttila, S.; Saarinen, N.; Kukko, A.; Vastaranta, M. Assessing forest traversability for autonomous mobile systems using close-range airborne laser scanning. Crojfe J. For. Eng. 2024, 45, 169–182. [Google Scholar] [CrossRef]
  19. Woellner, R.; Wagner, T.C. Saving species, time and money: Application of unmanned aerial vehicles (UAVs) for monitoring of an endangered alpine river specialist in a small nature reserve. Biol. Conserv. 2019, 233, 162–175. [Google Scholar] [CrossRef]
  20. Taş, İ.; Kaska, M.S.; Akay, A.E. Assessment of Using UAV Photogrammetry Based DEM and Ground-Measurement Based DEM in Computer-Assisted Forest Road Design. Eur. J. For. Eng. 2023, 9, 1–9. [Google Scholar] [CrossRef]
  21. Xu, D.; Wang, H.; Xu, W.; Luan, Z.; Xu, X. LiDAR Applications to Estimate Forest Biomass at Individual Tree Scale: Opportunities, Challenges and Future Perspectives. Forests 2021, 12, 550. [Google Scholar] [CrossRef]
  22. Sun, Y.; Jin, X.; Pukkala, T.; Li, F. Predicting Individual Tree Diameter of Larch (Larix olgensis) from UAV-LiDAR Data Using Six Different Algorithms. Remote Sens. 2022, 14, 1125. [Google Scholar] [CrossRef]
  23. Alvites, C.; Marchetti, M.; Lasserre, B.; Santopuoli, G. LiDAR as a Tool for Assessing Timber Assortments: A Systematic Literature Review. Remote Sens. 2022, 14, 4466. [Google Scholar] [CrossRef]
  24. Honkavaara, E.; Arbiol, R.; Markelin, L.; Martinez, L.; Cramer, M.; Bovet, S.; Chandelier, L.; Ilves, R.; Klonus, S.; Marshal, P.; et al. Digital Airborne Photogrammetry—A New Tool for Quantitative Remote Sensing?—A State-of-the-Art Review On Radiometric Aspects of Digital Photogrammetric Images. Remote Sens. 2009, 1, 577–605. [Google Scholar] [CrossRef]
  25. White, J.C.; Wulder, M.A.; Vastaranta, M.; Coops, N.C.; Pitt, D.; Woods, M. The Utility of Image-Based Point Clouds for Forest Inventory: A Comparison with Airborne Laser Scanning. Forests 2013, 4, 518–536. [Google Scholar] [CrossRef]
  26. Díaz, G.M.; Mohr-Bell, D.; Garrett, M.; Muñoz, L.; Lencinas, J.D. Customizing unmanned aircraft systems to reduce forest inventory costs: Can oblique images substantially improve the 3D reconstruction of the canopy? Int. J. Remote Sen. 2020, 41, 3480–3510. [Google Scholar] [CrossRef]
  27. Iglhaut, J.; Cabo, C.; Puliti, S.; Piermattei, L.; O’Connor, J.; Rosette, J. Structure from motion photogrammetry in forestry: A review. Curr. For. Rep. 2019, 5, 155–168. [Google Scholar] [CrossRef]
  28. Swayze, N.C.; Tinkham, W.T.; Vogeler, J.C.; Hudak, A.T. Influence of flight parameters on UAS-based monitoring of tree height, diameter, and density. Remote Sens. Environ. 2021, 263, 112540. [Google Scholar] [CrossRef]
  29. Dhruva, A.; Hartley, R.J.L.; Redpath, T.A.N.; Estarija, H.J.C.; Cajes, D.; Massam, P.D. Effective UAV Photogrammetry for Forest Management: New Insights on Side Overlap and Flight Parameters. Forests 2024, 15, 2135. [Google Scholar] [CrossRef]
  30. Tomaštík, J.; Mokroš, M.; Saloň, Š.; Chudý, F.; Tunák, D. Accuracy of Photogrammetric UAV-Based Point Clouds under Conditions of Partially-Open Forest Canopy. Forests 2017, 8, 151. [Google Scholar] [CrossRef]
  31. Zhang, H.; Aldana-Jague, E.; Clapuyt, F.; Wilken, F.; Vanacker, V.; Van Oost, K. Evaluating the potential of post-processing kinematic (PPK) georeferencing for UAV-based structure-from-motion (SfM) photogrammetry and surface change detection. Earth Surf. Dyn. 2019, 7, 807–827. [Google Scholar] [CrossRef]
  32. Đuka, A.; Tomljanović, K.; Franjević, M.; Janeš, D.; Žarković, I.; Papa, I. Application and Accuracy of Unmanned Aerial Survey Imagery after Salvage Logging in Different Terrain Conditions. Forests 2022, 13, 2054. [Google Scholar] [CrossRef]
  33. Klaas-Witt, T.; Emeis, S. The five main influencing factors for lidar errors in complex terrain. Wind. Energy Sci. 2022, 7, 413–431. [Google Scholar] [CrossRef]
  34. Stöcker, C.; Nex, F.; Koeva, M.; Gerke, M. Quality assessment of combined IMU/GNSS data for direct georeferencing in the context of UAV-based mapping. ISPRS 2017, 42, 355–361. [Google Scholar] [CrossRef]
  35. Tomaštík, J.; Mokroš, M.; Surový, P.; Grznárová, A.; Merganič, J. UAV RTK/PPK Method—An Optimal Solution for Mapping Inaccessible Forested Areas? Remote Sens. 2019, 11, 721. [Google Scholar] [CrossRef]
  36. Padró, J.C.; Muñoz, F.J.; Planas, J.; Pons, X. Comparison of four UAV georeferencing methods for environmental monitoring purposes focusing on the combined use with airborne and satellite remote sensing platforms. Int. J. Appl. Earth Obs. 2019, 75, 130–140. [Google Scholar] [CrossRef]
  37. Miller, Z.M.; Hupy, J.; Chandrasekaran, A.; Shao, G.; Fei, S. Application of postprocessing kinematic methods with UAS remote sensing in forest ecosystems. J. For. 2021, 119, 454–466. [Google Scholar] [CrossRef]
  38. Papa, I.; Popović, M.; Hodak, L.; Đuka, A.; Pentek, T.; Hikl, M.; Lovrinčević, M. Water and Vegetation as a Source of UAV Forest Road Cross-Section Survey Error. Forests 2025, 16, 507. [Google Scholar] [CrossRef]
  39. Craven, M.; Wing, M.G. Applying airborne LiDAR for forested road geomatics. Scand. J. For. Res. 2014, 29, 174–182. [Google Scholar] [CrossRef]
  40. Azizi, Z.; Najafi, A.; Sadeghian, S. Forest road detection using LiDAR data. J. For. Res. 2014, 25, 975–980. [Google Scholar] [CrossRef]
  41. Karjalainen, T.; Karjalainen, V.; Waga, K.; Tokola, T. Predicting the roadway width of forest roads by means of airborne laser scanning. Int. J. Appl. Earth Obs. 2024, 133, 104109. [Google Scholar] [CrossRef]
  42. Schmelz, W.J.; Psuty, N.P. Quantification of airborne lidar accuracy in coastal dunes (Fire island, New York). Photogramm. Eng. Remote Sens. 2019, 85, 133–144. [Google Scholar] [CrossRef]
  43. Liu, X. A new framework for accuracy assessment of LIDAR-derived digital elevation models. ISPRS 2022, 4, 67–73. [Google Scholar] [CrossRef]
  44. Elaksher, A.; Ali, T.; Alharthy, A. A Quantitative Assessment of LIDAR Data Accuracy. Remote Sens. 2023, 15, 442. [Google Scholar] [CrossRef]
  45. Park, H.C.; Rachmawati, T.S.N.; Kim, S. UAV-Based High-Rise Buildings Earthwork Monitoring—A Case Study. Sustainability 2022, 14, 10179. [Google Scholar] [CrossRef]
  46. Aruga, K.; Sessions, J.; Akay, A.E. Application of an airborne laser scanner to forest road design with accurate earthwork volumes. J. For. Res. 2005, 10, 113–123. [Google Scholar] [CrossRef]
  47. Matinnia, B.; Parsakhoo, A.; Mohamadi, J.; Jouibary, S.S. Study of the LiDAR accuracy in mapping forest road alignments and estimating the earthwork volume. J. For. Sci. 2018, 64, 469–477. [Google Scholar] [CrossRef]
  48. Chonpatathip, S. Earthwork Volume Measurement in Road Construction Using Unmanned Aerial Vehicle (UAV). Int. J. Geoinform. 2023, 19, 51–64. [Google Scholar] [CrossRef]
  49. Türk, Y.; Canyurt, H. Capabılıtıes of usıng UAVs to determıne forest road excavatıon volumes ın mountaınous areas. Šumarski List 2024, 137–150. Available online: https://ojs.srce.hr/index.php/sumlist/article/view/29496 (accessed on 13 November 2025). [CrossRef]
  50. Šikić, D.; Babić, B.; Topolnik, D.; Knežević, I.; Božičević, D.; Švabe, Ž.; Piria, I.; Sever, S. Tehnički Uvjeti za Gospodarske Ceste, 1st ed.; Znanstveni Savjet za Promet Jugoslavenske Akademije Znanosti i Umjetnosti: Zagreb, Croatia, 1989; pp. 1–78. [Google Scholar]
  51. Anonymous. Pravilnik o provedbi intervencije 73.08. “Izgradnja šumske infrastructure”. In iz Strateškog Plana Zajedničke Poljoprivredne Politike Republike Hrvatske 2023–2027; NN 90/2024; Government of Croatia: Zagreb, Croatia, 2024; Available online: https://narodne-novine.nn.hr/clanci/sluzbeni/full/2024_07_90_1681.html (accessed on 5 November 2025).
  52. CROPOS—Državna Mreža Referentnih Stanica Republike Hrvatske. Available online: https://www.cropos.hr/ (accessed on 15 August 2025).
  53. Vautherin, J.; Rutishauser, S.; Schneider-Zapp, K.; Choi, H.F.; Chovancova, V.; Glass, A.; Strecha, C. Photogrammetric accuracy and modeling of rolling shutter cameras. ISPRS 2016, 3, 139–146. [Google Scholar] [CrossRef]
  54. Zhou, Y.; Daakir, M.; Rupnik, E.; Pierrot-Deseilligny, M. A two-step approach for the correction of rolling shutter distortion in UAV photogrammetry. ISPRS 2020, 160, 51–66. [Google Scholar] [CrossRef]
  55. Štroner, M.; Urban, R.; Křemen, T.; Braun, J. UAV DTM acquisition in a forested area–comparison of low-cost photogrammetry (DJI Zenmuse P1) and LiDAR solutions (DJI Zenmuse L1). Eur. J. Remote Sens. 2023, 56, 2179942. [Google Scholar] [CrossRef]
  56. Urban, R.; Štroner, M.; Kovanič, L.; Blišťan, P.; Křemen, T.; Braun, J.; Peťovský, P.; Topitzer, B. Testing the accuracy and characteristics of data acquired using DJI Zenmuse L1 and L2 lidar systems and photogrammetric data acquired using DJI Zenmuse P1 in a quarry environment. Eur. J. Remote Sens. 2025, 58, 2595361. [Google Scholar] [CrossRef]
  57. Anonymous. Specifikacija Proizvoda, LiDAR Snimanje iz Zraka; Republika Hrvatska—Državna Geodetska Uprava: Zagreb, Croatia, 2022; pp. 1–50. Available online: https://dgu.gov.hr/UserDocsImages/dokumenti/Istaknute%20teme/Multisenzorno%20snimanje/LiDAR%20snimanje%20iz%20zraka.pdf (accessed on 10 October 2025).
  58. Chekole, S.D. Surveying with GPS, Total Station and Terresterial Laser Scaner: A Comparative Study. Master’s Thesis, Royal Institute of Technology (KTH), Stockholm, Sweden, May 2014. [Google Scholar] [CrossRef]
  59. Sokolović, D.; Bajrić, M. Volumen zemljanih radova pri izgradnji šumskih cesta na strmim terenima. Nova Meh. Šum. 2015, 36, 33–42. Available online: https://hrcak.srce.hr/file/227355 (accessed on 10 November 2025).
  60. Hrůza, P.; Mikita, T.; Tyagur, N.; Krejza, Z.; Cibulka, M.; Procházková, A.; Patočka, Z. Detecting Forest Road Wearing Course Damage Using Different Methods of Remote Sensing. Remote Sens. 2018, 10, 492. [Google Scholar] [CrossRef]
  61. Kınalı, M.; Çalışkan, E. Use of unmanned aerial vehicles in forest road projects. Bartın Orman Fakültesi Derg. 2022, 24, 530–541. [Google Scholar] [CrossRef]
  62. Kim, Y.H.; Shin, S.S.; Lee, H.K.; Park, E.S. Field Applicability of Earthwork Volume Calculations Using Unmanned Aerial Vehicle. Sustainability 2022, 14, 9331. [Google Scholar] [CrossRef]
  63. Gülci, S. The determination of some stand parameters using SfM-based spatial 3D point cloud in forestry studies: An analysis of data production in pure coniferous young forest stands. Environ. Monit. Assess. 2019, 191, 495. [Google Scholar] [CrossRef]
  64. Su, J.; Bork, E. Influence of vegetation, slope, and lidar sampling angle on DEM accuracy. Photogramm. Eng. Remote Sens. 2006, 72, 1265–1274. [Google Scholar] [CrossRef]
  65. Tinkham, W.T.; Smith, A.M.; Hoffman, C.; Hudak, A.T.; Falkowski, M.J.; Swanson, M.E.; Gessler, P.E. Investigating the influence of LiDAR ground surface errors on the utility of derived forest inventories. Canad. J. For. Res. 2012, 42, 413–422. [Google Scholar] [CrossRef]
  66. Salleh, M.R.M.; Ismail, Z.; Rahman, M.Z.A. Accuracy assessment of lidar-derived digital terrain model (DTM) with different slope and canopy cover in tropical forest region. ISPRS 2015, 2, 183–189. [Google Scholar] [CrossRef]
Forests 17 00265 g001
Figure 1. Study area.
Figure 1. Study area.
Forests 17 00265 g001
Forests 17 00265 g002
Figure 2. Workflow of data collection and processing: (A) ALSUAV, (B) UAVSfM, and (C) ALSDGU.
Figure 2. Workflow of data collection and processing: (A) ALSUAV, (B) UAVSfM, and (C) ALSDGU.
Forests 17 00265 g002
Forests 17 00265 g003
Figure 3. Forest road cross-section with highlighted parameters that were tested.
Figure 3. Forest road cross-section with highlighted parameters that were tested.
Forests 17 00265 g003
Forests 17 00265 g004
Figure 4. Difference in cut volume using the tested methods. Data is shown as mean value ± standard error. Methods sharing the same letter do not differ significantly based on Tukey’s HSD test (p < 0.05).
Figure 4. Difference in cut volume using the tested methods. Data is shown as mean value ± standard error. Methods sharing the same letter do not differ significantly based on Tukey’s HSD test (p < 0.05).
Forests 17 00265 g004
Forests 17 00265 g005
Figure 5. Fill volume using the tested methods. Data is shown as mean value ± standard error. Methods sharing the same letter do not differ significantly based on Tukey’s HSD test (p < 0.05).
Figure 5. Fill volume using the tested methods. Data is shown as mean value ± standard error. Methods sharing the same letter do not differ significantly based on Tukey’s HSD test (p < 0.05).
Forests 17 00265 g005
Forests 17 00265 g006
Figure 6. Mass haul graph of created designs.
Figure 6. Mass haul graph of created designs.
Forests 17 00265 g006
Forests 17 00265 g007
Figure 7. Terrain slope using the measuring methods. Data is presented as mean ± standard error. Letter a indicates no significant differences between measuring methods (Tukey’s HSD test, p < 0.05).
Figure 7. Terrain slope using the measuring methods. Data is presented as mean ± standard error. Letter a indicates no significant differences between measuring methods (Tukey’s HSD test, p < 0.05).
Forests 17 00265 g007
Forests 17 00265 g008
Figure 8. Carriageway value of the tested measuring methods. Data is presented as mean ± standard error. Letter a indicates no significant differences between measuring methods (Tukey’s HSD test, p < 0.05).
Figure 8. Carriageway value of the tested measuring methods. Data is presented as mean ± standard error. Letter a indicates no significant differences between measuring methods (Tukey’s HSD test, p < 0.05).
Forests 17 00265 g008
Forests 17 00265 g009
Figure 9. Impact of cross-terrain slope error on cut error: (A) ALSDGU design, (B) UAVSfM design.
Figure 9. Impact of cross-terrain slope error on cut error: (A) ALSDGU design, (B) UAVSfM design.
Forests 17 00265 g009
Forests 17 00265 g010
Figure 10. Impact of cross-terrain slope error on fill error: (A) ALSDGU design, (B) UAVSfM design.
Figure 10. Impact of cross-terrain slope error on fill error: (A) ALSDGU design, (B) UAVSfM design.
Forests 17 00265 g010
Forests 17 00265 g011
Figure 11. Impact of carriageway value error on cut error: (A) ALSDGU design, (B) UAVSfM design.
Figure 11. Impact of carriageway value error on cut error: (A) ALSDGU design, (B) UAVSfM design.
Forests 17 00265 g011
Forests 17 00265 g012
Figure 12. Impact of carriageway value error on fill error, (A) ALSDGU design (B) UAVSfM design.
Figure 12. Impact of carriageway value error on fill error, (A) ALSDGU design (B) UAVSfM design.
Forests 17 00265 g012
Table 1. Technical characteristics of used UAVs.
Table 1. Technical characteristics of used UAVs.
SpecsDJI Mavic 3 EDJI Matrice 350 RTK
Dimensions; Folded (without propellers)221.0 × 96.3 × 90.3 mm (L × W × H)810 × 670 × 430 mm (L × W × H)
Dimensions; Unfolded (without propellers)347.5 × 283 × 107.7 mm (L × W × H)430 × 420 × 430 mm (L × W × H)
Weight0.915 kgWithout batteries—3.77 kg
With two TB65 batteries—6.47 kg
Max Take-off Weight1.063 kg9.2 kg
Hovering AccuracyVertical:
±0.5 m (GNSS)
±0.1 m (RTK)

Horizontal:
±0.5 m (GNSS)
±0.1 m (RTK)		Vertical:
±0.5 m (GNSS)
±0.1 m (RTK)

Horizontal:
±1.5 m (GNSS)
±0.1 m (RTK)
RTK Positioning Accuracy (RTK FIX)1 cm + 1 ppm (horizontal)
1.5 cm + 1 ppm (vertical)		1 cm + 1 ppm (horizontal)
1.5 cm + 1 ppm (vertical)
Max Flight Speed15 m/s23 m/s
Max Take-off Altitude Above Sea Level6000 m5000 m or 7000 m
Max Wind Speed Resistance12 m/s12 m/s
Max Flight Time (no wind)45 min55 min
Operating Temperature Range−10° to 40 °C−20° to 50 °C
Class (EU)C1C3
SensorDJI Mavic 3E Wide Camera:
20 MP sensor
FOV: 84°
Format Equivalent: 24 mm
Aperture: f/2.8–f/11
Focus: 1 m to ∞
Electronic Shutter: 8–1/8000 s
Mechanical Shutter: 8–1/2000 s		DJI Zenmuse L2
Ranging Accuracy: 2 cm at a flight altitude of 150 m
Maximum Returns Supported: 5
Minimum Detection Range: 3 m
Scanning Modes: Non-repetitive scanning pattern and repetitive scanning pattern
Table 2. Cut volume descriptive statistics by designs based on different survey methods.
Table 2. Cut volume descriptive statistics by designs based on different survey methods.
MethodMean, m3Minimum, m3Maximum, m3RMSE, m3
ALSUAV17.990.0069.39-
ALSDGU17.510.0070.664.03
UAVSfM21.730.0074.885.82
Table 3. Fill volume descriptive statistics by designs based on different survey methods.
Table 3. Fill volume descriptive statistics by designs based on different survey methods.
MethodMean, m3Minimum, m3Maximum, m3RMSE, m3
ALSUAV5.780.0047.99-
ALSDGU6.920.0049.522.04
UAVSfM4.470.0037.982.92
Table 4. Cross-terrain descriptive statistics by designs based on different survey methods.
Table 4. Cross-terrain descriptive statistics by designs based on different survey methods.
MethodMean, %Minimum, %Maximum, %RMSE, %
ALSUAV15.623.4728.75-
ALSDGU15.672.7922.761.55
UAVSfM15.801.7224.291.20
Table 5. Carriageway value descriptive statistics by designs based on different survey methods.
Table 5. Carriageway value descriptive statistics by designs based on different survey methods.
MethodMean, mMinimum, mMaximum, mRMSE, m
ALSUAV0.14−0.711.23-
ALSDGU0.17−0.681.190.08
UAVSfM0.06−0.661.010.12
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Papa, I.; Hodak, L.; Popović, M.; Đuka, A.; Pentek, T.; Lovrinčević, M. ALS and SfM Field Data Survey as a Basis of Forest Road Design. Forests 2026, 17, 265. https://doi.org/10.3390/f17020265

AMA Style

Papa I, Hodak L, Popović M, Đuka A, Pentek T, Lovrinčević M. ALS and SfM Field Data Survey as a Basis of Forest Road Design. Forests. 2026; 17(2):265. https://doi.org/10.3390/f17020265

Chicago/Turabian Style

Papa, Ivica, Luka Hodak, Maja Popović, Andreja Đuka, Tibor Pentek, and Mihael Lovrinčević. 2026. "ALS and SfM Field Data Survey as a Basis of Forest Road Design" Forests 17, no. 2: 265. https://doi.org/10.3390/f17020265

APA Style

Papa, I., Hodak, L., Popović, M., Đuka, A., Pentek, T., & Lovrinčević, M. (2026). ALS and SfM Field Data Survey as a Basis of Forest Road Design. Forests, 17(2), 265. https://doi.org/10.3390/f17020265

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop