1

Fig. 1. Theoretical illustration of an ideal case of forest area of 100 ha, with 4 straight forest roads, mean road spacing (RS) of 250 m, road density (RD) of 40 m ha–1, calculative average extraction distance (AED0) of 62.5 m and calculative maximum extraction distance (MED0) of 125 m.

2

Fig. 2. A forest area of 2456 ha from Central Finland having forest roads of 62 406 m (25.4 m ha–1). With MED0 of 200 m (~RS/2), the accessibility index E = 0.708. The black vector line describes road network and dark grey the 200 m buffer around the road lines. Light grey describes the forest area that cannot be accessed with the MED0 of 200 m from the closest road.

3

Fig. 3. Illustration of grouping of segments. Small line segments (<20 m) are removed (red-coloured segments). Remaining line segments are grouped based on their orientation using the K-means algorithm (blue and green coloured segments).

4

Fig. 4. Illustration of the distance calculation between adjacent trails.

Table 1. Summary of the logging trail network metrics.
Definition Equation
Metrics that analyse quantitative aspects of the logging trail network TL Logging trail network length (m), derived from the logging trail network polyline
Atot Area (m2) of the logging site polygon
TD Logging trail network density (m ha–1)
TSn Nominal logging trail spacing (m)
DIST Distance travelled (m), calculated as the sum of segment lengths multiplied by the number of machine passes
Quantitative metrics considering buffer polygon regions AT Logging trail area (m2), calculated as TL multiplied by the nominal trail width (e.g., 4.5 m)
AT% Proportion of trail area within the logging site (%), calculated as AT/Atot
ANBRC Net boom reach coverage (m2), defined as the total area reachable within the nominal crane length (~10 m). Areas extending beyond the logging site polygon are clipped
ATSC Net coverage (m2) reached with the nominal logging trail spacing (TSn)
Spatial configuration metrics of the logging trail network CA Measures the share of the forest area that is reachable by the modern harvester
TD0 The density of the ideal logging trail network
CE Compares the actual trail length, TLact, to the ideal trail length, TL0, or actual trail length density, TDact, to the ideal trail length density, TD0
Ctot Structural-Coverage Efficiency index. Combines coefficients CA and CE
TSgc Average logging trail spacing
Table 2. Estimation of GNSS tracks per logging trail segment and the parameter values employed to consolidate the centerline (based on the algorithms by Cao et al. 2025).
Stand No. of tracks per logging trail segment Origin of GNSS Tracks H = harvester
H+F = harvester and forwarder 		Kernel density search radius Threshold value
No. Min Max Most frequent
1 1 5 2 H 1.5 0.70
2 1 14 4 H+F 1.5 0.70
3 1 14 14 H+F 0.65 0.85
4 1 14 2 H+F 1.5 0.70
5 1 5 2 H 2.5 0.80
6 1 5 2 H 1.5 0.70
7 1 14 6–7 H+F 2 0.80
8 1 8–9 2 H+F 2.5 0.80
9 1 14 6–7 H+F 1.5 0.70
10 1 14 4 H+F 1.5 0.70
5

Fig. 5. Part of logging site 1. The site polygon is shown in dark green, the 10 m buffer around the logging trails in light green, and the logging trail area in medium green. Yellow areas indicate overlaps of the boom reach, while red areas represent boom reach extending outside the stand polygon.

Table 3. Summary of logging trail network metric values by sample stand.
Stand No. Stand area
Atot
ha 		Trail length
TL
m 		Trail density
TD
m ha–1 		Trail spacing
TSgc
m 		Share of trail area
AT% 		Net boom reach coverage
ANBRC
m2 		Coverage Coefficient
CA 		Structural Efficiency
CE 		Structural – Coverage Efficiency index
Ctot = CA × CE–1
1 13.4 6558 487 20.0 21.4 10.9 0.81 0.98 0.83
2 19.8 9561 482 20.8 21.4 16.6 0.84 0.97 0.87
3 10.5 6244 594 20.3 25.7 19.0 0.87 1.19 0.73
4 13.9 9077 652 16.9 27.8 11.8 0.85 1.30 0.65
5 26.6 14292 537 21.4 23.7 23.2 0.87 1.07 0.81
6 7.83 4220 538 23.0 23.9 6.89 0.88 1.08 0.81
7 1.68 950 566 20.0 24.9 1.54 0.92 1.13 0.81
8 4.62 2526 546 20.0 23.9 3.92 0.85 1.09 0.78
9 6.29 3801 604 19.4 26.6 5.74 0.91 1.21 0.75
10 13.1 6590 502 23.4 22.2 11.3 0.86 1.0 0.86
6

Fig. 6. Logging trail network of stand 10 (a), demonstrating a well-designed layout, and of stand 4 (b), demonstrating a poorly executed layout. Color-coding represents the number of machine passes.

7

Fig. 7. In our small sample of ten sites, the geometrically corrected trail spacing (TSgc) was moderately correlated with the Structural Efficiency (r = –0.664, p = 0.036) (a) and with the Structural-Coverage Efficiency index (r = 0.753, p = 0.012) (b).