*3.2. Soil Property*

Figure 8 shows the grain size distributions of the obtained soil samples as a result of grain size analysis. All the samples were classified as GFS, fine-grained sandy gravel, of which the mechanical feature is classified as suitable for compaction [52]. Two samples, i.e., Site 1 FY2013 and Site 2 FY2021, were additionally classified to have a wide distribution, the uniformity coefficient of which is *U*<sup>c</sup> ≥ 10, and the coefficient of curvature *U*<sup>c</sup> - is between 1 and 3.

**Figure 8.** Grain size distribution of soil sampled at Sites 1 and 2.

The result of the grain size analysis indicated that the soil of both sites has a good property for compaction. It also can be said that there was no major difference between the soil properties of Site 1 and Site 2.

#### *3.3. Site 1: Measurement of Roadbed Strength*

The result of the two-way ANOVA on the *N*<sup>d</sup> values in Site 1 indicated that two factors were significant, whereas their interaction was not: A, relative position within the road width, and B, fiscal year of construction (Table 3). In other words, the population means of all the levels were not the same in Factors A and B, and there was no interaction effect in the combination of Factor A levels and Factor B levels.


**Table 3.** ANOVA table of *N*<sup>d</sup> in Site 1.

Total mean: 15.65.

From a brief inspection of the multiple comparison analysis of Factor A levels (Figure 9), it can be concluded that the roadbed soil strength was not even over the road width. In other words, it was stronger in the cut bunks and weaker in the fill bunks. Thus, the main objective of the Shimanto method was not achieved in the investigated spur road.

As for the effect of Factor B, level B3 (FY2014) had significantly greater *N*<sup>d</sup> values than the other levels (Figure 10). From a visual inspection of the site, a possible reason would be that hard rock materials were more distributed in the FY2014 section than the others. The age-related change, such that older roadbeds gained a larger strength [48,49], is not observed in Figure 10.

Figure 11 shows all the observations of the *N*<sup>d</sup> values in Site 1 accompanied with three thresholds: 1.4, 5.4, and 10.0. While no observation was lower than *N*<sup>d</sup> = 1.4, 15 observations in B4 (FY2015) and B5 (FY2016) and 4 observations in B1 and B2 (FY2013) were lower than *N*<sup>d</sup> = 5.4. In Figure 10, where the values were averaged, the age-related change was not clear. However, the lower limits of the values seem to have a tendency of age-related change and are hence lower for younger ages in Figure 11.

**Figure 9.** Main effect of Factor A, relative position within the road width, of Site 1 on the *N*<sup>d</sup> -value. The error bars represent the standard error (SE). Different letters indicate significant differences between the population means. Tukey test, *p* < 0.01, *n* = 24.

**Figure 10.** Main effect of Factor B, fiscal year of construction, of Site 1, on the *N*<sup>d</sup> -value. The error bars and different letters have the same meaning as in Figure 9. *p* < 0.01, *n* = 25 except for B3 (*n* = 20).

#### *3.4. Site 2*

3.4.1. Measurement of Roadbed Strength

In the ANOVA table for the *N*<sup>d</sup> values in Site 2, all the single factors and a few interactions were significant (Table 4). The insignificant interactions were pooled into the error.

**Figure 11.** *N*d—values of Site 1 accompanied with thresholds.

**Table 4.** ANOVA table of *N*<sup>d</sup> in Site 2.


Total mean: 9.79.

As for Factor A (relative position in the road width), A1 (0.5 m from center, 14.4 ± SE (Standard Error) 4.4) was significantly greater than the others (A2, 1.5 m from center, 7.5 ± SE 1.2; A3, 2.5 m from center, 7.4 ± SE 1.3; Tukey test, *p* < 0.05, *n* = 32), implying that the soil strength increased from natural 7.4–7.5 to compacted 14.4. The effect of Factor B should be evaluated as the interaction A × B. The average *N*<sup>d</sup> values of A1 × B2, the rut of the uphill side (21.1 ± SE 8.5), were significantly greater than the downhill side (7.6 ± SE 1.0), which was not significantly different from the other remaining interactions (Tukey test, *p* < 0.05, *n* = 16). The tendency is the same as that of Site 1 (Figure 9). Note that these *N*<sup>d</sup> values were averaged through a depth of 0–100 cm (D1: 0–50 cm and D2: 50–100 cm). The average *N*<sup>d</sup> values of D2 (15.5 ± SE 3.0) were significantly greater than those of D1 (4.1 ± SE 0.4; *p* < 0.01, *n* = 48). As for Factor C, the fiscal year of construction, the level C2 (FY2020, one year after construction, 17.2 ± SE 5.8) was significantly greater than the other levels (C1: FY2019, two years after construction, 6.6 ± SE 1.5; C3: FY2021-1, 0.5 year after construction, 8.4 ± SE 1.3; C4: FY2021-2, 0.5 year after construction, 7.0 ± SE 1.1; Tukey test, *p* < 0.05, *n* = 24).

The effect of Factor C can be more clearly observed as interactions between other factors. Figure 12 shows the interaction of Factors A and C. The average *N*<sup>d</sup> values of the combined levels of A1 × C2 were significantly greater than the others, implying a possibility that there were harder subsoils or rocks under the investigated section of C2.

**Figure 12.** Interaction of Factor A (relative position) and Factor C (fiscal year of construction) of Site 2 on the *N*d—value. The error bars and different letters have the same meaning as in Figure 9. *p* < 0.01, *n* = 8.

Figure 13 shows the interaction of Factors B, C, and D. From the figure, it can be observed that the average *N*<sup>d</sup> values of the combined levels B1 × C2 × D2 were significantly greater than the others, also implying a possibility that there were harder subsoils or rocks under the rut of the uphill side at a depth of 50–100 cm on the investigated section of C2.

**Figure 13.** Interaction of Factor B (downhill side or uphill side), Factor C (fiscal year of construction), and Factor D (penetrating depth) of Site 2 on the *N*<sup>d</sup> -value. The error bars and different letters have the same meaning as in Figure 9. *p* < 0.01, *n* = 6. B1: Downhill side, B2: Uphill side; D1: 0–50 cm, D2: 50–100 cm.

To provide an overview of all the aforementioned observations in Site 2, Figure 14 shows the all-observation average on Factor D, i.e., depth 0–100 cm for comparison with Figure 11 of Site 1. In the figure, the observations on the spur road are marked with filled circles, whereas those on natural ground are marked with open circles. From the figure, the *N*<sup>d</sup> values on the spur road (closed circles) have a tendency of age-related change, except for C1 (FY2019). The *N*<sup>d</sup> values on the natural ground (open circles) have a large variation, which probably reflects various situations of the natural ground.

**Figure 14.** *N*d—values of Site 2 accompanied with thresholds.

Compared with the threshold values for C1 (FY2019) and C4 (FY2021-2) × R2, the *N*<sup>d</sup> values on the spur road were lower than *N*<sup>d</sup> = 5.4. Thus, the compaction of the roadbed material soils, especially for FY2019, would have been insufficient to make the excavated soil sufficiently strong.

#### 3.4.2. Measurement of Bearing Capacity

The result of the three-way ANOVA on the in situ CBR values in Site 2 is summarized in Table 5. Although Factor R was not significant, it was not pooled into the error because the interaction B × C × R, which includes Factor R, was significant [53].

**Table 5.** ANOVA table of in-situ CBR in Site 2.


Total mean: 14.5%.

As for Factor B (downhill side or uphill side), B2 (uphill side, 16.5 ± SE 0.7%) was significantly greater than B1 (downhill side, 12.5 ± SE 0.7%; *p* < 0.01, n = 40), implying that the compaction of the road surface soil was better performed in the uphill side than in the downhill side. As for Factor C, the average in situ CBR values of C1 (FY2019, 17.1 ± SE 1.1) and C2 (FY2020, 16.6 ± SE 0.8) were significantly greater than those of C3 (FY2021-1, 12.5 ± SE 1.0) and C4 (FY2021-2, 12.0 ± SE 1.1; Tukey test, *p* < 0.01, *n* = 20). The tendency could be regarded as age-related change, i.e., natural compaction [49,50]. The effect of interaction among Factors B, C, and R is shown in Figure 15. In the combined levels of B1 × C1 × R2 and B2 × C4 × R2, the average values seem to be larger than the combined effects of Factors B and C. A reason would be the experimental error probably caused by cobbles or small rocks submerged in the road surface. However, according to the related studies that have evaluated the bearing capacities of the spur road surface using an in situ CBR tester [46–49], the total mean of 14.5% in Site 2 could be regarded as moderate, or not too low, for spur roads used for crawler-type machines.

**Figure 15.** Interaction of Factor C (fiscal year of construction), Factor R (repetition), and Factor B (downhill side or uphill side) of Site 2 on in-situ CBR. The error bars and different letters have the same meaning as in Figure 9. *p* < 0.05, *n* = 5. B1: Downhill side, B2: Uphill side.

#### **4. Conclusions**

We investigated spur roads in Kochi University Forest constructed using the Shimanto method in order to verify whether sufficient roadbed strength and road surface bearing capacity were achieved. The main conclusions are as follows:


In the present study, the year of construction was treated as time elapsed since construction. However, in order to properly evaluate the strength development of a single pavement section, it is necessary to test the same pavement section at different ages. It would have been better to compare soil strength for three different depths, e.g., 0–30 cm, 30–60 cm, and 60–100 cm, to check the difference between fill and cut slopes and the effect of natural compaction. Although such analysis was not achieved in the present study because the resolution of *N*<sup>d</sup> values was limited by the field situation, the authors would like to confirm the problem in the future study.

The lower strength of the roadbed on the downhill side of the spur roads or embankments was also observed in a related study on spur road construction. Furthermore, it was suggested that the soil under the road width should be excavated more widely toward the cut slope side before compaction to obtain the same roadbed strength on the downhill side as that on the cut slope side. Regarding the age-related change effect of the roadbed strength and bearing capacity, it was reported that in areas with low compaction ability soils, there is a spur road construction method that requires three years for completion. The method seems to effectively use the age-related change effect of the road strength. It is recommended that spur roads be left unused for a certain period of time before use in order to stabilize the materials of the constructed road body.

**Author Contributions:** Conceptualization, Y.S. and T.Y.; methodology, Y.S.; software, Y.S., S.H. and I.K.; validation, Y.S., S.H. and I.K.; formal analysis, Y.S., S.H. and I.K.; investigation, Y.S., S.H., H.A. and I.K.; resources, Y.S. and S.H.; data curation, Y.S. and S.H.; writing—original draft preparation, S.H. and I.K.; writing—review and editing, Y.S.; visualization, Y.S., S.H. and I.K.; supervision, Y.S.; project administration, Y.S.; funding acquisition, Y.S. and T.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI, grant numbers 21H03672, 22H03800, and 22K05747.

**Data Availability Statement:** The data used to support the findings of this study are available from the corresponding author upon request.

**Acknowledgments:** The authors are grateful to Ryohei Wada and Masa'aki Tateishi for supporting the field investigation in Site 2, and Naoyuki Hashimoto for supporting the GNSS survey in Site 2.

**Conflicts of Interest:** The authors declare no conflict of interest.
