3. RESULTS

3.1 Soil profile characteristics

The general characteristics of soil pedons are illustrated based on pedon 1 (Figure 2b). At 0–10 cm soil depth, the soil was dark red with numerous wormholes found due to the high organic matter content of the surface layer. Cracks began to appear below 10 cm, with a few soil profiles showing various crack configurations. For instance, pedon 1 contained a V-shaped crack at 18–21 cm and an I-shaped crack at approximately 19–33 cm, while most of the profiles contained more I-shaped cracks. Pedon 1 consisted of four genetic horizons, delineated as A (0–8 cm), B1 (8–21 cm), B2 (21–54 cm), and BC (54–60 cm) (Figure 2b), which were classified as loamy skeletal, mixed, super-active, or mesic, respectively (Soil Survey Staff, 1999).
This experiment applied GPR technology to physically probe soil cracks and identify natural crack configurations before the infiltration experiment was unfolded. High-signal areas floating on the radargrams were identified, and the envelopes were calculated as the average crack coverage amplitude (Table 3). Soil cracks existed or did not result in different envelopes. The envelope of non-crack profiles had no substantial fluctuation with reflection wave conduction, while the profile with cracks showed more signal fluctuations at 0.3–0.6 m. The extracted crack signal parameters TotAra, Maxa, Mina, and Acumasupport the phenomenon (Table 3). Furthermore, the amplitude fluctuation of the I-shaped crack was less marked than that of the V- and Λ-shaped cracks (Figure S3). Thus, the different soil crack configurations, as determined by GPR and excavation, could provide practical support for conducting infiltration experiments. Notably, geophysical methods cannot directly explain all the properties of soil cracks. Whether the crack width, inclusion, and signal fluctuations corresponded to each other based on the raw data remains unclear.

3.2 Comparison between Brilliant Blue FCF solution and water transport

Figure 3 shows the infiltration trajectories with time, showing that the Brilliant Blue FCF solution penetrates faster than water. The difference between water and dye penetration was observed at the beginning of infiltration, as early as 3 min. In the VR1.5 treatment (Figure 3c), the matrix flow was notably dominant throughout the infiltration process, and no preferential flow occurred. This phenomenon indicates that water was probably not transported down V-shaped cracks (Figure 3b). The matrix flow dominated before 30 min, while preferential flow dominated later in the IR1.5 and ΛR1.5 treatments (Figure 3a–b).
The dye-stained and wetting areas show a linear fit with the primary function, in which the R 2 values were all above 0.8 (Figure 4). Regarding the I-shaped crack filled with rock fragments, slight variabilities were found between Brilliant Blue FCF solution and water when the dye-stained area was less than 21%, at which point the former infiltrated marginally faster than the latter; however, these differences gradually became more potent after the turning point, indicating that there was a lag in the effects of the Brilliant Blue FCF solution compared to water (Figure 3–4). Comparison with the I-shaped crack showed that the retardation on the solution infiltration in the Λ-shaped crack appeared later, around a dye-stained area ratio of 24%–27%. This result may explain, at least in part, that the retardation on the Brilliant Blue FCF with time was influenced by the crack configuration, and the flow pathways of the Brilliant Blue FCF were not fully representative of water infiltration paths.

3.3 Effect of crack properties on infiltration and preferential flow

The effects of the crack properties (inclusion, crack width, and configuration) on the infiltration rate, maximum dye-penetration depth, cumulative infiltration, and wetting advancing rate are shown in Figure 5–7. Generally, a noticeable trend in the infiltration rate was observed, which could be divided into a sharp decline and a steady decline for all treatments. The turning point occurred at approximately 40 min, after which, the infiltration rate reached a relatively smooth decline for all treatments (Figure 5a, 6a, and 7a). The maximum dye penetration depth and cumulative infiltration increased at a slower rate with time after 40 min (Figure 5b–c, 6b–c, and 7b–c).

3.3.1 Crack inclusion

Regarding the influence of inclusions, the infiltration rate was faster by 18.8% on average when filled with rock fragments than when filled with sand grains (Figure 5a), and the existence of cracks increased the dye-penetration depth, cumulative infiltration, and wetting front depth by at least 5.2%, 63.2%, and 4.4% (Figure 5b–d). These results were apparent in the IR2, IS2, ΛR1.5, and ΛS1.5 treatments, indicating that cracks filled with rock fragments enhanced the infiltration capacity of the soil. The ΛR1.5 treatment had a notably higher dye penetration depth, cumulative infiltration, and wetting front depth at the end of infiltration (Figure 5b–d). For example, the 70-min wetting front depth of the ΛR1.5 treatment was 46.5 mm, whereas that of the ΛS1.5 treatment decreased by 17.0 % (Figure 5d). Similarly, the maximum dye-penetration depth and cumulative infiltration decreased by 14.9% and 4.4%, respectively (Figure 5b–c).

3.3.2 Crack width

The volumes corresponding to the three crack widths were 8.0, 17.5, and 31.5 cm3, respectively. Regarding crack widths, significant differences (p <0.05) were identified in the infiltration rates among the three treatments with different crack widths (Figure 8b). After the infiltration reached relative stability, the infiltration rate was faster for the 1.5-cm width crack (Figure 6a). The maximum dye-penetration depth, cumulative infiltration, and wetting advancing rate were the highest when the crack width was 1.5 cm (Figure 6b–d), and there was a significant difference (p <0.05) in the maximum dye-penetration depth with the 1.5-cm width crack (Figure 8b). Our experimental results indicate the existence of a certain optimal crack width between 1 and 2 cm, which allowed the infiltration rate to be maximised. After the crack volume reached a value range of 8.0–31.5 cm3, moisture was more inclined to be stored in the crack than to infiltrate downward.

3.3.3 Crack configuration

Regarding the configuration, the chart shows that the ΛR1.5 treatment had a substantially higher penetration rate than other configurations, ranging from 54% to 120% higher than other two configurations (Figure 7a). The maximum dye-penetration depth and cumulative infiltration of the Λ-shaped configuration increased steadily at the beginning of infiltration, crossing the line for the I-shaped configuration at approximately 50 min (Figure 7b and 7c). When infiltration proceeded to a certain process, the Λ-shaped configuration increased the rate of liquid passage, but the effect on the wetting advancing rate was not obvious. The wetting front depth nearly overlapped between the three crack configurations, with variations ranging from 0.03% to 38.39% (Figure 7d). Significant differences (p <0.05) were observed between the IR1.5 and ΛR1.5 treatments, but there were no significant differences (p <0.05) between these and the VR1.5 treatment (Figure 8c).

3. 4 Contribution of crack properties to preferential flow

The I- and Λ-shaped treatments, filled with rock fragments, generated preferential flow, whereas the V-shape was not sensitive to preferential flow (Table 2). Thus, the analysis of dye coverage and preferential flow indices was displayed only for IR1, IR1.5, IR2, and ΛR1, ΛR1.5, and ΛR2 treatments.
Figure 9 illustrates the turning range of the preferential flow ratio with increasing crack width for the same configuration and inclusion. The experiment proved that the preferential flow extensively slowed down the decline in dye coverage and remained steady in the IR1.5 and ΛR2 treatments (Figure 9b and 9f). For example, the percentage of preferential flow dye coverage in the IR1.5 treatment ranged from 60.7% at 15 cm to 14.4% at 20 cm, while in the ΛR1.5 treatment, it ranged from 75.9% at 14 cm to 7.5% at 17 cm. The variations indicated more preferential pathways in the ΛR1.5. However, a rebound point at a soil depth of 15 cm was observed in the IR2 treatment, in which the morphology of the stained end appeared radially orbicular (Figure 9c). This demonstrates the configuration has a greater effect on preferential flow and a wider linear crack filled with rock fragments likely leads to a lateral flow at the end of the crack.
Overall, the preferential flow ratio increased with increasing crack width, similar to the length index of the preferential flow. Generally, the preferential flow ratio and matrix flow depth showed opposite trends; however, this was not always the case (Table 4). Treatment with IR1.5, however, did not match the above regularity. The nonuniformity and spatial variability of the linear crack was more distinct, which led to an obvious preferential flow phenomenon. The response of the Λ-shaped crack to the preferential flow with time lagged. For instance, the depth of matrix flow for I-shape was 0.9%–33.3% lower than that for Λ-shape; however, the length index of preferential flow was higher by 3.9%–11.4% (Table 4).