Changes in CI and IR over time
The changes in CI over time for the HS and W-RS samples are shown in Fig. 4, and the change in IR over time is shown in Fig. 5. Because of the influence of the ordinate axis scale, the CI and IR changes do not appear to be significant when\(\theta=\)4.7% (Figs. 4(b) and 5(b)). Thus, we scaled the ordinate of CI and IR, and present the enlarged plots in Fig. 6.
Figure 4 shows that higher values of ISWC produce a larger CI in both the HS and W-RS samples. The difference is that CI exhibits smooth monotonic growth for HS (Fig. 4(a)), but a more variable increase for W-RS. The WR effect is very evident in these CI curves, similar to those measured by Moret-Fernández et al. (2019), where the CI has an inflection point and demonstrates smooth monotonic growth at the critical point (Fig. 4(b)). Lichner et al. (2013) defined the relationship between CI and the square root of time as “hockey-stick-like,” and suggested a procedure for estimating the WR cessation time. Figure 5 shows that larger values of ISWC give a larger IR. The IR of the HS sample decreases monotonously with time (Fig. 5(a)) because of the decreasing contribution of the capillary (or pressure head)-driven component of the hydraulic gradient that develops as the wetting front moves further away from the surface (Lichner et al., 2013). The IR then approaches steady infiltration. In contrast, the IR of the W-RS sample (Fig. 5(b)) exhibits a gradual decrease in the initial period (0–20 min), and then presents a single-peak curve that increases before decreasing. The larger ISWC values (θ=6.2% and θ=9.6%) produce an earlier inflection point (\(\cong\)310 min) than the smaller ISWC (θ=4.7%, \(\cong\)568 min). For all three ISWCs, the wave peak of the IR is narrow and the process of reaching the maximum IR is dramatic, and the attenuation process is slower after the peak. That is, the IR presents an asymmetric single-peak curve with a left-skewed distribution. From Fig. 5(b), we can observe that the stable IR after the peak is larger than that before the peak. In other words, the water enters the soil slowly before the inflection point, and then enters more quickly after this point. This result is in accordance with that reported by Moret-Fernández et al. (2019). The slower infiltration before the inflection point could be explained by the hydrophobicity preventing water from entering the soil, resulting in a low IR. In essence, the hydraulic gradient is mainly determined by the matrix in unsaturated flow (Vogelmann et al., 2017), and WR reduces the matric potential. Hence, the volume of water-conductive pores is lower, promoting a decrease in IR. Following the wetting of hydrophobic compounds, the repellency effect disappears after prolonged contact with water (Filipović et al., 2018), allowing the IR to increase as the WR disappears (Alagna et al., 2017). Another viewpoint is that any initially postponed infiltration increases after the CA between the water and the soil particles decreases (Bughici et al., 2016). The mechanisms for IR increase suggested by Alagna et al. (2017) and Bughici et al. (2016) are basically the same when stated in terms of CA.
The CI curve is relatively smooth before and after the inflection point (Fig. 4(b)), and the IR has a significant peak when \(\theta\)=6.2% and\(\theta\)=9.6% (Fig. 5(b)). However, Fig. 6(a) shows that the CI has several inflection points and a distinctly mutated inflection point at 580 min when \(\theta\)=4.7%. In this case, the CI exhibits a multi-stage growth trend. This is consistent with the results of Filipović et al. (2018) using HYDRUS (2D/3D) to simulate the hydraulic performance of W-RS under drought conditions.
Figure 6(b) shows that there are multiple peaks and troughs on either side of the maximum IR (580 min). This suggests that the infiltration process of W-RS has unstable infiltration properties when the ISWC value is small, which may be caused by spatial variability in the WR (Liu et al., 2019) leading to preferential flow characteristics in WR soils (Rye et al., 2017). Because of the irregular distribution of the priority path, the infiltration properties are unstable. Unstable infiltration may also be caused by the surface pressure head being less than the water-entry value, which is positive in repellent soils (Wang et al., 2000). However, taking 580 min as the cutoff when \(\theta\)=4.7%, the IR gradually increases before and gradually decreases after this point. From a macro perspective, this is still a single-peak curve.
Wang et al. (1998) suggested that when the air pressure ahead of the wetting front reaches an air-breaking value, soil air escapes from the surface, leading to an immediate decrease in the air pressure and an increase in the infiltration rate. When the air pressure falls below a certain air-closing value, air escape stops, the infiltration rate decreases, and the air pressure increases. This cyclic process repeats itself during the entire infiltration period, possibly explaining the inflection point in CI (Figs. 4(b) and 6(a)) and significant variability in IR (Figs. 5(b) and 6(b)), which are inconsistent with the traditional infiltration theory.
For the W-RS, before the experiment, we adjusted the water head difference (water-ponding depth)\(h_{0}\) between the Markova bottle and soil surface to 1 cm. After a period of infiltration, the soil surface was 2 cm higher than the initial surface as a result of volume expansion (Fig. 14(a)). The soil water was considered to be within 1 cm positive pressure infiltration for a short time after the beginning of the experiment, followed by \(h_{0}\)=1 cm − 2 cm=−1 cm negative pressure with a water-entry value of \(h_{\text{we}}>0\) (Wang et al., 2000). Therefore, the water head difference \({}_{h}\) between the water-ponding depth and the water-entry pressure head satisfied some pressure criterion, and when\({}_{h}=h_{0}-h_{\text{we}}=-1cm-h_{\text{we}}<0\), unstable flow occurred (Wang et al., 1998).