Variations in Soil Water Content, Inﬁltration and Potential Recharge at Three Sites in a Mediterranean Mountainous Region of Baja California, Mexico

: In this research, we examined temporal variations in soil water content ( θ ), inﬁltration patterns, and potential recharge at three sites with different mountain block positions in a semiarid Mediterranean climate in Baja California, Mexico: two located on opposing aspects (south- (SFS) and north-facing slopes (NFS)) and one located in a ﬂat valley. At each site, we measured daily θ between 0.1 and 1 m depths from May 2014 to September 2016 in four hydrological seasons: wet season (winter), dry season (summer) and two transition seasons. The temporal evolution of θ and soil water storage (SWS) shows a strong variability that is associated mainly with high precipitation (P) pulses and soil proﬁle depth at hillslope sites. Results shows that during high-intensity P events sites with opposing aspects reveal an increase of θ at the soil–bedrock interface suggesting lateral subsurface ﬂuxes, while vertical soil inﬁltration decreases noticeably, signifying the production of surface runoff. We found that the dry soil conditions are reset annually at hillslope sites, and water is not available until the next wet season. Potential recharge occurred only in the winter season with P events greater than 50 mm/month at the SFS site and greater than 120 mm/month at the NFS site, indicating that soil depth and lack of vegetation cover play a critical role in the transport water towards the soil–bedrock interface. We also calculate that, on average, around 9.5% (~34.5 mm) of the accumulated precipitation may contribute to the recharge of the aquifer at the hillslope sites. Information about θ in a mountain block is essential for describing the dynamics and movement of water into the thin soil proﬁle and its relation to potential groundwater recharge.


Introduction
Understanding the dynamics and transport of soil water content (θ) in thin mountain soils is essential for describing infiltration in superficial and deeper soil layers. Evaluation of the spatial and temporal distribution of θ is also crucial for assessing how recharge might vary with different locations in a mountain block [1,2]. Significant variations in θ depend on climatic forcing [3][4][5][6], soil properties, topography and vegetation [5,[7][8][9]. Each of these factors can influence θ over multiple spatiotemporal

Site Description
Chaparral shrubs cover entirely the study site of the sub-basin and the micro-basins under analysis (~10.3 ha). The natural ecosystem was clearly structured by variations in aspect. Figure 2 shows the vegetation differences between pole-facing and equator-facing slopes in our study site. In this case, a more xeric vegetation exists on the equator-facing slopes, while a more mesic vegetation prevails in the pole-facing slopes. The northern micro-basin (0.82 ha) is composed exclusively of SFS, which leads to sparse and xeric vegetation, small shrubs (less than 1 m) and the absence of grasses even in wet season. Dominant plant species are Eriogonum fasciculatum (California buckwheat) and Bahiopsis laciniata (San Diego County viguiera) [32,35]. In contrast, the southern micro-basin (4.53 ha) is dominated by NFS which cover about 80% of the area while the 20% remaining includes westfacing slopes. Here, vegetation is denser (more mesic), small and medium shrubs prevail, and herbaceous plants appear at ground level in wet season. Higher woody plant diversity is present, including Eriogonum fasciculatum, Fraxinus parryi (Chaparral ash), Condea emoryi (Desert lavender), Malosma laurina (Laurel sumac), Salvia apiana (White sage), Adenostoma fasciculatum (Chamise), Artemisia californica (California sagebrush), Bahiopsis laciniata and Baccharis salicifolia (Mule fat) [35]. The CT micro-basin (4.97 ha) is located in the transition of the southern and northern micro-basins, over a small inter-mountainous valley. At CT, vegetation is similar to the NFS in terms of cover and diversity of plant species. In addition, grasses are present in the wet season and there is a presence of small patches of trees.

Site Description
Chaparral shrubs cover entirely the study site of the sub-basin and the micro-basins under analysis (~10.3 ha). The natural ecosystem was clearly structured by variations in aspect. Figure 2 shows the vegetation differences between pole-facing and equator-facing slopes in our study site. In this case, a more xeric vegetation exists on the equator-facing slopes, while a more mesic vegetation prevails in the pole-facing slopes. The northern micro-basin (0.82 ha) is composed exclusively of SFS, which leads to sparse and xeric vegetation, small shrubs (less than 1 m) and the absence of grasses even in wet season. Dominant plant species are Eriogonum fasciculatum (California buckwheat) and Bahiopsis laciniata (San Diego County viguiera) [32,35]. In contrast, the southern micro-basin (4.53 ha) is dominated by NFS which cover about 80% of the area while the 20% remaining includes west-facing slopes. Here, vegetation is denser (more mesic), small and medium shrubs prevail, and herbaceous plants appear at ground level in wet season. Higher woody plant diversity is present, including Eriogonum fasciculatum, Fraxinus parryi (Chaparral ash), Condea emoryi (Desert lavender), Malosma laurina (Laurel sumac), Salvia apiana (White sage), Adenostoma fasciculatum (Chamise), Artemisia californica (California sagebrush), Bahiopsis laciniata and Baccharis salicifolia (Mule fat) [35]. The CT micro-basin (4.97 ha) is located in the transition of the southern and northern micro-basins, over a small inter-mountainous valley. At CT, vegetation is similar to the NFS in terms of cover and diversity of plant species. In addition, grasses are present in the wet season and there is a presence of small patches of trees. Throughout the sites, the soil type is classified as Haplic Phaeozem. Plant roots are evident in the shallow topsoil layer at the three sites. Deeper, soil horizons where poorly differentiated until reaching bedrock on the hillslopes sites. Total soil depth is shallower at SFS site (~0.4 m) than at the NFS site (~1 m), because SFS is more affected by fluvial erosion processes due to lack of vegetation cover. Residual parent material is formed by fractured siliceous rocks (granite, granodiorite and tonalite) [38]. It should be mentioned that the soil depth at the CT site was measured with a drilling machine and reach approximately 7.5 m depth.

Terrain Attributes
We employed a digital terrain model with a spatial resolution of 5 m derived from an aerial Light Detection and Ranging (LiDAR) survey [39] to calculate terrain characteristics for the study site. We calculated a set of topographic indices to analyze the influence of flow attributes (hypsometry, slope aspect, curvature), nonlocal morphometric variables (catchment area) and combined morphometric variables (topographic wetness index, [40]) on each micro-basin and our observation sites.

Site Instrumentation and Study Period
Soil moisture was measured with multi-sensor profilers PR2 (Delta-T Devices), with data loggers DL6 or GP2 (Delta-T Devices), recorded every 30 min to obtain θ in m 3 /m 3 . A PR2/4 profile probe with 4 sensors at different depth intervals was used at the SFS site (0.1, 0.2, 0.3 and 0.4 m), while at the CT and NFS sites a PR2/6 profile probe with 6 sensors at different depth intervals were installed (0.1, 0.2, 0.3, 0.4, 0.6 and 1 m). Detailed descriptions of sensor operation have been reported in several studies [41][42][43] and in the sensor operation manual [44]. Installation of the sensors was carried out using a drill bit to place an access tube for the multi-sensor probes. This procedure preserved the natural properties of the soil profile and allowed estimating soil thickness (~0.4 m at SFS and ~1 m at NFS). In addition, a Davis weather station was installed at the CT site to record atmospheric conditions (P and temperature) at 30-min intervals. P was measured with a calibrated precipitation gauge. Datasets were analyzed by event, month and hydrological seasons: wet season (WS; December-March), dry season (DS; June-September) and two transitional seasons (TS1 from wet to dry season; April-May; and TS2 from dry to wet season; October-November). The study period began in the middle of TS1 2014 (TS1-14; May) at the SFS and CT sites, while at the NFS site, data collection began at the end of WS 2015 (WS-15; March). Thus, data availability was ~2.3 years at SFS and CT, while ~1.6 years were available at NFS.

Soil Texture Analysis and Sensor Calibration
For each of the three sites, soil sampling was performed and analyzed to establish textural characteristics and hydraulic properties of each soil layer with hydrometer and sieve analyses (Table Throughout the sites, the soil type is classified as Haplic Phaeozem. Plant roots are evident in the shallow topsoil layer at the three sites. Deeper, soil horizons where poorly differentiated until reaching bedrock on the hillslopes sites. Total soil depth is shallower at SFS site (~0.4 m) than at the NFS site (~1 m), because SFS is more affected by fluvial erosion processes due to lack of vegetation cover. Residual parent material is formed by fractured siliceous rocks (granite, granodiorite and tonalite) [38]. It should be mentioned that the soil depth at the CT site was measured with a drilling machine and reach approximately 7.5 m depth.

Terrain Attributes
We employed a digital terrain model with a spatial resolution of 5 m derived from an aerial Light Detection and Ranging (LiDAR) survey [39] to calculate terrain characteristics for the study site. We calculated a set of topographic indices to analyze the influence of flow attributes (hypsometry, slope aspect, curvature), nonlocal morphometric variables (catchment area) and combined morphometric variables (topographic wetness index, [40]) on each micro-basin and our observation sites.

Site Instrumentation and Study Period
Soil moisture was measured with multi-sensor profilers PR2 (Delta-T Devices), with data loggers DL6 or GP2 (Delta-T Devices), recorded every 30 min to obtain θ in m 3 /m 3 . A PR2/4 profile probe with 4 sensors at different depth intervals was used at the SFS site (0.1, 0.2, 0.3 and 0.4 m), while at the CT and NFS sites a PR2/6 profile probe with 6 sensors at different depth intervals were installed (0.1, 0.2, 0.3, 0.4, 0.6 and 1 m). Detailed descriptions of sensor operation have been reported in several studies [41][42][43] and in the sensor operation manual [44]. Installation of the sensors was carried out using a drill bit to place an access tube for the multi-sensor probes. This procedure preserved the natural properties of the soil profile and allowed estimating soil thickness (~0.4 m at SFS and~1 m at NFS). In addition, a Davis weather station was installed at the CT site to record atmospheric conditions (P and temperature) at 30-min intervals. P was measured with a calibrated precipitation gauge. Datasets were analyzed by event, month and hydrological seasons: wet season (WS; December-March), dry season (DS; June-September) and two transitional seasons (TS1 from wet to dry season; April-May; and TS2 from dry to wet season; October-November). The study period began in the middle of TS1 2014 (TS1-14; May) at the SFS and CT sites, while at the NFS site, data collection began at the end of WS 2015 (WS-15; March). Thus, data availability was~2.3 years at SFS and CT, while~1.6 years were available at NFS.

Soil Texture Analysis and Sensor Calibration
For each of the three sites, soil sampling was performed and analyzed to establish textural characteristics and hydraulic properties of each soil layer with hydrometer and sieve analyses (Table 1).
Hillslope locations (SFS and NFS) had sand fractions ranging from 85-95%, while CT was only 64-87% in terms of sand fraction (Table 1). In the three soil profiles (0.4 m at SFS; 1 m at NFS; 1 m at CT), the sand fraction was the most dominant, with the remaining soil composed of fine sediments (silt and clay) with a lower percentage of clays (1-1.5% in both sites) ( Table 1). Overall soil textural characteristics and relative fractions are consistent in each depth-layer except for the deepest layer at the CT site, composed with 34% of clay fraction (Table 1).
To calibrate the multi-sensors readings, we followed the procedure described in [44] whose methodology is briefly described below. The purpose of calibration is to generate two coefficients (when the soil is moist and when it is dry), which can be used in a linear equation to convert voltage readings into soil moisture from the dielectric properties of the soil ( √ ε). We collected undisturbed soil samples for each depth-layer (16 samples in total) and measured the field voltage using an HH2 moisture meter and an ML3 sensor from Delta-T Devices. Then, we placed each sample inside of a metallic cylinder (volume~300 cm 3 ) and transported these to a laboratory where wet and dry weights were obtained before and after drying at 105 • C for~72 h. The output voltage was then measured for dry conditions to obtain both calibration coefficients. To verify the calibration, we related the dielectric soil measurements ( √ ε) to the sixteen volumetric soil water content samples. The PR2 sensors quoted an error of~±0.05 m 3 /m 3 or less with a soil-specific calibration.

Soil Moisture Analyses
We analyzed the temporal variation of θ in terms of SWS in the soil profile at each site during the study period and for the defined seasons. We performed a numerical integration to calculated SWS ij for site i on day j as: where the 1000 is a conversion factor from m to mm, the subscripts indicate the depth of the measurement (θ in m 3 /m 3 ), a denoting the sensor upper bound, b represents the sensor lowest bound and c is the total length of the sensor rod and d is the length of each profile sensor (0.1 m). We standardized the SWS values by carrying out a transformation: where Z ij is the standardized variable, subscripts i and j represent the site and day, respectively, (Z ij indicates that at site i conditions were relatively wet if Z ij > 0 or dry if Z ij < 0), the temporal mean ( SWS ti and standard deviation (σ ti ) are defined by the total observation days.
Soil moisture analyses at the three sites were related to the precipitation event amount, intensity and monthly accumulation. In total, 14 months were selected with P ranging from 12.6 to 126.6 mm/month. For the remaining months, P pulses were less than 7 mm (29 months). For periods with rainfall, vertical infiltration (Inf in mm) was defined as the total of water amount entering the soil when the wetting front passed the first layer (~0.1 m). Potential recharge (Pr in mm) was then defined as the total of water amount that reached the deepest soil layer in each profile. We calculate both fluxes following [45], where infiltration continues with uniform θ above the wetting front [46] and we estimate infiltration with a modified equation from [46,47], as follows: where θ a is the maximum value of θ recorded after a precipitation pulse (m 3 /m 3 ), θ p represents θ measured preceding the pulse (m 3 /m 3 ), wf is the wetting front depth (m), z symbolizes the terrain surface, 1000 is a conversion factor from m to mm and d is the depth interval (0.1 m). We define the estimation of Inf assuming that θ is uniform at 0.1 m above the depth layer being measured (e.g., at 0.2 m depth, θ is uniform from 0.1 to 0.2 m) and that θ is uniform before the wetting front passage.  (Figure 3a; black bars). Small precipitation events with P < 2 mm/day were the most representative in the entire period (69% of the total events), while larger events occurred less frequently: P > 2 and <5 mm/day (10%), P > 5 and <10 mm/day (11%), and P > 10 mm/day (10%). The maximum rainfall intensity during the period was 15 mm/h, measured in January 2016. Some days with P > 20 mm/day occurred from January to May. where is the standardized variable, subscripts i and j represent the site and day, respectively, ( indicates that at site i conditions were relatively wet if > 0 or dry if < 0), the temporal mean (SWS ti ) and standard deviation (σti) are defined by the total observation days.

Precipitation Analyses
Soil moisture analyses at the three sites were related to the precipitation event amount, intensity and monthly accumulation. In total, 14 months were selected with P ranging from 12.6 to 126.6 mm/month. For the remaining months, P pulses were less than 7 mm (29 months). For periods with rainfall, vertical infiltration (Inf in mm) was defined as the total of water amount entering the soil when the wetting front passed the first layer (~0.1 m). Potential recharge (Pr in mm) was then defined as the total of water amount that reached the deepest soil layer in each profile. We calculate both fluxes following [45], where infiltration continues with uniform θ above the wetting front [46] and we estimate infiltration with a modified equation from [46,47], as follows: where θa is the maximum value of θ recorded after a precipitation pulse (m 3 /m 3 ), θp represents θ measured preceding the pulse (m 3 /m 3 ), wf is the wetting front depth (m), z symbolizes the terrain surface, 1000 is a conversion factor from m to mm and d is the depth interval (0.1 m). We define the estimation of Inf assuming that θ is uniform at 0.1 m above the depth layer being measured (e.g., at 0.2 m depth, θ is uniform from 0.1 to 0.2 m) and that θ is uniform before the wetting front passage.  (Figure 3a; black bars). Small precipitation events with P < 2 mm/day were the most representative in the entire period (69% of the total events), while larger events occurred less frequently: P > 2 and <5 mm/day (10%), P > 5 and <10 mm/day (11%), and P > 10 mm/day (10%). The maximum rainfall intensity during the period was 15 mm/h, measured in January 2016. Some days with P > 20 mm/day occurred from January to May.

Terrain Attributes at the Study Sites
We compared the three sites through the distribution of the accumulated area and elevation using hypsometric curves. Convex up profiles of hypsometric curves show active micro-basins at the two hillslope sites (SFS and NFS). Terrain attributes can be used to describe soil erosion and landslide development, as well as to make inferences on soil water and runoff processes [48,49] (Figure 4a,b). The CT site show an intermediate stage basin (Figure 4a,b), wherein a complex interaction of landscape processes is taking place [50,51]. In about 20% of the area, hillslope processes are taking place, while the remaining 80% indicate the development of fluvial processes where mechanisms are different to those described at the hillslope sites. Slope aspect impacts θ due to differences in insolation (SFS tends to have higher radiation than NFS in the northern hemisphere). Slope aspect at SFS basin was clearly south (about 80% area), with SW and SE slopes covering the remaining area. By contrast, the NFS basin had northerly aspects (40% N, 30% NW and 10% NE), with the remaining area characterized with west-facing orientations. At the CT site, NW aspect is dominant (50%), followed by W (20%) and N (20%), with the remaining 10% corresponding to NE facing-slopes (Figure 4c). The topographic wetness index (TWI) is affected by the magnitude of slope and the upslope contributing area. Figure 4d shown the relative trend of TWI in each micro-basin. We observed relatively lower values at hillslope sites than at CT site due to their steeper slope gradients. Curvature (Cur) variations among the micro-basins are shown in Figure 4e. Overall, positive and negative values of curvature are well balanced, although the SFS micro-basin showed more extreme values than NFS. CT micro-basin values tend to be more lineal on the surface (Cur = 0) (Figure 4e). Cur values at SFS, CT and NFS sites show negative values (SFS < NFS < CT), indicating that surface and subsurface flows converge and decelerate, this features indicate that soil moisture content tends to increase at these points, with more potential at the SFS site.

Terrain Attributes at the Study Sites
We compared the three sites through the distribution of the accumulated area and elevation using hypsometric curves. Convex up profiles of hypsometric curves show active micro-basins at the two hillslope sites (SFS and NFS). Terrain attributes can be used to describe soil erosion and landslide development, as well as to make inferences on soil water and runoff processes [48,49] (Figure 4a,b). The CT site show an intermediate stage basin (Figure 4a,b), wherein a complex interaction of landscape processes is taking place [50,51]. In about 20% of the area, hillslope processes are taking place, while the remaining 80% indicate the development of fluvial processes where mechanisms are different to those described at the hillslope sites. Slope aspect impacts θ due to differences in insolation (SFS tends to have higher radiation than NFS in the northern hemisphere). Slope aspect at SFS basin was clearly south (about 80% area), with SW and SE slopes covering the remaining area. By contrast, the NFS basin had northerly aspects (40% N, 30% NW and 10% NE), with the remaining area characterized with west-facing orientations. At the CT site, NW aspect is dominant (50%), followed by W (20%) and N (20%), with the remaining 10% corresponding to NE facing-slopes ( Figure 4c). The topographic wetness index (TWI) is affected by the magnitude of slope and the upslope contributing area. Figure 4d shown the relative trend of TWI in each micro-basin. We observed relatively lower values at hillslope sites than at CT site due to their steeper slope gradients.   Figure 5 presents daily soil water content for varying depths at the SFS, CT and NFS sites along with daily P during the study period. At the three sites, we found significant differences in daily θ  Figure 5 presents daily soil water content for varying depths at the SFS, CT and NFS sites along with daily P during the study period. At the three sites, we found significant differences in daily θ for each soil layer (ANOVA, p < 0.01). High daily θ was observed in winter months (WS) with great precipitation amounts (151.2 and 198.4 mm in WS-15 and WS-16, respectively). Lower θ amounts occurred during transitional seasons (TS1 and TS2) and the summer dry season (DS) with low or no rainfall. TS1 exhibited daily θ superior than TS2 due to more frequent precipitation. Furthermore, daily θ at the monitored depths almost was completely depleted or removed in the dry summer (DS) and in the transition from dry to wet months (TS2) at all sites, indicating a resetting of dry conditions each year. As expected, there was a greater variation of daily θ in the shallower soil layers that were more exposed to precipitation and evaporation. As [12,52,53] showed, for deeper soil layers, the soil water content generally tended to be more stable. Soil moisture observations at hillslope sites show the active response of vertical and lateral θ after a long rainfall pulses of low intensity (e.g., Figure 5; December 2015 at SFS) and long rainfall pulses with medium-high intensity (e.g., Figure 5; March 2015 and January 2016 at SFS and NFS sites; and January 2016 at CT site) respectively. We found at hillslopes sites an atypical increase of θ, near the bottom of the soil profile at the contact with bedrock (e.g., 0.4 m at SFS and 1 m at NFS) (Figure 5e,g). Interestingly, this mechanism was evident at SFS and NFS showing a sharp increase in daily θ that was superior to amounts in shallower (overlying) layers. This could be indicative of gravitational water quickly drained to the soil profile bottom. If bedrock is fractured, this process could promote rapid infiltration towards the mountain block aquifer from that point. However, it could be also indicating a lateral flow from uphill locations transported along the soil-granite bedrock interface, this mechanism has not been documented frequently in opposing slope aspects with varying plant species. For instance, Penna et al. [54] showed the opposite trend, that daily θ on hillslopes decreases with depth up to 20 cm. Gutiérrez-Jurado et al. [55] also showed evidence of lateral water fluxes in different aspect hillslopes after a large storm, but did not attribute this to the effect of the soil-bedrock interface in promoting subsurface transport.  (TS1 and TS2) and the summer dry season (DS) with low or no rainfall. TS1 exhibited daily θ superior than TS2 due to more frequent precipitation. Furthermore, daily θ at the monitored depths almost was completely depleted or removed in the dry summer (DS) and in the transition from dry to wet months (TS2) at all sites, indicating a resetting of dry conditions each year. As expected, there was a greater variation of daily θ in the shallower soil layers that were more exposed to precipitation and evaporation. As [12,52,53] showed, for deeper soil layers, the soil water content generally tended to be more stable. Soil moisture observations at hillslope sites show the active response of vertical and lateral θ after a long rainfall pulses of low intensity (e.g., Figure 5; December 2015 at SFS) and long rainfall pulses with medium-high intensity (e.g., Figure 5; March 2015 and January 2016 at SFS and NFS sites; and January 2016 at CT site) respectively. We found at hillslopes sites an atypical increase of θ, near the bottom of the soil profile at the contact with bedrock (e.g., 0.4 m at SFS and 1 m at NFS) (Figure 5e,g). Interestingly, this mechanism was evident at SFS and NFS showing a sharp increase in daily θ that was superior to amounts in shallower (overlying) layers. This could be indicative of gravitational water quickly drained to the soil profile bottom. If bedrock is fractured, this process could promote rapid infiltration towards the mountain block aquifer from that point. However, it could be also indicating a lateral flow from uphill locations transported along the soil-granite bedrock interface, this mechanism has not been documented frequently in opposing slope aspects with varying plant species. For instance, Penna et al. [54] showed the opposite trend, that daily θ on hillslopes decreases with depth up to 20 cm. Gutiérrez-Jurado et al. [55] also showed evidence of lateral water fluxes in different aspect hillslopes after a large storm, but did not attribute this to the effect of the soil-bedrock interface in promoting subsurface transport.   Figure 6 summarizes the temporal dynamics of θ through the mean and standard deviation for each hydrological season and depth. Shallower soil depths had more variable θ across each season, while seasonal variations decreased with soil depth, as noted in other studies (e.g., [16,56]). The decrease and variability in θ at each layer was primarily caused by climatic factors (upper layers) and the uptake of soil water by roots [57] (Figure 6, bars) occurring within the season as well. As noted previously, the high and variable amounts of soil water at depth was an indication of rapid lateral transport at the soil-bedrock interface. Since the overlying shallow layers did not exhibit the same behavior, the soil water during wet seasons is likely to derive from uphill locations. This suggests that steep areas with shallow soils in mountain block systems transport water laterally on the soil-bedrock surface [58][59][60] thus effectively bypassing the shallower soils which are more exposed to evapotranspiration.

Temporal Variations of Soil Water Content with Depth
Water 2018, 10, x FOR PEER REVIEW 9 of 18 Figure 6 summarizes the temporal dynamics of θ through the mean and standard deviation for each hydrological season and depth. Shallower soil depths had more variable θ across each season, while seasonal variations decreased with soil depth, as noted in other studies (e.g., [16,56]). The decrease and variability in θ at each layer was primarily caused by climatic factors (upper layers) and the uptake of soil water by roots [57] (Figure 6, bars) occurring within the season as well. As noted previously, the high and variable amounts of soil water at depth was an indication of rapid lateral transport at the soil-bedrock interface. Since the overlying shallow layers did not exhibit the same behavior, the soil water during wet seasons is likely to derive from uphill locations. This suggests that steep areas with shallow soils in mountain block systems transport water laterally on the soil-bedrock surface [58][59][60] thus effectively bypassing the shallower soils which are more exposed to evapotranspiration.

Comparison of Soil Water Storage Profiles Across Sites
Statistical descriptions of soil water storage (SWS) (up to 1 m at NFS and CT sites, and up to 0.4 m at SFS site) in each profile during the hydrological seasons are presented in Table 2

Comparison of Soil Water Storage Profiles Across Sites
Statistical descriptions of soil water storage (SWS) (up to 1 m at NFS and CT sites, and up to 0.4 m at SFS site) in each profile during the hydrological seasons are presented in Table 2. Comparisons across sites are possible for the overlapping periods from WS-15 to DS-16 in terms of average profile water stored conditions. The highest values of SWS occurred during the winter months (rainy season). The NFS and CT sites showed the highest values (39 and 40 mm) over the entire period, and about 37% more SWS than the SFS site (25 mm). This indicates the NFS and CT sites exhibited higher soil water storage capacity over the entire profile. The highest average values of profile water stored during the WS-15 occurred at the hillslope sites (NFS, 61 mm; SFS, 37 mm), while the flat area (CT, 31 mm) exhibited lower values during this hydrological season. As an opposite case, it was observed during the following rainy season (WS-16) that the flat site (CT, 70 mm) had an accumulation of water stored higher than the hillslopes sites (NFS, 56 mm; SFS, 32 mm) ( Table 2). Relating the SWS in the three soil profiles, we found that south-facing slope has persistently lower amounts of SWS than north-facing slope and CT site (Figure 7a; Table 2). This suggests that the aspect and slope effects on soil moisture are consistent with expectations in semiarid Mediterranean regions. Some studies have reported that south-facing slopes tend to be drier with lower infiltration rates [30,[61][62][63]. Subsequently, temporal SWS patterns were compared under the same soil profile depth at the three sites (up to 0.4 m) ( Table 3). The average totals of SWS up to 0.4 m, in this case, reveals that south-facing slope allowed more SWS than north-facing slope and CT in the entire period excepting in WS-16 in which CT was more representative. We attribute SWS amount at SFS site due to the lack of vegetation cover, which permits additional water to reach the soil surface and possible lateral subsurface water transport from uphill locations along the soil-bedrock interface. To further investigate the SWS at hillslope and flat sites, Figure 7 presents the daily variation of SWS profile (Figure 7a), daily variation of SWS up to 0.4 m profile (Figure 7b) and the corresponding standardized values to detect trends equivalent to relatively wetter and drier soil profile regimes (Figure 7c). Dry periods are those whose SWS is lower than the standardized value of Z = 0 (Figure 7c). In general, the soil moisture that accumulates during the winter season dominates the overall wetter conditions of the hydrological seasons in all three sites (Figure 7a,b). An analysis of Pearson's correlation coefficient indicates that there was a positive linear relation between SWS at the sites due to the probable common presence of precipitation pulses, similar soil textural characteristics, topographical features (SFS and NFS) and common presence of plant cover (at NFS and CT sites): SFS-NFS (R 2 = 0.92), SFS-CT (R 2 = 0.61) and NFS-CT (R 2 = 0.67).
The SWS up to 0.4 m was superior at SFS site than NFS for the entire period and greater than CT excluding WS-16 (Figure 7b). At SFS site, soil water content was highly enriched in the deepest layer (0.4 m), potentially due to rapid flow into the layer from local changes in the soil-bedrock configuration. As a result, increased SWS at SFS persisted along the hydrologic seasons.
Water 2018, 10, x FOR PEER REVIEW 11 of 18 The SWS up to 0.4 m was superior at SFS site than NFS for the entire period and greater than CT excluding WS-16 (Figure 7b). At SFS site, soil water content was highly enriched in the deepest layer (0.4 m), potentially due to rapid flow into the layer from local changes in the soil-bedrock configuration. As a result, increased SWS at SFS persisted along the hydrologic seasons. A comparison across stations indicated wetter standardized conditions at SFS during rainfall peaks in winter (WS) and wet-dry transition (TS1) seasons in nearly all occurrences as compared to NFS, but soil water depletion occurs more quickly. Likewise, during a prolonged period in WS-16, CT exhibited higher standardized values than the other sites, suggesting water inputs from local precipitation or precipitation excess from surrounding areas. Soil water depletion occurred more rapidly at SFS and NFS during this period with durations of approximately 1 month (Figure 7c), due to steeper slopes and predominance of sandy fractions.
Overall, the soil moisture dynamics in each profile were consistent with prior Mediterranean climate studies showing that soils are generally saturated in the winter and exhibit recessions during the intervening dry periods (e.g., [4,64]). At our sites, dry soil conditions were persistent and lasted for nearly 70% of the time. Furthermore, if precipitation pulses were smaller than 7 mm during the transition seasons, dry conditions extended through those periods. Interestingly, the dry periods did not seem to reset the soil moisture differences across sites that were established during the wet winter seasons. As a result, the north and south-facing slopes had a wetter standardized conditions during P events followed by a faster depletion and the flat area permitted progressive accumulation of SWS after P events which depleted at a slower rate. Above average wet periods could also lead to much higher wetness at the flat CT site as compared to the hillslope locations.

Infiltration and Potential Recharge During Precipitation Events
Vertical infiltration that reached 0.1 m depth, was estimated at each station for precipitation events larger than 7 mm during the study period. We estimate vertical infiltration in only 14 months, however, the comparison between the totals of the three stations was made since March 2015 to September 2016 (total P = 374 mm) (Figure 8). Figure 8 shown monthly precipitation and monthly A comparison across stations indicated wetter standardized conditions at SFS during rainfall peaks in winter (WS) and wet-dry transition (TS1) seasons in nearly all occurrences as compared to NFS, but soil water depletion occurs more quickly. Likewise, during a prolonged period in WS-16, CT exhibited higher standardized values than the other sites, suggesting water inputs from local precipitation or precipitation excess from surrounding areas. Soil water depletion occurred more rapidly at SFS and NFS during this period with durations of approximately 1 month (Figure 7c), due to steeper slopes and predominance of sandy fractions.
Overall, the soil moisture dynamics in each profile were consistent with prior Mediterranean climate studies showing that soils are generally saturated in the winter and exhibit recessions during the intervening dry periods (e.g., [4,64]). At our sites, dry soil conditions were persistent and lasted for nearly 70% of the time. Furthermore, if precipitation pulses were smaller than 7 mm during the transition seasons, dry conditions extended through those periods. Interestingly, the dry periods did not seem to reset the soil moisture differences across sites that were established during the wet winter seasons. As a result, the north and south-facing slopes had a wetter standardized conditions during P events followed by a faster depletion and the flat area permitted progressive accumulation of SWS after P events which depleted at a slower rate. Above average wet periods could also lead to much higher wetness at the flat CT site as compared to the hillslope locations.

Infiltration and Potential Recharge During Precipitation Events
Vertical infiltration that reached 0.1 m depth, was estimated at each station for precipitation events larger than 7 mm during the study period. We estimate vertical infiltration in only 14 months, however, the comparison between the totals of the three stations was made since March 2015 to September 2016 (total P = 374 mm) (Figure 8). Figure 8 shown monthly precipitation and monthly infiltration at each site. As a result, at the three sites, monthly totals shown that between 28 and 42% of monthly P was vertically infiltrated into the top soil layer profiles (SFS < CT < NFS) and between 72 (SFS) and 58% (NFS) of the precipitation was lost to other superficial processes such as rainfall interception (not at SFS), evaporation and runoff (SFS > CT > NFS).
Water 2018, 10, x FOR PEER REVIEW 12 of 18 infiltration at each site. As a result, at the three sites, monthly totals shown that between 28 and 42% of monthly P was vertically infiltrated into the top soil layer profiles (SFS < CT < NFS) and between 72 (SFS) and 58% (NFS) of the precipitation was lost to other superficial processes such as rainfall interception (not at SFS), evaporation and runoff (SFS > CT > NFS). In Figure 9a,b, we show the nineteen precipitation events that occurred in the fourteen months under analysis. The cumulative precipitation, precipitation intensities, duration and vertical infiltration for each precipitation event was estimated since March 2015. It is interesting to note the totals of the precipitation-infiltration analysis revealed that water loss at ~0.1 m depth was lower at SFS than NFS and CT (between 15 and 20%,) this mechanism could be driven by a transpiration process at NFS and CT.  In Figure 9a,b, we show the nineteen precipitation events that occurred in the fourteen months under analysis. The cumulative precipitation, precipitation intensities, duration and vertical infiltration for each precipitation event was estimated since March 2015. It is interesting to note the totals of the precipitation-infiltration analysis revealed that water loss at~0.1 m depth was lower at SFS than NFS and CT (between 15 and 20%,) this mechanism could be driven by a transpiration process at NFS and CT.
Water 2018, 10, x FOR PEER REVIEW 12 of 18 infiltration at each site. As a result, at the three sites, monthly totals shown that between 28 and 42% of monthly P was vertically infiltrated into the top soil layer profiles (SFS < CT < NFS) and between 72 (SFS) and 58% (NFS) of the precipitation was lost to other superficial processes such as rainfall interception (not at SFS), evaporation and runoff (SFS > CT > NFS). In Figure 9a,b, we show the nineteen precipitation events that occurred in the fourteen months under analysis. The cumulative precipitation, precipitation intensities, duration and vertical infiltration for each precipitation event was estimated since March 2015. It is interesting to note the totals of the precipitation-infiltration analysis revealed that water loss at ~0.1 m depth was lower at SFS than NFS and CT (between 15 and 20%,) this mechanism could be driven by a transpiration process at NFS and CT.  Furthermore, we found the fraction of infiltration that passed to underlying soil layers required precipitation events greater than~10.5 mm for infiltration fronts to pass below the shallow soil layer (underneath 0.1 m; about~10% of annual P events). Successive events with accumulations between 48 mm (December 2014) and 104 mm (January 2016) (Figure 9a) were needed for θ to increase in deeper layers ( Figure 10). When precipitation events were intense (e.g., March 2015, 12 mm/h; 6 January 2016, 15.2 mm/h; 31 January 2016, 12.6 mm/h; Figure 9b; white circles), infiltration was limited substantially because the precipitation rate was greater than the capacity of the soil to transport vertical flow (Figure 10). On the other hand, during a sequence of low-intensity pulses (e.g., December 2014; Figure 9b), infiltration occurred progressively and the wetting front passed gradually to lower layers [60,65]. This mechanism was appreciated mainly during DS and the intermediate seasons TS1 and TS2 without the soil water content reaching the deepest soil layer ( Figure 10). Furthermore, we found the fraction of infiltration that passed to underlying soil layers required precipitation events greater than ~10.5 mm for infiltration fronts to pass below the shallow soil layer (underneath 0.1 m; about ~10% of annual P events). Successive events with accumulations between 48 mm (December 2014) and 104 mm (January 2016) (Figure 9a) were needed for θ to increase in deeper layers (Figure 10). When precipitation events were intense (e.g., March 2015, 12 mm/h; 6 January 2016, 15.2 mm/h; 31 January 2016, 12.6 mm/h; Figure 9b; white circles), infiltration was limited substantially because the precipitation rate was greater than the capacity of the soil to transport vertical flow ( Figure 10). On the other hand, during a sequence of low-intensity pulses (e.g., December 2014; Figure 9b), infiltration occurred progressively and the wetting front passed gradually to lower layers [60,65]. This mechanism was appreciated mainly during DS and the intermediate seasons TS1 and TS2 without the soil water content reaching the deepest soil layer (Figure 10).   Figure 10d, dark blue bar). We calculated that these two precipitation events in total generated 34.4 mm considered as potential recharge at SFS site   (Figure 10c,d; dark blue bars "d6" for CT and NFS). Overall, we estimated that on average about 9.5% of the accumulated precipitation (~374 mm; from March 2015 to September 2016) could contribute to the recharge of the mountain block aquifer at the two sites with opposing aspects (SFS and NFS; 34.5 mm average potential recharge). It is worth mentioning that SFS with less precipitation amounts allows infiltration water to reach the deepest layer of the soil profile, leading to greater probability of potential recharge in that site (e.g., March 2015 = 53.8 mm; Figure 10b) than the opposite NFS (with this amount of P, wetting front could not reach to the last layer "d6").

Conclusions
This study presents a comparison of soil water content measurements, soil water storage, infiltration and potential recharge at three sites representing different topographic positions in a mountain block within a semiarid Mediterranean climate ecosystem. Results of the study reveal: 1.
The evolution of soil water content showed a strong variability on the opposing hillslopes (SFS and NFS) principally in the winter and wet-dry transition seasons in the shallowest depths. Likewise, the hillslope sites exhibit a sudden increase in soil water content during P peaks followed by rapid depletion. In contrast, the flatter site (CT) showed slower soil water content decreases and the accumulation of water from upland areas during significant events. During high-intensity P events, sites with opposing aspects reveal an increase of soil water content at the soil-bedrock interface (~0.4 m in SFS;~1 m in NFS) suggesting lateral subsurface fluxes, while vertical soil infiltration decreases noticeably, signifying the production of surface runoff.

2.
The seasonal patterns of SWS were similar at the three sites. Precipitation in winter (WS) and wet-dry transition months (TS1) replenish SWS at each location, which is depleted in the summer-fall months most notably at the hillslope sites. NFS and CT sites exhibit relatively high SWS due to their larger soil profile thickness, but SWS comparison until 0.4 m revealed higher values at SFS sites. This is attributed to the lack of vegetation cover, which permits additional water to reach the soil surface and infiltrate in vertical and horizontal directions, and the possible effect of lateral subsurface water transport from uphill locations at the soil-bedrock interface. 3.
The relation of SWS at hillslope sites showed strong positive correlation. Soil water storage was completely depleted in the soil profiles at hillslope sites after WS, meaning that there is no water available in the soil until the next WS. Water deficit occurred in summer months (DS) and dry-wet transition months (TS2) and lasted for nearly 70% of the study period at the three sites.

4.
Potential recharge occurred only in WS with P events greater than 50 mm/month at SFS site and greater than 120 mm/month at NFS site, indicating that soil depth and lack of vegetation cover play a critical role in the transport of water towards the soil-bedrock interface. We calculate that on average, around 9.5% of accumulated precipitation (~374 mm; from March 2015 to September 2016) could contribute to the recharge of the aquifer by the two sites with opposing aspects (~34.5 mm).
The results presented here provide insights into the dynamics, storage and transport of soil water content at different temporal scales. This is the first research work carried out with these objectives in the region, and although results are spatially limited, they allow inferences to be made about the potential contribution of the hillslopes sites towards aquifer recharge in the mountain block. Funding: This study was funded by the Consejo Nacional de Ciencia y Tecnología (CONACYT, Grant 235295) and the Fondo Institucional de Fomento Regional para el Desarrollo Científico, Tecnológico y de Innovación (FORDECYT, Project 146606).