Estimation of Precipitation Fraction in the Soil Water of the Hillslope Vineyard Using Stable Isotopes of Water

Abstract

This paper presents research related to the estimation of the precipitation fraction in the soil water of a sloped vineyard at the SUPREHILL Critical Zone Observatory (CZO) in Zagreb, Croatia. Numerous investigations have shown that exploration of hillslope soils can be very challenging due to the existence of heterogeneity and different soil properties, as well as due to anthropogenically induced processes, which can affect precipitation infiltration and soil water flow. Within this research, physicochemical soil properties, soil water content (SWC), and isotopic composition of soil water and precipitation (δ²H and δ¹⁸O) have been examined. The isotopic signature of soil water was monitored in 24 points, at 4 depths, throughout the hillslope vineyard. Soil water isotopic composition from all monitoring points coincided with the Local Meteoric Water Line (LMWL), with almost no variability at 100 cm depth, which was consistent with the smallest variation of SWC at 80 cm depth and indicated that most of water mixing takes place in the shallower part of the hillslope. Results suggested the existence of heterogeneity, uneven erosion processes in the footslope of the observed vineyard, and different infiltration patterns. Fractions of precipitation varied significantly depending on the depth and position in the vineyard, from approximately 1% up to 98%, where more precipitation fraction has been determined in the surface and subsurface runoff. Additionally, statistical analysis and a more detailed evaluation of precipitation fractions at the 40 cm depth, where wick lysimeters are installed, have shown that C_org content is related to the silt fraction, while the first results indicate that the infiltration patterns were dependent on the common influence of all observed physicochemical properties.

1. Introduction

Investigation of hillslope soils can be very challenging. Numerous hydrological studies of soils in the last years have been focused on problems related to the determination of different types of flows in soil, as well as to human practices, processes, slope, and soil characteristics that affect soil water movement [1,2,3,4,5,6,7,8]. In general, infiltration of precipitation presents a complex process that can be described as preferential or piston flow. Preferential flow water is channeled through more permeable pathways, while with piston flow, water from recent precipitation forces the older soil water to flow downward [9].

It has been shown that soil porosity, soil water content, soil structure, soil texture, soil compaction, and vegetation can affect infiltration of surface water [9,10], but also that soil hydrophobicity can reduce water infiltration into the soil [11]. Organic matter can also be important because some research showed that soil organic carbon, which corresponds to the high levels of organic matter, can enhance permeability and water availability, as well as hydraulic conductivity of soil by improving aggregate stability and porosity [12,13,14,15]. Previous results also showed that subsurface hillslope flow is different in wet and dry conditions. In wet conditions, subsurface lateral flow prevails, while in dry conditions, flow is dominantly vertical [16,17,18]. Furthermore, it has been shown that soil erosion can play a crucial role in hillslope hydrology, causing soil heterogeneity [5], where the footslope can have a deeper loose layer compared to the hilltop [8]. The soil heterogeneity can also cause preferential flow on the whole hillslope, local preferential flow, or surface runoff [2]. Additionally, to determine water flow and solute transport, chemical tracer experiments [19,20] as well as dye tracers have been used [3,21].

Stable isotopes of oxygen and hydrogen in soil water have been used in numerous studies of soil and unsaturated zone [22,23,24,25,26], while some of them were related to the hillslopes [27,28]. In general, there are two types of water that should be distinguished: mobile and bulk water. Mobile water should follow the one of the precipitation but can have high variability of water isotopic composition in a defined soil depth over space and time [29,30]. However, some research showed that macropore flow can have a different isotopic signature with respect to the precipitation [31] and can possibly indicate mixing between water from the soil matrix and the macropore flow [32]. It has been shown that the kinetic fractionation usually cannot be seen in mobile water, but can be seen in bulk soil water, and is globally limited to the upper 30 cm of the soil in most soil water isotope hydrology studies [32,33]. It must be emphasized that local precipitation can have strong temporal variation in isotope composition due to temperature and different isotope effects [34], which can consequently result in isotopic heterogeneities [35,36] where small pores may contain water similar to older precipitation, while larger pores could consist of more recent precipitation [37,38]. The newest research has shown that the isotopic composition of soil bulk water should be different from evaporating water, although the difference can be too small to affect the evaporative water loss calculation [39]. Furthermore, it has been shown that soil tension, water–mineral interaction, and soil particle surface adsorption can have an influence on the soil water isotopic signature and its variations [40,41,42,43,44].

In the last few years, much research has been conducted in the wider Zagreb area. Most of the research was focused on the identification of the relationship between Zagreb aquifer and Sava River using statistical [45,46] and isotope techniques [47,48,49]. Soil and unsaturated zone research were focused to the estimation of water flow and the transport of potentially toxic metals [50], and the evaluation of physical and chemical properties on soil permeability [51]. Newest soil research is related to the sorption of cadmium, zinc, and cooper in the dominant soils of the Zagreb aquifer [52], the definition of nickel and chromium origin in Fluvisols [53], as well as to the estimation of precipitation infiltration through the unsaturated zone into the Zagreb aquifer using cross-correlation analysis and stable isotopes of water [54]. Regarding investigated hillslope vineyard, the first initial isotopic insight suggested that preferential flow can be expected to a maximum depth of 80 cm [55], while it has been also found that erosion-affected soil structural properties can influence soil hydraulic properties and soil water dynamics [8].

The objectives of this study were to: (1) investigate soil water isotopic signature in 24 monitoring points at 4 different depths in the hillslope vineyard, (2) estimate the precipitation fraction in soil water using the isotope mass balance mixing model, and (3) determine soil properties that influence water infiltration.

2. Materials and Methods

The research area (SUPREHILL Critical Zone Observatory (CZO); https://sites.google.com/view/suprehill/) ((accessed on 11 November 2022) is located in the agricultural sloped area within the experimental field Jazbina in Zagreb, Croatia (Figure 1). It is a vineyard with rows separated by grassed inter-row areas (2 m wide). The soil type is classified according to World Reference Base (WRB) classification system as Dystric Luvic Stagnosol [8]. The hillslope is separated into three segments: hilltop, backslope, and footslope. The sloping between the segments was determined using an unmanned aerial vehicle (UAV) and Agisoft Metashape software (Agisoft LLC., St. Petersburg, Russia, ver. 1.7.5), with the sloping between the hilltop and backslope being 17.5%, and the sloping between backslope and footslope was 25.4%. Additional to the vineyard-row-oriented south-east sloping, perpendicular-to-row sloping is present, with 7.3% at the hilltop and backslope and 4.4% at the footslope.

In Figure 1 locations of all the used monitoring points can be seen. Within this research, four different types of instruments have been installed for sampling soil water. At the surface level, for capturing surface water runoff, self-constructed instruments were installed at the footslope (2 × 2 m) in 3 repetitions (vineyard rows). At the 40 cm depth, for the measurement of near surface drainage, passive wick lysimeters were used. They were installed at the hillslope, backslope, and footslope, between two vines, in three repetitions. They have dimensions of 250 × 250 × 40 mm. Their surface is covered with a filter mesh (1 × 1 mm) to prevent clogging, while fiberglass is placed inside the lysimeter. At 60 cm depth, a self-constructed instrument for collecting subsurface runoff was installed, also in 3 repetitions in rows at the footslope. For sampling water from 100 cm depth, suction probes (UGT Umwelt-Geräte-Technik, Müncheberg, Germany) with a ceramic cup were installed at the hillslope, backslope, and footslope, also in 3 repetitions (vineyard rows). To summarize, soil water was sampled from depths of 20 cm, 40 cm, 60 cm, and 100 cm. From 20 cm (SR_2, SR_4 and SR_6) and 60 cm (SS_1, SS_2 and SS_3) depth water was sampled only at the footslope, while from 40 cm (WL_2, WL_6, WL_10, WL_14, WL_18, WL_22, WL_26, WL_30, WL_34) and 100 cm (SP_1 to SP_9) depth water was sampled additionally at the hillslope and backslope. In this research, the closest monitoring points in rows were used for data evaluation, resulting in a total of 24 monitoring points of soil water plus rain sampler (RS). For the construction of the Local Meteoric Water Line (LMWL) and further analysis, all data related to the precipitation isotopic signature have been used except for August 2022, due to technical issues with rain sampler and significant evaporation of the sample which resulted in negative d-excess and significant shift from the LMWL.

Data used in this research can be generally divided into four groups: meteorological data, soil water content (SWC), soil physicochemical parameters, and stable isotopes of oxygen and hydrogen in precipitation and soil water. All data are related to the period from February 2021 to September 2022. Meteorological and SWC data are presented as daily data, and stable isotopes are presented as monthly data, while physicochemical parameters were defined at the establishment of the SUPREHILL CZO. All data was processed, and basic statistical parameters (average, median, maximum, minimum, standard deviation-SD, and coefficient of variation–CV) were calculated with Microsoft Excel, while Tibco Statistica (version 14.0.0.15) was used for the construction of boxplots and correlation analysis.

Within this paper, the sum of daily precipitation and evapotranspiration data is presented, as well as daily average air temperature. Daily reference evapotranspiration was calculated via the Penman–Monteith equation [56], while all meteorological parameters were gathered using a meteorological station (ATMOS41, METER Group, Inc., Pullman, WA, USA) installed in the backslope position of the hillslope at a 2 m height.

SWC was measured at the hourly intervals with capacitance sensors (TEROS 10/TEROS 12, METER Group, Inc., Pullman, WA, USA). Sensors were installed in vineyard rows, in between 2 vines, in 3 repetitions (3 different vineyard rows) at 20, 40, 60, and 80 cm depth, to capture the soil water dynamics near the installed soil water sampling instruments. Sensors at 40 and 80 cm depth were installed at the hillslope, backslope, and footslope, near the wick lysimeters and suction probes, while at 20 and 60 cm, they were installed at the footslope near the surface and subsurface runoff instruments.

Sampling of disturbed soil samples was performed at 3 depths (0–30, 30–60, and 60–90 cm) at the hillslope, backslope, and footslope, in 3 repetitions. Granulometric soil composition analysis was performed according to [57], while organic carbon content (C_org) was determined using sulfochromic oxidation according to [58]. Values of the bulk density have been estimated using HYPROP-FIT for the hilltop, backslope, and footslope for 4 depths (0–20, 20–40, 40–60, and 60–90 cm) and 3 vineyard rows [59], while, for this research, only the average values for depth up to 40 cm have been used. According to the United States Department of Agriculture soil classification system (USDA) [60], soils have been classified as silt loam or silty clay loam for all three depths [55].

Stable isotopes of oxygen and hydrogen (δ¹⁸O and δ²H) in precipitation and soil water were determined at the Laboratory for Spectroscopy of the Faculty of Mining, Geology and Petroleum Engineering, University of Zagreb, using laser absorption spectroscopy (Liquid Water Isotope Analyzer LWIA-45-EP, Los Gatos Research, San Jose, USA). It has been shown that analyzers which use laser spectroscopy are fully suitable for the high precision isotopic analysis of natural waters [61]. Data were analyzed using Laboratory Information Management System (LIMS for lasers 2015) [62]. The results are reported as δ values per mill (‰) relative to the standard [63,64,65]:

$$\delta =\frac{R_{sample}}{R_{standard}}-1$$

(1)

where R_sample and R_standard stand for the isotope ratio in the sample (R = ²H/¹H and R = ¹⁸O/¹⁶O) and standard material, respectively. This means that δ value expresses the relative difference in ‰ of the isotope ratio R in the sample to the defined (known) isotope ratio in the international reference material (standard). Within this research, all results are presented with respect to VSMOW (Vienna Standard Mean Ocean Water), with measurement precision of ±0.19‰ for δ¹⁸O and ±0.9‰ for δ²H. Deuterium excess (d-excess) has been defined according to Dansgaard [66] with a global mean slope of 8. For the δ¹⁸O and δ²H analysis of cumulative monthly precipitation samples, a rain collector RS-1 (Palmex Ltd., Zagreb, Croatia), whose design allows evaporation-free rain sampling [67] and is suitable for most hydrology investigations [68], was installed at the top of the hillslope at 1 m above ground.

In this research, the isotopic signature of soil water has been evaluated in 24 monitoring points. Soil water from the deepest observed soil depth (100 cm) was approximated as non-mobile water due to small variability of the isotopic composition (presented in the results and discussion section), which is probably related to the bulk water and/or older precipitation, i.e., precipitation that fell before the investigated time period. Small variability of the isotopic composition also suggests that most of the water mixing in the hillslope vineyard is occurring in the first 100 cm depth (Figure 2). The isotopic signature of recent precipitation (the one that fell within the investigated time period) was taken as the only input of recharge, while precipitation infiltration was estimated through calculation of precipitation fractions at 20 cm, 40 cm, and 60 cm depth using the two-end member mass balance equation, where the sum of the end member contributions is usually expressed as fractions (f) equal to 1 [47,69,70,71]. In general, infiltration presents the process by which water that falls on the surface enters the soil. However, it must be emphasized that the fraction of recent precipitation in the observed monitoring point presents the part of the water which has infiltrated into the soil, but which has been captured by the instrument. From that perspective, it presents infiltrated water, but probably does not present the total amount of the infiltrated water. Precipitation fractions were estimated by two-component mixing model using following equations:

f_snmw + f_p = 1

(2)

f_snmw × δ¹⁸O_snmw + f_p × δ¹⁸O_p = δ¹⁸O_smw

(3)

where f_snmw and f_p are the soil non-mobile water and precipitation fractions (end members, Figure 2), δ¹⁸O_smw is the isotopic composition of oxygen in the soil water of the investigated monitoring point (isotopic composition is the consequence of mixing between end members), while δ¹⁸O_snmw and δ¹⁸O_p are the isotopic compositions of oxygen in the soil non-mobile water and precipitation, respectively.

3. Results and Discussion

3.1. Evaluation of Meteorological Parameters

Variations of precipitation, evapotranspiration, and air temperature are shown in Figure 3. In the observed period, 247 days had precipitation. Daily precipitation varied from 0.017 mm to 54.98 mm, with average of 5.27 mm. Average daily temperature was 13.91 °C and ranged from −5.14 °C to 29.9 °C. According to the values of average yearly air temperature, which can be seen in the Croatian Meteorological and Hydrological Service [72], year 2021 in the study area has been defined as a warm year, while year 2022 has been defined as an extremely warm year. On the other hand, yearly sums of precipitation suggest normal, expected values in both 2021 and 2022. Daily reference evapotranspiration reached up to 6.30 mm, with an average of 2.49 mm per day. The total daily reference evapotranspiration during the observed period was 1510.4 mm, higher than the total precipitation of 1301.79 mm. In the most monitoring points, soil water was only captured in the autumn to spring period due to the existence of warm and extremely warm summer periods. This is consistent with the measured volumes presented in Figure 4. In general, within the summer months, isotopic composition of soil water was measured only for surface runoff. In total, without considering samples related to the precipitation, for 228 samples the isotopic composition has been defined. Due to the lack of data, the mass balance model has been applied using average values calculated based on months where at least half of the monitoring points had measured isotopic signature, resulting in the usage of 191 of 228 samples. Volume from suction cups is not presented due to challenging sampling of soil water in the deepest observed layer, i.e., very small amounts were available.

3.2. Variability of Soil Water Content and Granulometric Composition

Variation of the SWC is presented in Figure 5, while basic statistical parameters for all monitoring points are shown in

Table 1. Basic statistical parameters related to the soil water content (SWC) in 24 monitoring points.

Monitoring Point	Average	Median	Maximum	Minimum	SD	CV
SR_2	0.26	0.28	0.37	0.15	0.06	0.25
SR_4	0.27	0.29	0.34	0.18	0.05	0.19
SR_6	0.31	0.33	0.38	0.20	0.06	0.18
SS_1	0.31	0.33	0.36	0.21	0.05	0.17
SS_2	0.32	0.34	0.39	0.22	0.06	0.18
SS_3	0.32	0.35	0.40	0.21	0.07	0.21
WL_2	0.32	0.35	0.42	0.18	0.08	0.25
WL_6	0.31	0.35	0.41	0.20	0.07	0.23
WL_10	0.34	0.37	0.41	0.23	0.06	0.17
WL_14	0.31	0.34	0.41	0.20	0.07	0.23
WL_18	0.29	0.33	0.41	0.18	0.07	0.25
WL_22	0.33	0.35	0.42	0.21	0.07	0.21
WL_26	0.31	0.34	0.41	0.18	0.07	0.21
WL_30	0.34	0.36	0.43	0.21	0.07	0.21
WL_34	0.34	0.37	0.42	0.22	0.07	0.20
SP_1	0.32	0.35	0.37	0.27	0.04	0.13
SP_2	0.30	0.33	0.37	0.23	0.05	0.16
SP_3	0.32	0.33	0.41	0.21	0.08	0.24
SP_4	0.33	0.35	0.37	0.25	0.04	0.12
SP_5	0.31	0.34	0.36	0.24	0.05	0.15
SP_6	0.33	0.35	0.36	0.26	0.03	0.10
SP_7	0.31	0.33	0.38	0.20	0.06	0.19
SP_8	0.29	0.30	0.37	0.20	0.06	0.21
SP_9	0.34	0.36	0.37	0.27	0.04	0.12

. It can be clearly seen that probes in all locations and depths showed similar reaction in the summer months, i.e., SWC drastically dropped. Additionally, it can be seen that variation of SWC was not identical if the location on the hillslope and depth of the sensor are taken into account. For example, SR_2 (SENS_7) had the lowest average (0.26 cm³ cm⁻³) SWC, but one of the biggest CV (0.25). Some sensors above suction probes (SP_3 and SP_8) had high (0.24 and 0.21, respectively) CV values, while some (SP_9) had one of the highest average (0.34) SWC and one of the lowest CV (0.12). However, when evaluating average values of all sensors at the same depths, the results suggested that surface runoff corresponded to the lowest average and median value, as well as the lowest minimum value. Subsurface runoff had similar average and median values, but a lower standard deviation and CV value, when comparing them to the sensors at 40 cm depth. The deepest sensors had the highest minimum values as well as the lowest standard deviation and a much lower CV. This suggested that variation of the SWC in 100 cm depth, where suction probes were installed, was even lower. Additionally, this corresponds to the results of the previous research, where results suggested that most of the preferential flow can be expected at up to 80 cm depth [55]. As shown in the Figure 1, different slopes have been observed in the investigated hillslope vineyard. The slope between the hilltop and backslope is 17.5%, while between backslope and footslope it is 25.4%. In addition to the vineyard-row, which has south-east sloping, perpendicular-to-row sloping has been also observed, with 7.3% at the hilltop and backslope, and 4.4% at the footslope. When observing all locations in the vineyard, using data from both 40 and 80 cm depths (WL and SP monitoring points), the average standard deviation of soil water content at all locations varied from 0.05 to 0.07, while CV varied from 0.15 to 0.21. Although differences in slope have been determined, preliminary results suggest that slope does not have a lot of impact on the variability of the SWC, which is consistent with the similar research where it was shown that soil water retention in slopes was mostly related to the soil properties [5].

Granulometric composition at three depths in all nine locations, as well as values of C_org, are presented in Figure 6. In general, results showed that silt fraction is dominant at all positions, that sand has the highest average occurrence in the backslope and goes up to a maximum of 19%, and that clay content varies from 14% to 40%. C_org shows a similar trend in all locations, i.e., a decrease with depth, with the highest average values in the footslope. As previously mentioned, values of bulk density are presented in detail in Krevh et al. [59]. In this research, only average values of bulk density to a depth of 40 cm were used for the correlation analysis and interpretation purposes. Values of bulk density varied from 1.46 g/cm³ to 1.59 g/cm³, with the highest average values in the backslope and smallest in the hilltop.

3.3. Isotopic Composition and Water Mixing in the Hillslope Vineyard

Stable isotopes of oxygen and hydrogen from the precipitation and soil water are presented in Figure 7, while their variability can be additionally seen in Figure 8, where boxplots for δ²H, δ¹⁸O and d-excess are shown. The slope of the LMWL line is similar to the LMWL Zagreb [73] and LMWL Velika Gorica [53], which is expected due to the small distance between all three monitoring points.

In Figure 7a, results suggested a similar isotopic pattern between the precipitation and surface runoff, which is reasonable to expect due to soil water which is captured in shallowest depth (20 cm). Values of δ²H of the surface runoff mostly varied from −40‰ to −70‰, while δ¹⁸O values in general varied from −7‰ to −10‰. D-excess values varied from 8‰ to 13‰ (Figure 7a and Figure 8). As shown in Figure 7b, the isotopic signature of subsurface runoff also follows the LMWL, but, in Figure 7b, it can be clearly seen that the variability in the monitoring point SS_1, also located in the northwestern part of the footslope, was much lower with respect to the other two points. Subsurface runoff showed an isotopic signature mostly between −50‰ and −75‰, and between −8‰ and −11‰ for δ²H and δ¹⁸O, respectively. D-excess, in most cases, varied from 9 to 11‰. In general, results suggested smaller variability in the northwestern part of the footslope in monitoring point SR_2 at 20 cm depth and SS_1 at 60 cm depth. There are numerous factors that can present a possible reason for that. Firstly, it is possible that, in this part of the vineyard, C_org has big influence on water flow, due to the determined highest values at all three depths (Figure 6). It has been shown that soil organic carbon can enhance permeability and water availability [13], as well as enhance hydraulic conductivity by improving aggregate porosity and stability [12,13,14,15]. However, in this location, except for the highest values of C_org, there is the highest average value of silt and lowest average value of clay (if all data from the whole vineyard are considered). The lowest share of clay could enable faster flow, in which less water could be captured with monitoring equipment and, in the end, generate less variation in the isotopic signature. This corresponds to the lowest average SWC for SR_2 and SS_1, which can be seen in Table 1, in comparison with other monitoring points which capture surface and subsurface flow. Furthermore, this also corresponds to the fact that, in the positions SR_2 and SS_1, much less volume has been captured (Figure 4). Additionally, it has been shown that footslope can have a deeper loose layer compared to the hilltop [8], which can cause heterogeneity and influence soil water dynamics. In this case, it is very interesting to see that, in the northwestern part of the footslope, there are the lowest values of bulk density in the first 40 cm depth, and the highest in the deepest parts. On the other two points in the footslope, values of bulk density are totally opposite with depth, as shown by Krevh et al. [59]. In the end, the influence of the slope of the vineyard on the surface and subsurface flow should not be neglected, which is from the northwest to southeast (Figure 1), although results from this study indicate that slope does not have a lot of impact on the variability of the soil water content, which corresponds to previous research where it was shown that the retention of soil water on the slope is more influenced by soil properties [5].

Isotopic signature from wick lysimeters follows the LMWL but shows much less variability than surface and subsurface runoff (Figure 7c). Values of δ²H and δ¹⁸O varied from approximately −50‰ to −60‰ and −8‰ to −9‰, respectively. D-excess values in general varied from 10‰ to 11‰. The smallest variation, within the limit of the precision of the instrument, can be seen for the isotopic signature of soil water from the suction probes (Figure 7d). This is expected because they are capturing soil water at 100 cm depth, and it is consistent with, in most cases, the lowest SWC variation at sensors located at 80 cm depth, as shown in Figure 5 and Table 1. Average and median δ²H values of all analyses (n = 53) were −51.1‰ and −51.00‰, respectively, while the average and median of δ¹⁸O value were −7.77‰ and −7.74‰. D-excess values, in almost all cases, varied from 10‰ to 12‰. In general, d-excess values from all water samples, especially those related to the soil water, suggest that evaporation did not have a significant impact on the isotopic composition. Although very similar, in Figure 8 it can be clearly seen that in the footslope isotopic signature shows slightly higher variation (SP_7, SP_8, and SP_9). All results together suggest that soil water from the suction probes presents mostly non-mobile water, probably associated with older precipitation and/or bulk water, and not related to the recent precipitation, i.e., precipitation which fell within the investigated time period. It must be emphasized that these kinds of instruments are usually used for sampling mobile water, while, for the purpose of sampling bulk soil water methods such as cryogenic vacuum extraction [74,75] and other laboratory or in situ techniques are used [76,77]. However, increased intention related to the differentiation between sampling of bulk and mobile water has been pronounced in recent research [33], as well as the necessity to explore differences between their isotopic signatures [78]. Within this research, it is also not clear whether the soil water captured by suction probes presents bulk water or older precipitation, or a mixture of water between them, which points to the need for more detailed research in future. On the other side, it is clear that recent precipitation did not have a lot of influence on the isotopic composition of soil water at 100 cm depth, at least in the observed time period, which indicates that water mixing mainly took place in the shallower parts of the vineyard.

In order to define precipitation fractions in soil water at different monitoring points, a two-component mixing model using δ¹⁸O values has been applied. It was performed in a way that soil water captured from suction probes was defined as non-mobile water, while precipitation was the only input, i.e., the only source of recharge. As mentioned before, not all months were included in the mixing model due to lack of data. In the end, it was decided to use only months where at least half of the monitoring points, i.e., 12 locations, had determined isotopic composition. These months generally coincide with the period from September to May, for which most of the soil water isotopic composition was available (191 of 228). For the δ¹⁸O value of the precipitation, the average value from the used months was calculated, i.e., −9.65‰. All data from suction probes have been checked for outliers and extreme values using a box-plot, which resulted in the exclusion of five values and the final average value of −7.73‰. For surface runoff, subsurface runoff, and wick lysimeters, the average values from the selected time period have been used for each location separately (

Table 2. Results of two-component mixing model.

Monitoring Point	Average δ18O (‰)	Precipitation Fraction in Soil Water (%)
SR_2	−9.27	80.48
SR_4	−9.36	84.83
SR_6	−9.60	97.50
SS_1	−7.90	8.88
SS_2	−9.61	97.97
SS_3	−8.69	50.24
WL_2	−8.15	21.98
WL_6	−8.33	31.55
WL_10	−8.63	46.98
WL_14	−8.31	30.45
WL_18	−8.29	29.16
WL_22	−7.91	9.84
WL_26	−8.57	44.11
WL_30	−8.62	46.29
WL_34	−7.75	1.16

). Results suggest different values of precipitation fraction within the soil water. The highest values can be seen for the surface runoff, as expected, from approximately 80% up to more than 97%. Precipitation fraction within subsurface runoff varies greatly, from about 9% in the northwestern part of the footslope up to almost 98% percent in the central part of the footslope, which suggests a very heterogeneous area as well as different influences of soil physicochemical parameters and erosion processes on the water flow. Precipitation infiltration to wick lysimeters also shows different patterns within the vineyard, from fractions of more than 1% up to almost 47%. In general, a very small precipitation fraction can be seen at the backslope of the vineyard, which is especially related to the point WL_22, which had the smallest average values of C_org as well as the highest share of sand. An additional very interesting isotopic signature has been observed in point WL_34, located in the southeastern part of the footslope. Within this point, the smallest precipitation fraction is observed, totally different from the other two points in the footslope, which suggests much higher precipitation infiltration, i.e., about 45%. The possible reasons for that are related to the highest clay content of the footslope when observing first 30 cm depth, as well as the slightly higher average bulk density in first 40 cm depth. Previous results showed that proportion of clays in soil determines the occurrence of soil water repellency [79] and that soils with a predominance of clay have stronger water repellency than coarse-textured soils [80].

Results related to the estimation of the precipitation fractions captured by wick lysimeters suggested great variability and dependence to different physicochemical parameters in different parts of the vineyard, which is expected due to heterogeneity and more and less pronounced erosion, especially in the footslope part. In order to statistically define the relationship between precipitation infiltration and physicochemical parameters, correlation analysis has been performed using values for first 30 cm depth, which are related to the granulometric composition, and C_org, while average bulk density values for first 40 cm depth were used based on the HYPROP-FIT estimation presented in Krevh et al. [59]. In

Table 3. Correlation matrices between precipitation fraction in soil water and soil physicochemical parameters.

Parameter	Precipitation Fraction (%)	Corg (g/kg)	Sand (%)	Silt (%)	Clay (%)	Bulk Density (g/cm3)
Precipitation fraction (%)	1.00	-	-	-	-	-
Corg (g/kg)	0.47	1.00	-	-	-	-
Sand (%)	0.46	−0.05	1.00	-	-	-
Silt (%)	0.48	0.73	0.29	1.00	-	-
Clay (%)	−0.57	−0.59	−0.60	−0.94	1.00	-
Bulk density (g/cm3)	−0.55	−0.64	−0.20	−0.35	0.37	1.00

, correlation matrix is presented, while statistically significant results are marked in red (α = 0.05). Although not statistically significant, it can be clearly seen that precipitation infiltration is positively correlated with C_org, sand and silt, and negatively correlated with clay content and bulk density. This also implies that precipitation infiltration depends on the joint influence of the different soil parameters, which quantification should be one of the focuses of the future research. Statistically significant negative correlation was observed between silt and clay content, while statistically positive significant correlation was related to the connection between C_org and silt content. Results also showed that C_org in this case is not positively correlated with the clay content. Although maybe not that common, research conducted in arid and semi-arid regions has shown that fine silt can contribute more organic carbon then clay when its proportion is relatively higher than clay [81], which can also be the case in this research.

Furthermore, it must be emphasized that long-term monitoring has to be established in order to achieve more reliable results. This is especially related to the sampling of soil water in the summer months, which was not possible in all monitoring points within the investigated period. Furthermore, for a more detailed statistical inspection of all observed parameters, as well as to define their relationship under different hydro-meteorological conditions, additional data is necessary [82,83,84,85,86,87,88]. In addition to long-term data collection, future research should be focused to the development of water flow and transport models, evaluation of seasonal patterns of different hydrometeorological variables, and definition of the isotopic difference between mobile, non-mobile, and bulk water.

4. Conclusions

This paper presents research related to the estimation of precipitation fractions in the soil water at the SUPREHILL CZO, located in the City of Zagreb, Croatia. It is a sloped vineyard where the monitoring of soil water isotopic composition was conducted in 24 points at 4 depths. It has been shown that soil water isotopic composition from all monitoring points coincides with the calculated LMWL, with almost no variability at 100 cm depth, which is consistent with the smallest variation of SWC. This suggests the existence of non-mobile soil water, on which the precipitation influence can not be seen in the observed period, and that most of the water mixing occurs in first 100 cm depth. Results also suggest the existence of heterogeneity, uneven erosion processes in the footslope, and different infiltration patterns. Fractions of precipitation in soil water varied depending on the depth and position in the vineyard, from approximately 1% up to 98%, where more precipitation fraction can be seen in the surface and subsurface runoff. Statistical analysis and detailed evaluation of the physicochemical parameters and isotopic composition of soil water captured by wick lysimeters has shown that C_org content is, in general, related to the silt fraction, while the first results indicate that infiltration patterns are dependent on the common influence of all observed physicochemical properties. Furthermore, it has been shown that captured water volumes varied significantly with respect to the position in the vineyard, sampling depth, and used equipment, while sampling of soil water in summer months was challenging due to dry conditions. However, monitoring must be continued in order to achieve more reliable results, while future research should be focused on the development of a soil water flow model calibrated using stable isotopes of water, an evaluation of the differences in the isotopic signature of mobile, non-mobile, and bulk water, as well as on the examination of the relationship between the observed parameters in different hydrometeorological conditions.