Understanding soil water dynamics and accurately estimating groundwater recharge are essential steps in achieving efficient and sustainable management of groundwater resources in regions with deep vadose zones. The objective of this study was to understand transient data and the dynamics nature of water from deep sections at the thick vadose zone, and to estimate groundwater recharge by applying Darcy's law of unsaturated water fluxes. The study was conducted during year 2009–2013 at Luancheng Agro-ecosystem Experimental Station of Chinese Academy of Sciences, which is located in the North China Plain. The water contents were measured with water probes and matric suctions using pressure transducers at depths of 9 and 11 m and were combined with laboratory measurements of unsaturated hydraulic conductivity to estimate groundwater recharge. The results indicated that the soil water content at 9- and 11-m depths increased following the rainy season and then gradually stabilized. And the intensity and continuity of precipitation events played an important role in soil water changes. The soil water dynamics between different depths (9 and 11 m) indicated a time lag (approximately 5–11 days). The groundwater recharge ranged from 7.60 to 19.75 mm resulting from hysteresis over the study period.
Generally, groundwater is the fundamental water source needed to meet the rapidly increasing industrial, agricultural and economic development requirements, particularly in arid and semiarid zones. However, excessive groundwater use would result in a rapid decline in groundwater levels and in aggravated environmental problems (Green et al. 2008; Lu et al. 2011). Groundwater recharge from precipitation and irrigation, which refers to water that arrives at the groundwater surface, is an important component of water cycles driving sustainable groundwater management (Walker et al. 2002). Furthermore, groundwater recharge is influenced by many factors: land use, crops, evaporation and transpiration, the thickness of the unsaturated zone, and the properties of the vadose zone (Kendy et al. 2004; Wang et al. 2008; von Rohden et al. 2010; Huang et al. 2013; Turkeltaub et al. 2014). Therefore, groundwater-recharge estimation is one of the most challenging issues in water resources research, especially in a region with a thick vadose zone.
Many approaches have been used to estimate groundwater recharge and deep drainage, which have included conventional water-balance methods (Finch 2001; Favreau et al. 2002; Kendy et al. 2003; Hubbell et al. 2004; Scanlon et al. 2005), modeling techniques (Kendy et al. 2004), groundwater-table measurements (Scanlon et al. 2007) and environmental tracers such as tritium and chloride (von Rohden et al. 2010; Huang et al. 2013). Scanlon et al. (2002) presented a comprehensive research review of the methods using vadose zone data for groundwater-recharge estimations. Conventional water-balance methods have involved large errors due to many aspects of recharge estimation for (semi)arid areas (Finch 2001; Favreau et al. 2002). Furthermore, numerical unsaturated-zone simulations, which are a useful tool for estimating the magnitude and timing of recharge, require calibration from a dataset of shallow depths or verification of the parameters (Kendy et al. 2004). Water-level measurements, the most direct measure of groundwater recharge, are not an effective approach when the groundwater level is affected by pumping (Scanlon et al. 2007). This is particularly true when the unsaturated zone is very thick, when precipitation or irrigation water may take a long time to reach the water table. Environmental tracers have always been applied to estimate groundwater recharge over recent years (Si & de Jong 2007; Wang et al. 2008; von Rohden et al. 2010; Huang et al. 2013). These studies have usually estimated groundwater recharge using a soil profile measurement, which has been represented by the long-term cumulative state of the vadose zone, and which has estimated the recharge process on time scales of years to decades. Moreover, environmental tracer methods do not often provide information about pore-scale preferential flow (Rimon et al. 2011).
Despite numerous studies, the estimation of groundwater recharge has remained uncertain and inconsistent in many regions, particularly in regions with a thick vadose zone. For example, in the North China Plain (NCP), which has experienced a steeply declining water table and increased thickness of the unsaturated zone, Kendy et al. (2004) found that the recharge rates ranged from 5 to 109 cm/a based on the quantity of precipitation and irrigation; Wang et al. (2008) determined that the recharge rate ranged from 0.14 to 0.67 mm/d using tritium and bromide tracing from 2003 to 2005; however, von Rohden et al. (2010) obtained an effective recharge rate of approximately 0.3 m/a using environmental tracers (3H-3He, noble gases, stable isotopes 18O and 2H). And, most previous studies have focused primarily on estimating groundwater recharge from measuring the root zone. Few researchers monitored the dynamics of soil water in the deep vadose zone or estimated groundwater recharge from precipitation and irrigation in the deeper unsaturated zone (Huang et al. 2013; Turkeltaub et al. 2014). At sites with thick unsaturated zones, can precipitation and irrigation recharge groundwater? And, if so, what is the amount of groundwater recharge? Estimating groundwater recharge at an increased depth may be more representative than the past estimations from shallow depths. Additionally, the inconsistent understanding of groundwater-recharge estimation in the near-surface environment could be eliminated with this approach.
Therefore, the objective of this study was to understand transient data and the dynamics nature of water from deep sections at the thick vadose zone, and to estimate groundwater recharge by applying Darcy's law of unsaturated water fluxes in the deep vadose zones. This would be helpful to sufficiently understand the dynamic nature of percolation and transport processes and to manage groundwater resources more efficiently and sustainably in a region with a thick unsaturated zone.
MATERIALS AND METHODS
|.||Groundwater recharge (mm)|
|Soil depth .||2009.8–2009.12 .||2010 .||2011 .||2012 .||2013 .|
|.||Groundwater recharge (mm)|
|Soil depth .||2009.8–2009.12 .||2010 .||2011 .||2012 .||2013 .|
Instrument setup and data collection
The soil water content and matric suction at deep soils were measured continuously in an access well using time domain reflectometry (TDR) probes and with pressure transducers, respectively. The 15-cm long TDR probes (CS630, Campbell Scientific Co., USA) were horizontally inserted 15 cm into the soil at 9 and 11 m depth to reduce the effects of water flow disturbance from the wall of the well. And the pressure transducers (Watermark probe, Campbell Scientific Co., USA) were used to determine the soil matric suction at 8.75 m, 9.25 m, 10.75 m and 11.25 m depth, respectively; they were also used to further calculate the corresponding water potential gradient at 9 and 11 m depth. First, to install the transducers, a hole perpendicular to the well wall was drilled to approximately the diameter of the probe; second, a pressure transducer was inserted into the hole horizontally and the soil mud was filled around the transducer into the hole to ensure contact between the transducer and the soil. Furthermore, T-type thermocouples were inserted at the same soil depth to calculate the matric suction. The soil water content and matric suction were automatically recorded using a TDR100 system (Campbell Scientific Co., USA) every 6 h from August 2009 to December 2013.
Soil hydraulic conductivity
Unsaturated hydraulic conductivity is an important parameter of hydraulic properties for calculating groundwater recharge. In this study, the unsaturated hydraulic conductivity function was determined using an evaporation measurement device (HYPROP, HYdraulic PROPerty analyzer, Germany) in a laboratory (Schindler et al. 2010). The undisturbed soil samples at 9 and 11 m depths were obtained with stainless steel cylinders (0.08 m diameter, 0.05 m height). The soil column was slowly saturated by placing it in a pan of distilled water to minimize trapped air. To guarantee complete saturation, the pressure head was adjusted several times during the process.
After saturating the soil column, tensiometers were installed into the soil core in the upward direction. Then, the tensiometer assembly was placed on a scale, and the whole assembly and scale were connected to a computer through a signal cable for data recording. When the soil column started to evaporate, the water tension values and the sample mass were collected simultaneously at 10-min intervals. The water tension values, water content, and water fluxes were used to derive the water retention curve and the unsaturated hydraulic conductivity function. This experiment was conducted in a laboratory at a temperature of 18–25°C and under atmospheric pressure. A detailed description of this method is given by Schindler et al. (2010).
RESULTS AND DISCUSSION
Temporal dynamics of soil water content
The temporal dynamics of the soil water content exhibited the same behavior at both depths from August 2009 to December 2013 (Figure 3). The soil water content changed slowly over time, ranging from 0.03 to 0.06 cm3/cm3 at the 9 m depth and from 0.27 to 0.31 cm3/cm3 at the 11 m depth. Note that some of the apparent variability in Figure 3 may be attributed to measurement artifacts. This corresponds with the findings of other studies (Turkeltaub et al. 2014; Min et al. 2015). For example, Min et al. (2015) indicated that soil water content changed by no more than 0.05 cm3/cm3 slightly below the root zone. Meanwhile, during the study period, soil water content at the deep unsaturated zones can have a response to the precipitation, and displayed temporal dynamics corresponding to the precipitation. The soil water content increased gradually after the rainy season (from June to September), followed by a slow decline in the water content to a stable level in the drought season of each year. This corresponds with the results of Turkeltaub et al. (2014), which found that soil water content showed slow and gradual increases at a >2 m depth after rainfall (Turkeltaub et al. 2014). The wetting front of water propagated in a step-like pattern (Zimmermann et al. 1966; Turkeltaub et al. 2014). The increase in the water content in the deep soil resulted from infiltrated rainwater from the upper soil layers: rainwater gradually moved downward into the deeper soil horizon, and the soil water content at depth increased and then became constant when the wetting front arrived.
Although the temporal dynamics of the soil water content were relatively similar in each year, the magnitude of the increasing water content was different. Variations in soil water clearly depended on rain events. Especially in 2009 and 2012, the obvious increasing soil water content trend was most likely attributable to greater rainfall events and rainfall intensities. This is consistent with a study by Turkeltaub et al. (2014), which showed that the wetting-front process of the deep vadose zone was 61–94 days later than the upper layer after rain. Also, continuous rain happened in 2009 and 2012. As a result, the soil water content at 9 and 11 m depths rose drastically by 0.04 cm3/cm3 (Figure 3). Multiple rainfall events merge into one recharge process when the vadose zone is thick enough (Huo et al. 2014). The consecutive rainfalls caused the infiltrated water to far exceed the soil water capacity, and the infiltrated water continued to move towards the deep soil. This is also supported by the findings of other studies (Huang et al. 2013; Huo et al. 2014; Turkeltaub et al. 2014). For example, Turkeltaub et al. (2014) indicated that only a significant amount of surplus water progressed into the deep soil layer.
Furthermore, the dynamics of the soil water content displayed hysteresis between 9 and 11 m of depth (Figure 3). The peak soil water at 9 and 11 m occurred on 12 October 2009 and 23 October 2009, respectively, indicating an 11-day delay and thus an average water movement velocity of approximately 0.13 m/day. For each year from 2010 to 2013, the lag time between the two different depths was 9 days, 8 days, 5 days, and 10 days, respectively. This is consistent with other studies in the NCP (Qiu 1992; Rimon et al. 2007), which found that the mean propagation velocity of the wetting front was approximately 0.15–0.2 m/day. Similarly, Min et al. (2015) found that the velocity of the wetting front was approximately 0.13 m/day below the root zone (2 m). With a velocity of downward water movement of 0.13 m/day, it would take 323 days for rainwater to reach the groundwater table at 42 m. Huang et al. (2013) reported that the groundwater took more than 60 years to be recharged in the Loess Plateau of China. Min et al. (2015) showed that the response time of the water table to rainfall input might be no more than 1 year. However, Lu et al. (2011) indicated that the response time would be approximately 1 month, based on numerical modeling. The disparity might result from the difference in soil layers and soil texture-related parameters in the numerical model. Water movement was primarily delayed between the soil layers. If the soil profile had a uniform texture and structure, the water infiltration time would be shorter. In addition, the time lag of recharge is prolonged with increase in thickness of the vadose zone (Huo et al. 2014).
Additionally, three to five irrigation applications were carried out each year during extreme drought events. Irrigated water was applied to the topsoil mainly to support crop growth. The soil water content displayed no obvious changes after irrigation. Therefore, irrigation had no effect on the soil water increases in the deep soil. Furthermore, during the experimental period, the water content at 11 m was approximately 0.25 cm3/cm3 higher than at 9 m depth.
Estimation of groundwater recharge
Table 1 lists the groundwater recharge at 9 and 11 m depths as estimated by Darcy's law over five consecutive years. During the whole measurement period, the vertical recharge from rainfall decreased with depth. The groundwater recharge was much greater at 9 m depth than at 11 m depth. The total groundwater recharge reached nearly 60 mm at 9 m depth and 6.26 mm at 11 m depth. At 11 m depth, the groundwater recharge was no more than 1 mm during each of the years between 2010 and 2013. Moreover, groundwater recharge differed greatly in different years. In 2009.8–2009.12, the groundwater recharge was the greatest, at 19.57 mm at 9 m and 5.42 mm at 11 m depth, respectively. In 2013, groundwater recharge was the lowest, at 7.60 mm at 9 m and 0.05 mm at 11 m depth, respectively. In this study, the groundwater recharge is far lower than the results of other studies (Kendy et al. 2004; von Rohden et al. 2010; Huo et al. 2014; Min et al. 2015). Kendy et al. (2004) reported that the recharge rates ranged from 50 to 1,090 mm/a based on the water balance method; Lu et al. (2011) showed that the average recharge rate was 180 mm/a according to the modeled results. The reason for the disparity might be the uncertainty in the parameters of the groundwater recharge estimation. Hubbell et al. (2004) indicated that the deep drainage was closely related to the hydraulic conductivity, with uncertainty of orders of magnitude. The change in the parameters (θr; θs; α; n) and hydraulic conductivity (Ks) would decrease or increase the groundwater recharge by 10% at most (Min et al. 2015).
Recharge relied on many factors, such as rainfall duration, intensity, antecedent water content, and evapotranspiration (Stephens & Robert 1986; Cao et al. 2012). The intensities and durations of the precipitation events were important for groundwater recharge, especially in the NCP. Fei (1988) attributed 70–80% of the recharge to precipitation and the rest mainly to surface water seepage. Other studies indicated that the areal recharge increases with water inputs increase (Kendy et al. 2004; Wang et al. 2008; Lu et al. 2011; Min et al. 2015). In this study, the precipitation and number of rainfall events were considerably greater in 2009 and 2012 compared with the average annual precipitation. Furthermore, the number of precipitation events with an intensity over 5 mm were also greater in 2009 and 2012 compared with other years. Twenty-five and 23 events with precipitation amounts over 5 mm, one rainstorm event (93.7 and 96.2 mm), and five and four precipitation events lasting nearly 3 days were observed in 2009 and 2012, respectively (Figure 3). The temporal change of the groundwater recharge was primarily caused by different water input (Kendy et al. 2004; Wang et al. 2008; Min et al. 2015). Thus, the rainfall events in 2009 and 2012 were abundant and provided favorable conditions for water recharge in the deeper soils.
Furthermore, the hysteresis of water movement also influenced the recharge. Precipitation in 2010 was lowest among the measurement periods. However, greater estimated water fluxes of 16.03 mm at 9 m in 2010 were observed. It seemed that part of the greater recharge in 2010 was likely due to the greater precipitation and intensity events in 2009 of 557.3 mm. And the low water flux in 2011 might be attributed to the accumulation of low precipitation and intensity in 2010 and 2011. Previous studies reported that the temporal change of the groundwater recharge was primarily caused by different water input and climate factors (Stephens & Robert 1986; Wang et al. 2008; Min et al. 2015). Clearly, the 2 m interlayer between the 9 and 11 m depths delayed water movement. This corresponds to the hysteresis of the water content change at 9 and 11 m depths, as mentioned above. This is consistent with other studies, which showed that the soil water flux is depth dependent (Hubbell et al. 2004; Turkeltaub et al. 2014; Min et al. 2015). Turkeltaub et al. (2014) found that soil water flux showed a significant variation with depth. The recharge from the water input is different at different depths in the deep vadose zone (Carrera-Hernandez et al. 2012). It indicated that hysteresis plays an important role in estimating the actual groundwater recharge in the NCP. Therefore, in this study, the percent of precipitation entering the groundwater could not be estimated accurately over periods of months or years.
Due to the time lag between the precipitation and groundwater recharge in thick unsaturated zones, the infiltrated water from precipitation would take a long time to reach the groundwater table. In areas with deep water tables, virtually all of the water far below the root zone is assumed to have escaped from evapotranspiration and could eventually recharge groundwater. This is also supported by the findings of other studies (Huo et al. 2014; Min et al. 2015). And due to soil water moving continually downward, the cumulative water flux increased gradually over time. Therefore, from a long-term perspective, it could actually be regarded as groundwater recharge from precipitation in the region.
In this study, soil water content and matric suction were determined using TDR probes and pressure transducers at depths of 9 and 11 m to monitor the long-term hydrologic activity in the NCP in a thick vadose zone. Darcy's law was used to estimate the vertical water flux and groundwater recharge. The results indicated that the precipitation significantly influenced the soil water content at 9 and 11 m depth, especially after a rainstorm. The soil water increased by approximately 0.03–0.04 cm3/cm3 after the rainy season. Furthermore, the soil water dynamics were influenced relative to consecutive precipitation events. Groundwater recharge ranged from 7.60 to 19.57 mm over different measurement periods due to different precipitation trends and buffering of the thick unsaturated zone.
This research was funded by National Natural Science Foundation of China (41271241) and the Innovation Knowledge Project of the Chinese Academy of Sciences (Grant Nos. KSCX2-EW-J-5; KZCX2-EW-415).