Deep soil water recharge response to precipitation in Mu Us Sandy Land of China

Yi-en Cheng*,Hong-in Zhan,Wen-in Yang,Fang Bao

aSchool of Soil and Water Conservation,Beijing Forestry University,Beijing 100083,China

bInstitute of Desertification Studies,Chinese Academy of Forestry,Beijing 100093,China

cDepartment of Geology&Geophysics,Texas A&M University,College Station,TX 77843-3115,USA

Abstract Soil water is the main form of water in desert areas,and its primary source is precipitation,which has a vital impact on the changes in soil moisture and plays an important role in deep soil water recharge(DSWR)in sandy areas.This study investigated the soil water response of mobile sand dunes to precipitation in a semi-arid sandy area of China.Precipitation and soil moisture sensors were used to simultaneously monitor the precipitation and the soil water content(SWC)dynamics of the upper 200-cm soil layer of mobile sand dunes located at the northeastern edge of the Mu Us Sandy Land of China in 2013.The data were used to analyze the characteristics of SWC,in filtration,and eventually DSWR.The results show that the accumulated precipitation(494 mm)from April 1 to November 1 of 2013 significantly in fluenced SWC at soil depths of 0-200 cm.When SWC in the upper 200-cm soil layer was relatively low(6.49%),the wetting front associated with 53.8 mm of accumulated precipitation could reach the 200-cm deep soil layer.When the SWC of the upper 200-cm soil layer was relatively high(10.22%),the wetting front associated with the 24.2 mm of accumulated precipitation could reach the upper 200-cm deep soil layer.Of the accumulated 494-mm precipitation in 2013,103.2 mm of precipitation eventually became DSWR,accounting for 20.9%of the precipitation of that year.The annual soil moisture increase was 54.26 mm in 2013.Accurate calculation of DSWR will have important theoretical and practical significance for desert water resources assessment and ecological construction.

Keywords:Mu Us Sandy Land;Sandy land;DSWR;Precipitation;Wetting front

1.Introduction

In semi-arid sandy lands,water is a scarce resource and a key factor restricting the vegetation growth(Rodríguez-Iturbe and Porporato,2005).In semi-arid areas with low amounts of precipitation,high evaporation,and low sandy soil water content(SWC),soil moisture becomes a primary factor in fluencing vegetation growth and composition(Laio et al.,2001),and it determines the occurrence or reversal of land desertification(Imeson,2012).As precipitation is the main source of recharge to soil moisture in semi-arid sandy areas(Pye and Tsoar,1987),understanding the relations between precipitation and soil moisture in those areas is significant to the efficient use of limited water resources,as well as land desertification prevention and control.

Scientists have conducted a large amount of research on the relations between precipitation and soil moisture in sandy land ecological systems(Qiu et al.,2001;Wang et al.,2006).The case study of the Horqin Sandy Land of China shows that soil moisture of mobile dunes at depths of 40-300 cm is closely associated with the amount of precipitation that occurred in previous months(Katoh et al.,1998;Li et al.,2009).A precipitation amount of about 4 mm can affect soil moisture as deep as 20 cm(Liu et al.,2006),and subsequent moisture transport can also affect the change of soil moisture at depths of 20-60 cm.However,even in a wet year with a precipitation amount of 450 mm,the potential recharge of the precipitation to groundwater is extremely weak.Laboratory experimental results from sandboxes(Liu et al.,2006)showed that soil moisture at depths of 20-140 cm was highly variable under the in fluence of precipitation.The soil moisture distribution at depths greater than 140 cm was relatively stable.Specifically,when the depth was greater than 200 cm,evaporation completely ceased,and the precipitation-induced in filtration eventually recharged the groundwater system.A precipitation amount of 13.4 mm in one event was the threshold for generating measurable deep soil water recharge(DSWR)below 140 cm.The field experiment conducted at the Mu Us Sandy Land showed that an event with a precipitation amount of larger than 15 mm generates noticeable DSWR(Yang et al.,2014).This 15-mm precipitation threshold in the field was only slightly greater than the 13.4-mm precipitation threshold in the laboratory study for DSWR generation.There have already been many studies on the in fluence of precipitation on soil moisture in sandy areas(Li et al.,2004).For instance,a study on mobile dunes showed that soil moisture at a depth of 100 cm or a depth of 140 cm was not affected by evaporation(Zhang et al.,2008).Some studies have suggested that the precipitation threshold in a single event that can penetrate a 200-cm soil layer(thus making direct contribution to DSWR)is 15 mm(Ayalon et al.,1998).Some other studies have indicated that the precipitation threshold in a single event that can penetrate a 30-cm soil layer(but cannot penetrate a 200-cm depth)is 8 mm or 13.4 mm(Dekker and Ritsema,1994).These findings are debatable(Pan and Mahrt,1987),and they did not address the issue of how the SWC in the preprecipitation period can affect the in filtration process in the soil(Jin et al.,2009).

However,there is a lack of in situ observations on deeper soil moisture dynamics and recharge effects in arid and semiarid regions.The existing studies often report different and sometimes contradictory conclusions on the wetting front of mobile dunes,with the wetting front defined as the lowest position at which SWC is greater than the field capacity(Cassel and Nielsen,1986).It is also unclear how much precipitation can recharge deeper soil water.

The objective of this study was to use in situ experiments to investigate deep in filtration processes or DSWR in the Mu Us Sandy Land of China.This study tried to answer the following questions:

(1)What is the soil moisture distribution in different soil layers under different precipitation events in sandy land?

(2)With different SWCs during the pre-precipitation period,how long will it take for precipitation to in filtrate into a soil layer at a depth of 200 cm?

(3)How can precipitation-induced recharge to the groundwater system be evaluated and what percentage of precipitation can eventually become DSWR?

2.Materials and methods

2.1.Overview of study area

The study area is located on the northeastern edge of the Mu Us Sandy Land of China.This area is part of Chagan Nur Gacha(39°5′N,109°36′E),Ejin Horo Banner,in the Inner Mongolia Autonomous Region of China.The location of Ejin Horo Banner is 38°56′N to 39°49′N,and 108°58′E to 110°25′E.It is in the southeastern part of Ordos City,with an elevation between 1070 m and 1556 m above mean sea level(MSL),and a water table usually 5 m below the sandy land surface(Zhang et al.,2012).The area has a typically temperate continental monsoon climate,with a wind speed of 3.6 m/s,2900 h of sunlight(Xu et al.,2015),a relative humidity of52%,precipitation of 358.2 mm,and potential evaporation of 2563 mm,which is about 7.2 times the precipitation amount,all measured in annual averages.The annual average temperature is 6.2°C,and the frost-free period lasts for 127-136 d.Soil types in this area include dark loess soil,loess soil,chestnut soil,and aeolian soil.

The Ordos Basin is a large groundwater basin,which can be divided into an open shallow groundwater circulation system and a closed deep groundwater system.The shallow groundwater circulation system includes a strong groundwater exchange belt,a half-open runoff belt,and a weak runoff belt.The Cenozoic loose rock pore aquifer system and the upper portion of the Carboniferous-Jurassic clastic rock fissure aquifer system have a circulation depth of about 300 m.The Cambrian-Ordovician carbonate karst aquifer system has a maximum circulation depth of 1800 m.The Cretaceous clastic rock fracture aquifer system can reach up to 1200 m.The closed deep groundwater system mainly refers to the stagnation of groundwater,including the Cambrian-Ordovician carbonate rock karst aquifer system and the lower portion of the Carboniferous-Jurassic clastic rock aquifersystem.The boundary between the shallow groundwater circulation system and the deep groundwater system in the Ordos Basin is basically the boundary between fresh water(including brackish water)and salt water.Precipitation is the main recharge source of groundwater in the Ordos Basin.There is a hydraulic relationship between the shallow and deep aquifer systems.Interaction of surface water and shallow groundwater occurs in various locations.In the Ordos Basin,precipitation and condensation water are the only two sources of recharge to deep soil water in this region.Groundwater can only discharge through evaporation and/or base flow to surface water.Within the Ordos Basin,groundwater generally flows from the inner basin area to the boundary of the basin,where the Yellow River and its tributaries serve as the discharge channel of the groundwater.

2.2.Experimental design

On September 1,2012,a mobile sand dune within the study area was chosen as the monitoring plot,located at 39°5′16′′N,109°36′19′′E,with an elevation of 1310 m above MSL.The upper 220-cm soil pro file was excavated,and then the monitoring plot went through a settling process of one year to simulate the natural environment.Multiple soil samples were collected from different depths within the soil pro file using cylinders with a diameter of 3.5 cm and a length of 3.5 cm and were used to evaluate the average bulk density,which was found to be 1.64 g/cm3.The particle sizes of the samples were as follows:3.23% of the samples had particle sizes of 0.5-1.0 mm,86.59%of the samples had particle sizes of 0.1-0.5 mm,and 10.18%of the samples had particle sizes below 0.1 mm.The thickness of the studied sand soil was nearly homogeneous without any clear physical layering.The term layer is used here mostly for the convenience of describing the in filtration period and water content at different depths.With six water content sensors installed at different depths,we arbitrarily collected soil samples at different depths for analysis of the soil bulk density and soil particle size distributions.As the soil was nearly homogeneous,the results obtained with our sampling method were representative of the averaged data of the entire thickness of concern.In addition,we used cutting rings to collect soil samples at individual layers to detect field capacity.Soil moisture sensors(ECH2O-5,Decagon Devices Inc.,United States;precision:±3%)were inserted into the soil pro file at depths of 10,30,60,90,150,and 200 cm,and then the pro file was back filled with excavated soil.In filtration passing through the 200-cm depth was generally regarded as DSWR in this study and DSWR data were collected with a new lysimeter that was explained in detail in Cheng et al.(2017).This new lysimeter can measure DSWR without disturbing the soil structure.It does not have the sealed side walls of the traditional lysimeter,and thus will not cut off the lateral soil moisture migration.The measurement accuracy of the new lysimeter is 0.2 mm and the measurement interval is half an hour.

It is worthwhile to point out that some other investigators may use different depth thresholds for defining DSWR.For instance,Zhang et al.(2008)used a depth of 140 cm instead of 200 cm as the threshold to define DSWR.A precipitation sensor(AV-3665R,AVALON,UnitedStates;precision:0.2 mm)was placed above ground at the site.A soil water content data collector(CR200X,Campbell,USA)was used to record precipitation and SWC data.The SWC data were recorded every hour,and the precipitation data were recorded every half hour.The new lysimeter described above was used to collect DSWR.

2.3.Data analysis

The evaporation amount was calculated with the following water balance equation:

where P is precipitation,E is evaporation(there is no transpiration at the study site because of lack of vegetation),R is the surface runoff,ΔS is the soil moisture storage variation,and D is the DSWR.Besides precipitation,there is no other recharge source in this experimental area.The experimental area consists of bare sandy land,with rapid in filtration and no surface runoff,and thus R is zero.ΔS is obtained by checking the soil moisture difference from April 1(the start of the experiment)to November 1(the end of the experiment).It is notable that if P,R,ΔS,and D can be measured directly,then one can calculate E from Eq.(1),without knowing the actual meteorological condition at the study site,which together with rainfall intensity and duration will affect E.

The soil water storage capacity(SWSC)of the ith layer,denoted as Wi,is calculated as

where diand Ciare the thickness and soil water content of the ith layer,respectively.The total soil water storage capacity over N layers,denoted as Wtotal,is simply the summation of SWSC of each layer:

In order to avoid the effect of freeze-and-thaw action,the experiment was conducted between April 1 and November 1,2013.For the rest of the year(from January 1 to March 31 and November 2 to December 31),the upper soil of the study area was usually frozen,so the precipitation and evaporation were essentially zero(Zhang and Wang,2001).

3.Results and discussion

3.1.Precipitation and DSWR characteristics during experiment

The annual accumulated DSWR reaches 103.2 mm,which is 20.9%of the total annual precipitation(494 mm).The soil water storage was 144.16 mm on April 1(the start of the experiment)and 198.42 mm on November 1(the end of the experiment),resulting in a soil moisture increase of 54.26 mm.Therefore,based on the water balance equation(Eq.(1)),E was calculated to be 336.54 mm.This evaporation amount is far below the pan evaporation amount,which is approximately 2200 mm.

There were 75 precipitation events within the study area from April 1 to November 1,2013,with an overall precipitation amount of 494 mm,as shown in Fig.1.Of the 75 events,54 events had precipitation of less than 5 mm during a 24-h measurement period,with a total amount of 76.6 mm,accounting for 15.5%of the total precipitation.Nine of the 75 events had precipitation of between 5 mm and 10 mm during a 24-h measurement period,with a total amount of 70.8 mm,accounting for 14.31%of the total precipitation.Fourteen of the 75 events had precipitation of above 10 mm(one of them had precipitation of above 30 mm)during the 24-h measurement period,with a total amount of 252.0 mm,accounting for 51.0%of the total precipitation.Under drought conditions,runoff did not occur in this area even when the precipitation intensity reached 12.0 mm/h.During the experimental period,the largest daily precipitation was 38.0 mm during a 24-h measurement period(occurring on September 17),and such a precipitation event lasted for several days and brought in a total precipitation amount of 88.6 mm.The peak precipitation intensity for the event was 11.2 mm/h,which was greater than the heavy rain threshold of 10 mm/h used by some investigators(Goswami et al.,2006).

Fig.1.Relationship between precipitation and in filtration.

The results mentioned above show that the precipitation events that occurred during the experimental period were not strong enough to generate surface runoff,which was therefore neglected in the analysis described below.Therefore,the redistribution of precipitation includes three parts:evaporation,soil water storage,and DSWR.

3.2.Dynamic response of soil moisture to precipitation

The antecedent SWC affects in filtration.The maximum and minimum SWCs at different soil depths are shown in Table 1.Variance analysis shows that there are significant differences between the minimum SWC and the field capacity for each depth.For all the 75 precipitation events within the study area from April 1 to November 1,2013,the minimum SWC was often seen in May and June,and the maximum SWC was found in September.

Fig.2 shows the dynamic changes of SWC with time at different depths over the experimental period.As can be seenin Fig.2,for the measurements at the depth of 10 cm,SWC is highly variable and closely related to the precipitation events.For such a shallow depth,the occurrence time of the peak value of SWC corresponds closely to the peak time of precipitation,as the time needed for in filtration over the upper 10-cm soil layer is very short.

Table 1 Minimum and maximum SWCs at different soil depths.

From April 1 to June 17,there were 14 precipitation events with a total precipitation amount of 31.0 mm.The heaviestprecipitation in a single precipitation event occurred on May 28 with an amount of 8.4 mm,as shown in Fig.2.However,one can only observe eight responses(when precipitation reaches a certain soil layer,the SWC sensor will respond with a rise,as shown in Fig.2)from the SWC at the 10-cm depth.This implies that six of the precipitation events were not intense enough to reach even the 10-cm soil depth.However,during this same time period,SWCs at depths of 30,60,and 90 cm all continued to decline with time,showing no responses to the 14 precipitation events.This suggests that precipitation with an amount of below 8.4 mm is unlikely to penetrate a soil layer with a depth greater than 30 cm.

From June 18 to November 1,there were 61 precipitation events with a total precipitation amount of 368.4 mm.It can be seen that the SWC over the entire 200-cm soil layer responded to the 61 precipitation events to various degrees:one can observe seven responses at the 60-cm depth, five responses at the 90-cm depth,four responses at the 150-cm depth,and two responses at the 200-cm depth.The response time of SWC at the upper 150 cm was usually less than 7 d after precipitation events,with one exception on July 3,after which a delay of 13 d occurred.Precipitation events and their responses at depths greater than 90 cm indicate that those responses are related more closely to the multiple-day precipitation events,such as the four days of precipitation starting from June 19 with a total amount of 53.8 mm,the three days of precipitation starting from July 14 with a total amount of 30.4 mm,the two days of precipitation starting from August 4 with a total amount of 28.6 mm,and the three days of precipitation starting from September 16 with a total amount of 88.6 mm.

Fig.2.Changes of soil water contents at different depths of mobile sand dunes in Mu Us Sandy Land.

3.3.Relationships between antecedent SWCs and wetting front arrival time

During the experimental period,the minimum antecedent SWCs at all depths occurred before June 18.After June 18,the SWCs at all depths reached their first maximum values at various times before July 20,because of two events with precipitation amounts of 17.8 mm and 24.2 mm on June 19 and July 14,respectively.Therefore,these two precipitation events were chosen to analyze the relationship between the antecedent SWCs and the wetting front.The antecedent SWCs at different depths for the precipitation events on June 19 and July 14 are listed in Table 2.

The precipitation event on June 19 increased SWCs at all depths.The first time that the SWC at the depth of 60 cm reached its peak value was on June 20,one day after the precipitation occurring on June 19.The first time that the SWC at the depth of 200 cm reached its peak value was on July 18,four days after the precipitation occurring on July 14(Fig.1).We can analyze the time series of the wetting fronts between June 19 and July 18 based on Table 3.In particular,Table 3 covers a period of 306 h starting from 10:00 a.m.on June 19,and it includes a major precipitation event that occurred on June 19.Table 4 covers a longer period of 671 h starting from 10:00 a.m.on June 19(the same starting time as in Table 3),and it includes a major precipitation event thatoccurred on July 14 in addition to the one that occurred on June 19.

Table 2 Antecedent SWCs at different depths for precipitation events on June 19 and July 14.

The accumulated precipitation amounts at different times are also shown in Table 3.The precipitation event that started on June 19(lasting for 71 h,with a total amount of 53.8 mm)could be used to analyze the wetting front that started with a relatively low antecedent SWC of 6.49%(Table 3).Based on the hourly data of SWC,we could calculate the average SWC over a 3-h time interval,which could be used to determine the relationships between precipitation,the wetting front,and the arrival time of the maximum SWC at different depths.

Table 3 shows that starting from a relatively low antecedent SWC(6.49%),the precipitation amounts required for the wetting front to reach soil layer depths of 10,30,60,90,150,and 200 cm are 4.6 mm(at 3 h),12.6 mm(at 9 h),28.8 mm(at 24 h),53.8 mm(at 71 h),55.0 mm(at 254 h),and 62.8 mm(at 306 h),respectively.When the antecedent SWC is relatively high(10.22%),the precipitation amounts needed for the wetting front to reach depths of 10,30,60,90,150,and 200 cm are 2.0 mm(at 2 h),4.4 mm(at 3 h),11.2 mm(at 6 h),17.4 mm(at 10 h),24.2 mm(at 27 h),and 24.2 mm(at 56 h),respectively.Comparing the movement time of the wetting front at different antecedent SWCs,one can see that the in filtration speed is fast at a high SWC.

From Table 3,one can see that 62.8 mm of accumulated precipitation and 306 h are required for the wetting front to reach the 200-cm depth of the soil layer for the antecedent SWC of 6.49%,and 24.2 mm of accumulated precipitation and 56 h are required for the wetting front to reach the 200-cm depth of the soil layer for the antecedent SWC of 10.22%.This means that the required precipitation amount and wetting front arrival time with the relatively low antecedent SWC of 6.49%are,respectively,2.59 and 5.82 times of those with the relatively high antecedent SWC of 10.22%,for the wetting front to reach the 200-cm depth.Based on this,one can conclude that the antecedent SWC is a key factor controlling in filtration at the study site.

Table 3 Precipitation amounts required for wetting front to reach different soil layer depths and corresponding times.

In Table 3,the accumulated precipitation amount is 53.8 mm at 71 h and 9-mm precipitation occurred between 71 h and 306 h.As suggested in section 3.2,any precipitation event below 8.4 mm is unlikely to penetrate a 30-cm depth.Therefore,we can conclude that the precipitation that occurred between 71 h and 306 h makes no contribution to deep in filtration or DSWR at a depth of 200 cm.In other words,DSWR occurring at the 200-cm depth should be attributed to the precipitation event prior to 71 h,in which the accumulated precipitation amount is 53.8 mm.

Table 4 shows that starting from a relatively low antecedent SWC of 6.49%,when the SWCs reach the field capacities for layers at depths of 10,30,60,90,150,and 200 cm,the accumulated precipitation amounts are 17.2 mm(at 18 h),32.6 mm(at 26 h),53.8 mm(at 71 h),98.6 mm(at 617 h),124.4 mm(at 649 h),and 132.0 mm(at 671 h),respectively.It is notable that the field capacities are 10.94%,12.01%,12.55%,19.13%,17.96%,and 16.91%for depths of 10,30,60,90,150,and 200 cm,respectively.Different field capacities at different depths are a re flection of soil heterogeneity.This is consistent with previous studies which indicated that the field capacity in sandy land can be as high as 23%(Basso et al.,2013).

As shown in Table 4,with a relatively high antecedent SWC of 10.22%,when the SWCs reach the field capacities for layers at depths of 10,30,60,90,150,and 200 cm,the accumulated precipitation amounts are 4.6 mm(at 3 h),11.2 mm(at 6 h),21.8 mm(at 11 h),24.0 mm(at 12 h),24.2 mm(at 39 h),and 30.4 mm(at 74 h),respectively.At such a relatively low antecedent SWC of 6.49%,it takes 9.12 times longer for precipitation-induced in filtration to penetrate to the 200-cm depth than it does for a case with a relatively high antecedent SWC of 10.22%.From this,we can see that the antecedent SWC also controls the in filtration rate considerably.Table 4 shows that a single precipitation event with an amount of 30.4 mm is capable of allowing the soil layer at a 200-cm depth to reach its field capacity.

3.4.Precipitation-induced sandy soil moisture replenishment

At the study site,the shallow soil layer usually began to freeze on November 6,and thus soil evaporation is expected to be zero(Pitman et al.,1999;Ek et al.,2003).Therefore,we chose November 1 as the date to calculate the end of soil water storage capacity.The soil water storage capacities for different soil layers were calculated and are shown in Table 5.From this table,we can see that the soil water storage on April 1,2013 was 144.16 mm,which is close to the soil water storage of 144.17 mm from April 1,2014.This implies that the overall soil water storage capacity probably remains almost consistent annually.

However,the soil water storage capacity in each layer(especially at depths of 100-150 cm)may experience relatively large variationsannually.Thisshowsthat,although the shallower soil layers may freeze from aroundNovember 1 to April 1 of the following year at the study site,the deeper soil layer below a depth of 100 cm may still experience dynamic change of soil moisture with time.The total water storage capacity of 0-200 cm on November 1 was 54.26 mm higher than that on April 1.This means that,during the period from April 1 to November 1 of 2013,11.0%of the precipitation could penetrate a depth of 200 cm and become DSWR.

Table 4 Precipitation amounts required for different soil layer depths to reach field capacity and corresponding times.

Table 5 Soil water storage capacities on April 1 and November 1,2013,and on April 1,2014 for different soil layers.

In semiarid regions,precipitation varies considerably every year.It is true that the year 2013 may not be representative of the long-term average behavior of DSWR in this region as the precipitation during this year was higher than the average annual precipitation of 358.2 mm.Instead,it may be more representative of a wet year.To determine the long-term behavior of DSWR in the semiarid regions such as the Mu Us Sandy Land,one must carry out a multi-year(preferably decade-long)experiment.

4.Conclusions

This study analyzed the response of SWC in mobile dunes to precipitation under different antecedent SWC conditions in the Mu Us Sandy Land of China,and determined the layers that are affected by precipitation.The precipitation-induced in filtration passing through a 200-cm soil depth is regarded as DSWR.The following conclusions can be drawn from this study:

(1)The antecedent SWC is a key factor in fluencing precipitation-induced in filtration and arrival time of wetting fronts at different soil depths,and eventually controlling DSWR.

(2)According to the trajectory of the wetting front,when the antecedent SWC is relatively low(6.49%),precipitation takes 306 h to penetrate to the soil layer with a depth of 200 cm;when the antecedent SWC is relatively high(10.22%),precipitation takes 56 h to penetrate to the soil layer with a depth of 200 cm.

(3)The 494-mm accumulated precipitation significantly in fluences the changes in the SWC at depths of 0-200 cm.When the antecedent SWC is relatively low(6.49%),53.8 mm of accumulated precipitation can generate a wetting front that can reach the 200-cm depth,and 132.0 mm of accumulated precipitation can increase the SWC at depths of 0-200 cm to its field capacity.When the antecedent SWC is relatively high(10.22%),a single event with 24.2 mm of precipitation can drive the wetting front beyond the depth of 200 cm,and 30.4 mm of accumulated precipitation can increase SWC at depths of 0-200 cm to its field capacity.

(4)The field measurement results show that the annual evaporation amount over this experimental area is 336.57 mm,which is substantially less than the pan evaporation of approximately 2200 mm at the site.

The results of this study will provide a basis for accurate assessment of soil water storage capacity and groundwater recharge amount,as well as references for analysis of soil moisture response to precipitation in sandy areas.However,post-precipitation SWC redistribution and its recharge to deep soil water and groundwater is a complicated process.This calls for long-term monitoring,as well as a systematic analysis of different precipitation patterns and meteorological factors that have an in fluence on soil moisture,a subject that is not included in this study but will be investigated in the future.

Acknowledgements

WewouldliketothankNanDangandWeiFengforcollecting data and creating graphics.We also sincerely thank anonymous reviewers whose constructive comments provided invaluable assistance to us in revising and improving the manuscript.