Journal of Hydrology 552 (2017) 241–248

Contents lists available at ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier .com/ locate / jhydrol

Research papers

Method to measure soil matrix infiltration in forest soil

http://dx.doi.org/10.1016/j.jhydrol.2017.06.0320022-1694/� 2017 Published by Elsevier B.V.

⇑ Corresponding author at: China Agricultural University, No. 17, Qinghua DongluRd, Haidian District, Beijing 100083, PR China.

E-mail address: leitingwu@cau.edu.cn (T. Lei).

Jing Zhang a, Tingwu Lei a,b,⇑, Liqin Qu c, Ping Chen a, Xiaofeng Gao a, Chao Chen a, Lili Yuan a,Manliang Zhang d, Guangxu Su d

aCollege of Water Resources & Civil Engineering, China Agricultural University, Beijing 100083, PR Chinab State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, CAS and MWR, Yangling, Shaanxi 712100, PR Chinac State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research, Beijing 100048, PR Chinad Tianshui Soil and Water Conservation Station, Yellow River Conservation Commission, Tianshui, Gansu Province 741000, PR China

a r t i c l e i n f o a b s t r a c t

Article history:Received 13 November 2016Received in revised form 15 April 2017Accepted 17 June 2017Available online 20 June 2017This manuscript was handled by K.Georgakakos, Editor-in-Chief, with theassistance of Venkat Lakshmi, AssociateEditor

Keywords:Forest soilMatrix infiltrationDouble-ring infiltrometerPreferential flow

Infiltration of water into forest soil commonly involves infiltration through the matrix body and prefer-ential passages. Determining the matrix infiltration process is important in partitioning water infiltratinginto the soil through the soil body and macropores to evaluate the effects of soil and water conservationpractices on hillslope hydrology and watershed sedimentation. A new method that employs a double-ring infiltrometer was applied in this study to determine the matrix infiltration process in forest soil.Field experiments were conducted in a forest field on the Loess Plateau at Tianshui Soil and WaterConservation Experimental Station. Nylon cloth was placed on the soil surface in the inner ring andbetween the inner and outer rings of infiltrometers. A thin layer of fine sands were placed onto the nyloncloth to shelter the macropores and ensure that water infiltrates the soil through the matrix only. BrilliantBlue tracers were applied to examine the exclusion of preferential flow occurrences in the measured soilbody. The infiltration process was measured, computed, and recorded through procedures similar tothose of conventional methods. Horizontal and vertical soil profiles were excavated to check the successof the experiment and ensure that preferential flow did not occur in the measured soil column and thatinfiltration was only through the soil matrix. The infiltration processes of the replicates of five plots wereroughly the same, thereby indicating the feasibility of the methodology to measure soil matrix infiltra-tion. The measured infiltration curves effectively explained the transient process of soil matrix infiltra-tion. Philip and Kostiakov models fitted the measured data well, and all the coefficients ofdetermination were greater than 0.9. The wetted soil bodies through excavations did not present evi-dence of preferential flow. Therefore, the proposed method can determine the infiltration processthrough the forest soil matrix. This method can also be applied to explore matrix infiltration in otherland-use types.

� 2017 Published by Elsevier B.V.

1. Introduction

Infiltration of water into forest soil involves soil matrix infiltra-tion and preferential flow (Stumpp and Maloszewski, 2010; Wanget al., 1994) because preferential flow commonly occurs in mostforest soils through macropores (Flury and Flühler, 1995;Stagnitti et al., 1995; Jarvis, 2007). Matrix infiltration in forest soilis related only to the process of water infiltration through the soilbody and is solely controlled by soil matrix capillary action. Deter-mining matrix infiltration in forest soil is highly important in par-titioning water infiltrating the soil matrix or going through

preferential passages. The water infiltrating through the soil matrixdetermines the amount of water stored in the soil matrix which isimportant to evaluate the effect of soil and water conservationpractices on hillslope hydrology and watershed sedimentation.The water infiltrating the soil body is transformed into soil wateravailable for plant water uptake and vegetation restoration. Vege-tation restoration can improve soil structure, control soil erosion,and conserve soil water resources. The preferential flow inducedby macropores is the main cause of pollution transport andgroundwater circulation and contamination (Doerr et al., 2007;Mooney and Morris, 2008).

The commonly used methods for soil infiltration measurementinclude double-ring infiltrometers, rainfall simulator and discinfiltrometer. A double-ring infiltrometer measures soil infiltrationunder the ponding water layer, which commonly involves

242 J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248

preferential flow. Jabro et al. (1994) used double-ring infiltrome-ters to estimate the preferential movement by bromide tracerunder field conditions, to find that water evidently moved throughmacropores to deep soil layers. In Sander and Gerke (2007) andKodešová et al.’s studies (2012), a rectangular frame was ham-mered into soil before the tracer solution was poured into the infil-tration plot under 5 cm ponding depth to evaluate the nature of thepreferential flow. The infiltration process measured by the rainfallsimulator also includes preferential flow. For example, Hardie et al.(2011) applied a Morin rotating-disk rainfall simulator to spray4 g L�1 Brilliant Blue solution to study the type, depth, and rateof dye tracer movement into the soil profile. Double-ring infiltrom-eters and rainfall simulators, when used in conventional ways, can-not directly measure forest soil matrix infiltration. A discinfiltrometer applies water to soil at negative water potentials pre-viously determined by the operator, which causes water to infil-trate through soil matrix. Given that water is driven bycapillarity force to enter the soil matrix body and the device worksunder negative water potential, this infiltrometer can only mea-sure unsaturated soil hydraulic parameters in situ (Logsdon andJaynes, 1993; Angulo-Jaramillo et al., 2000; Šimunek andGenuchten, 1996). To ensure good contact and level the soil sur-face, a sand layer is typically used between the disc and soil sur-face. The size and thickness of the sand used influence the waterflow through the sand layer and could cause randommeasurementerrors. Furthermore, a disc permeameter cannot measure the freematrix infiltration process on soil surfaces under positive pressureconditions. Even soil saturated hydraulic conductivity measure-ment under ponded infiltration conditions is not guaranteed bythe disc method (Reynolds and Zebchuk, 1996; Reynolds andElrick, 1991).

Li and Ghodrati (1997) constructed soil columns with macrop-ores of known shape and geometry in laboratory to visualizematrix infiltration and preferential flow processes. Köhne andMohanty (2005) also designed a laboratorial soil column experi-ment with artificial preferential flow paths. The laboratory experi-ment, including drainage, upward, and downward infiltrationmeasurements, conducted by applying a tension infiltrometer toquantify the preferential water flow processes. These studies reliedon advanced indoor measurement technologies and speciallydesigned devices, but did not directly relate to field conditions. Asoil column with macropores cannot easily imitate field conditionsand is not completely representative of field conditions. Underfield conditions, a lysimeter was used to partition water infiltrationthrough the soil matrix and through preferential passages based onmass balance (Greve et al., 2010; Schoen et al., 1999; Stumpp andMaloszewski, 2010). Everts and Kanwar (1990) used a lysimeterand measured tracer concentrations and the flow rate of drainagewater to estimate the preferential and matrix flow componentsof subsurface drainage. Stumpp and Maloszewski (2010) usedstable isotope contents, hydrological data, and a lysimeter to quan-tify preferential and matrix flows by measuring the seepage waterof the lysimeter for five years. However, the installation of lysime-ters is time consuming and can be expensive (Allaire et al., 2009). Alysimeter cannot quantify the amount of water infiltrating throughthe soil matrix before the soil matrix becomes saturated. Lysime-ters are also inapplicable for the rapid measurement of the tran-sient infiltration of soil matrix. No detailed study has reported onthe direct measurement method for the transient infiltration pro-cess of soil matrix in field conditions.

Reynolds and Zebchuk (1996) suggested that a layer of suitablecontact materials should be placed between an infiltrometer discand the soil surface to establish and maintain good hydraulic con-ductivity. Perroux and White (1988) reported that the hydraulicconductivity of a contact material should be no less than that ofthe measured soil. A layer of fine sand or sieved soil is widely used

as the contact material. For instance, in Kodešová et al.’s study(2010), 1 mm layer of the same materials as the tested soil sievedthrough a 2 mm mesh was lined between the soil surface and thedisk to ensure close contact of the disc with the surface soil.

The objectives of this study are to determine the soil matrixinfiltration process in forest soil under ponding conditions in thefield and examine the spatial variability of soil matrix infiltrationunder the same land use pattern.

2. Theory

Preferential flow occurrence is related to the size and connec-tivity of macropores to the soil surface (Wu et al., 2009). Forinstance, Allaire-Leung et al. (2000a) reported that the wettingfront movements were similar to the control (treatment withoutmacropores), C–C (close at the soil surface and close at the lowerboundary), and C–O (close at the soil surface and open at the lowerboundary) treatments and found that preferential flows were com-pletely sheltered when the top end of a macropore was closed,such as in the cases of C–O and C–C macropores. Wang et al.(2010) reported that the number of the macropores with radii lar-ger than those 0.3 mm, especially larger than 1.5 mm, was themost important one that caused the occurrence of preferentialflow. In Logsdon’s study (1995), when the diameter of soil porewas 6 mm, the water flow in the soil pore was controlled com-pletely by gravity, and preferential flow occurred. The results ofSun’s (2013) showed that macropores with diameter larger than0.7 mmwere the preferred means of preferential flow. The fractionof small capillary pores (pores of diameter <0.04 mm) representedthe soil matrix domain (Kodešová et al., 2012). If the macroporesare not connected to the soil surface to the open air or are shel-tered, preferential flow does not occur. The lower soil layer withmacropores could have the same characteristics as those of thematrix domain (Allaire et al., 2002). Therefore, the infiltration pro-cess measured by rainfall simulation or with a double-ring infil-trometer may only provide the soil matrix infiltration rate whenthe macropores are disconnected to the open air at soil surfaceor sheltered.

Therefore, in the current study, a 1–3 cm layer of fine sandswere spread on a nylon cloth on top of the measured soil surfaceto shelter the macropores on the soil surface to ensure that waterinfiltrates soil profile only through the matrix. The saturatedhydraulic conductivity of the used sand was 290.9 mm/h, as mea-sured with the constant head method (Klute and Dirksen, 1986).This value is greater than the saturated hydraulic conductivity ofthe measured soil.

A coarse sand layer with high saturated hydraulic conductivitycovers a less conductive finer-textured soil layer. The infiltrationrate is initially controlled by the highly conductive top sand layer,but when the wetting front reaches and penetrates into the finer-textured sublaying soil, the infiltration rate is expected to drop toapproximately be equal to that of the soil. Thus, the soil layer withlower conductivity than that of the sand layer controls the infiltra-tion process (Hillel, 1998). Therefore, this arrangement can mea-sure the soil infiltration through the matrix. Therefore, the 1–3 cm depth of fine desert sand did not confine the soil matrix infil-tration process in this study. Fine sand sheltered the connectivityof macropores on the soil surface and did not affect the measuredsoil matrix infiltration process (Fig. 1).

The area sheltered by sands on the top of a macropore may stillleak a certain amount of water flow (Fig. 1). The leaked water flowhas a very low rate, which may be affected by the surface tension ofthe soil underneath the sand shelter and comes into contact withthe soil of the wells of the macropore. This causes the water toreturn to the soil matrix to increase the soil matrix flow. In turn,

Fig. 1. Schematic of the theoretic concept to measure matrix infiltration.

J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248 243

the increase in matrix flow underneath the surface can reducewater infiltration into the matrix. To some extent, the matrix infil-tration is rebalanced. Therefore, infiltration as a whole is well rep-resented by the water flux through the sand layer and theinfiltration process measured by a double-ring infiltrometer withgood shelter of macropores should be capable of representing thesoil matrix infiltration capacity.

Fig. 2. Experimental plots to measure soil matrix infiltration in forest soils on theLoess Plateau.

3. Methods and materials

3.1. Study area

The study area is located in Qiaozi East watershed, Tianshui,Guansu Province of China (34�360–34�370N, 105�420–105�430 E)and situated in the southwestern hilly–gully region of the ChineseLoess Plateau. The area is influenced by dry and continental cli-mates, with annual mean precipitation of 469–628 mm, with highseasonal and annual fluctuations. The average annual temperatureis 10.7 �C, with a minimum daily mean temperature of �2.3 �C inJanuary and a maximum daily mean temperature of 22.6 �C in July.The highest and lowest recorded temperatures are 38.2 �C and�19.2 �C, respectively.

3.2. Measurement method

The experimental site is located in a 2500 m2 relatively levelforested area under 50-year-old locust trees with an average chestdiameter of 31 cm. Five adjacent experimental plots were selectedat intervals of about 2 m apart, as five replicates, as shown in Fig. 2.After a level experimental plot was selected, the litter layer wascleaned carefully, and the soil surface was revealed. ASTM indi-cated that cylinders with diameters of about 30 and 60 cm couldbe used for ring infiltriometers. Infiltrometers, with inner andouter ring diameters of about 30 and 60 cm, have been widelyapplied to measure the soil infiltration process (Clancy and Alba,2011; Ronayne et al., 2012). Hammering a double-ring infiltrome-ter into the soil could disturb the soil structure and produce leak-age passages, thus affecting the measured infiltration rate andcausing a measurement error in the initial infiltration rate rangingfrom 207% to 121% (Bagarello and Sgroi, 2004; Lei et al., 2013;Zhang et al., 2016). In Lassabatere et al.’s study (2006), the ringswere positioned on the soil surface and inserted 1 cm into the soilto measure the infiltration process. Therefore, in the current study,

the double-ring height of 8 cm was gently inserted about 1 cm intothe soil by using a rubber hammer while ensuring that the ringremained horizontal during insertion. The same soil materials wereused to the outside of the inner and outer rings to avoid leakage ofwater from inner and outer rings. A nylon cloth with a diameter of40 cmwas placed on the soil surface inside the inner ring and a cir-cular nylon cloth with an inner diameter of 20 cm and an outerdiameter of 80 cmwas placed on the soil surface between the innerand outer rings. Afterward, a layer of fine sands of 1–3 cm thickwas evenly fed onto the nylon cloth inside the inner and bufferrings. To minimize the risk of scouring the fine sand layers duringadding water into the rings, 5 � 5 cm paper boards were placed onthe sand surface. A mixed solution of water with Brilliant Blue wasput into the inner ring, and water without Brilliant Blue waspoured into the outer ring. Computed amount of water was care-fully poured onto the paper board to fill the water to a depth of4 cm as quickly as possible for the inner and outer rings at thesame time. The falling head method was used to measure the soilinfiltration process. Thereafter, the time durations for water levelto drop 0.5 cm in the inner ring were recorded with a stopwatchuntil the infiltration time did not change for three consecutivemeasurements and steady-state infiltration was assumed. Waterwas refilled into the ring whenever the water level dropped to1 cm depth. The infiltration rate was calculated with Eq. (1). The

Fig. 3. Schematic of the vertical and horizontal excavation profile (unit: cm).

Table 1Grain size distribution of the experimental sites.

Depth (cm) Sand (%) Silt (%) Clay (%) IWC (%) SD (g/cm3) TP (%)1–0.1 mm 0.05–0.005 mm <0.005 mm

Sand 99.78 0.22 0.00 0.12 1.41 46.790–20 2.21 54.00 43.79 12.49 1.49 43.7720–40 2.20 56.00 41.80 10.69 1.53 42.2640–60 4.20 56.00 39.80 10.30 1.53 42.2660–80 6.20 56.00 37.80 10.70 1.62 38.8780–100 8.23 58.00 33.77 11.89 1.59 40.00

Fig. 4. Experimental sands for sheltering macropores connecting to soil surface inthe inner and outer rings.

Fig. 5. Infiltration and cumulative infiltration curves

244 J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248

steady-state infiltration rate was calculated based on the last threeinfiltration rates. In the entire infiltration process, the water levelin the outer ring was kept equal to that in the inner ring.

i ¼ DIDt

� 600 ð1Þ

where i is the infiltration rate (mm/h), DI is the cumulative infiltra-tion in the inner ring during Dt time period (cm), Dt is the period ofinfiltration time (min), and 600 is the conversion coefficient.

After the infiltration experiment, the infiltrated water wasallowed to drain over night, after the rings were removed. On thefollowing day, vertical soil profiles were excavated first, startingfrom a trench tangent to the position where the inner ring waslocated, then at the positions of 1/4 and 1/2 of the inner ring diam-eter (Fig. 3a). After excavation of the vertical profiles, horizontalexcavations were made at 4 depths of 5, 10, 20, and 30 cm fromthe soil surface (Fig. 3b), in the remained half of the wetted soil

for the five experimental plots in the forest soils.

J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248 245

body. The lateral width and vertical depth of the wetted volumewere measured with a 1000 mm steel ruler with precision of1 mm. Thereafter, soil samples were collected at depths of 0–10,10–20, 20–40, 40–80, and 80–100 cm on the other side of the exca-vation trench for the measurement of soil properties in the labora-tory. Three samples were collected from each soil layer. Soilproperties measured included soil texture, bulk density, and satu-rated hydraulic conductivity. The measured soil properties areshown in Table 1.

Seven profiles, including vertical and horizontal, were exca-vated in each experiment. After each profile was exposed, the soilsurface was cleaned with a portable electric fan to remove loosesoil particles and reveal strained areas. A digital camera was uti-lized to record vertical and horizontal profiles, to identify any pos-sible spots stained by the tracer dye, as evidence of preferentialflow.

The sand materials used as shelter materials for exclusion ofpreferential flow, shown in Fig. 4, were collected from Shenmue,Ordos City, in Inner Mongolia Autonomous Region (40�230–40�250N, 110�170–110�190 E). The grain size distribution of the experi-mental samples is shown in Table 1.

Fig. 6. Vertical wetted soil profiles in (a) tangent, (b) 1/4, and (c) 1/2 locations of theinner ring 24 h after infiltration.

4. Results and discussion

4.1. Forest soil matrix infiltration process

According to the laboratory experimental results, the watertook only several seconds to penetrate the sand layer. The soilmatrix infiltration rate was recorded 1 min after the commence-ment of the experiment, with Eq. (1). The measured matrix infiltra-tion rate and cumulative infiltration curves of the fiveexperimental plots were shown in Fig. 5. The steady soil matrixinfiltration rate ranged from 53 to 65 mm/h and the cumulativesoil matrix infiltration ranged from 130 to 149 mm (Fig. 5). How-ever, the steady infiltration rate of the forest soil on the LoessPlateau measured by directly applying a double-ring infiltrometerranged from 300 to 720 mm/h (Jiang and Huang, 1986). The steadymatrix infiltration rate of forest soil measured in this study was farless than that including soil matrix infiltration and preferentialflow.

According to Darcy’s equation, the soil matrix infiltration rate isequal to the water flux on the soil surface. It is given by the govern-ing differential equation:

iðtÞ ¼ qð0; tÞ ¼ �Ks@wm

@zþ Ks ð2Þ

where i(t) is the soil matrix infiltration rate at time t; q(0,t) is thewater flux on the soil surface; wm is the soil water potential, with@wm@z as the hydraulic gradient on the soil surface; z is the verticalcoordinate with a positive downward direction; and Ks is saturatedhydraulic conductivity.

Table 2The model representations of matrix infiltration into the forest soil.

Site Philip Kostiakov

S K R2 A B R2

1 71.44 28.09 0.91 63.43 �0.35 0.902 74.61 27.91 0.97 65.04 �0.35 0.953 67.82 32.41 0.97 66.07 �0.33 0.984 67.37 29.08 0.94 63.16 �0.34 0.965 57.67 31.34 0.95 58.73 �0.34 0.95

246 J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248

As t approaches zero, @wm@z approaches negative infinity and i

approaches positive infinity. As infiltration proceeds and waterenters the soil, the absolute value of @wm

@z decreases, thereby causing

i to decrease with t. When t approaches positive infinity, @wm@z

approaches zero and i approaches Ks. Thus, the change in the soilmatrix infiltration rate with t is initially high, decreases with time,and eventually reaches equilibrium. The measured soil matrix infil-tration rate (Fig. 5a) could well express the matrix infiltrationconcept.

Infiltration into soils with macropores is regarded as two-domain flow. The first domain is the soil matrix, in which wateris subjected to capillary suction, and infiltration is addressed withPhilip’s sorptivity concept. The second domain is the soil macrop-ore system, in which water moves only under gravity (Weiler,2005) and flow is addressed with kinematic wave theory (Legoutet al., 2009). Therefore, preferential flow is driven by gravity(Germann et al., 2007) and cannot be described by Richard’s norDarcy-type equations and existing infiltration models (Stumppand Maloszewski, 2010; De Vries and Chow, 1978; Weiler, 2005).The process is therefore multidimensional, which depends on thegeometry of the cracks, the conductivity of the soil matrix, andthe mode and duration of wetting (Hillel, 1998; Edwards et al.,

Fig. 7. Horizontal wetted soil sections at (a) 0, (b) 5, (c) 1

1988; Bouma et al., 1982). Therefore, to validate the method formeasuring soil matrix infiltration rate, the most commonly usedphysical infiltration model (Philip model) and the most commonlyapplied empirical model (Kostiakov model) were used to representthe measured matrix infiltration, as shown in Table 2. These twomodels fitted well with the matrix infiltration rate and cumulativeinfiltration, with R2 being greater than 0.90 and 0.99, respectively,indicating that the measured infiltration process could well repre-sent the soil matrix infiltration.

4.2. Comparison of wetting front in the field and laboratory

Brilliant Blue possesses predicted lower sorption and highermobility (Flury and Flühler, 1995). It is mainly affected by prefer-ential flow controlled by gravity and has been applied to studieson the flow pattern of water or to estimations of the frequency,size, and spatial distribution of pores in soils. Brilliant Blue dye isan important tool to visualize preferential flow pathways and areasin soil profile as flow paths. For the soil matrix without macrop-ores, Brilliant Blue FCF moved through the surface of pseudofea-tures into the subjacent soil matrix at a distance of 0.5–1 mm(Nobles et al., 2004). The vertical and horizontal profiles shown

0, (d) 20, and (e) 30 cm depth 24 h after infiltration.

+ tangent* 1/4 O 1/2

Fig. 8. Wetted areas of vertical profiles with the semielliptical fitted by Eq. (4).

J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248 247

in Figs. 6 and 7 indicate that no stained area presented. This findingindicated that the sand layer effectively sheltered the preferentialflow through macropores. Therefore the measured infiltration pro-cess is indeed the one of soil matrix.

Under the ring infiltrometer, the wetting front in a vertical pro-file as controlled by soil capillary force was approximately close toa semielliptical shape, as shown in Fig. 8 (Zhang et al., 2016;Angulo-Jaramillo et al., 2000; Chowdary et al., 2006). To quantifythe shape of the wetted area, the wetting front of different verticalwetted areas were fitted by:

x2

a2þ z2

c2¼ 1 ð3Þ

where x and z are the lateral and vertical advance distances of thewetting front, L, respectively, and a and c are the long and shortsemi-axes of the elliptical, L, respectively.

For a soil column containing preferential flow, De Vries andChow (1978) reported that the wetted area during infiltration intoa highly porous forest soil developed so irregularly that it could notbe used to calculate Darcy-type infiltration. Fig. 8 shows also thefitting results of the wetted profiles, which were very close tothe given semielliptical, with all the coefficients of determinationgreater than 0.95 (R2 = 0.99, 0.95, and 0.97 at tangent, 1/4, and1/2 locations of the inner ring, respectively). This result indicatesthat the infiltration process in the wetted soil column was con-trolled only by soil capillary action.

4.3. Spatial variability of forest soil matrix infiltration

Fig. 9 showed that the coefficients of variation (CV) of parame-ters S, Ks, a, b, and is as well as CI used to analyze the spatial vari-ability of soil matrix infiltration rate and cumulative infiltration

Fig. 9. Coefficients of variation of parameters S, K

(CI). The spatial variability of matrix infiltration rate and CI waslow for all experimental plots, with the largest CV being 11.7%.The steady soil matrix infiltration rate ranged from 53 to 65 mm/h. CI ranged from 130 to 149 mm. This conclusion is consistentwith Das Gupta et al.’s (2006) results, to demonstrate that matrixflow exhibits low variability, characterized by low K(w). This isunderstandable. The loess soil was the deposition of wind dusts(Zhu and Ren, 1992). It is commonly believed that the soil textureand the soil matrix structures are relatively uniform. Therefore thematrix infiltration should also be relatively uniform. But the pref-erential flow through macropores of random distribution couldpresents high variability.

In the results of Liu et al. (2010) and Yu (2011), the initial andsteady infiltration rates measured by a conventional double-ringinfiltrometer and by the samples taken with the corer methodshowed significant spatial variability. This spatial variability mightbe attributed to the spatial randomness of macropores and prefer-ential flows. Preferential flow processes varies temporally and spa-tially (Legout et al., 2009). The variability of preferential flowexerts a considerable influence on the hydrological response of soil(Weiler, 2005). Preferential flow is affected by many factors, suchas macropore tortuosity (Allaire-Leung et al., 2000b), macroporecontinuity (Allaire-Leung et al., 2000a), and antecedent soil mois-ture (Hardie et al., 2011). Spatially-distributed macropore charac-teristics differs significantly among similar vegetation types (Huet al., 2016). As for different soil types, preferential flow also pre-sents high variability, as verified by the staining patterns withinthe vertical and horizontal field-scale images of dye distributionin Kodešová et al.’s study (2012).

5. Conclusions

Infiltration through forest soil matrix plays very important rolein water resource conservation and root water uptake. A methodwas developed in this research to measure the matrix infiltrationcapability of forest soil. A fine desert sand layer was used on topof the forest soil surface to shelter the macropores and ensure thatwater infiltrates the soil surface only through the soil matrix. Thehigher infiltration capability of the sand layer than that of theunderlying soil ensures that the measured soil infiltration rate isnot limited by the shelter sand layer. The replicated measurementsat 5 plots indicated that the measured steady soil matrix infiltra-tion rate ranged from 53 to 65 mm/h and the matrix cumulativeinfiltration ranged from 130 to 149 mm, with low variability. Themeasured matrix infiltration expressed the concept of soil matrixinfiltration and was consistent with the values produced by well-accepted infiltration models; with coefficients of determinationall above 0.9. The result also indicates that forest soil matrix infil-tration on the Loess Plateau exhibits low spatial variability. The

s, a, b, is and CI of the five experimental sites.

248 J. Zhang et al. / Journal of Hydrology 552 (2017) 241–248

proposed method provides a new means to measure the transientprocesses of forest soil matrix infiltration. Further research shouldbe conducted on additional land use patterns.

Acknowledgements

This research was supported by the Natural Science Foundationof China under Project No. 41230746, No. 51621061 and No.41501303.

References

Allaire, S.E., Gupta, S.C., Nieber, J., Moncrief, J.F., 2002. Role of macropore continuityand tortuosity on solute transport in soils: 2. Interactions with modelassumptions for macropore description. J. Contam. Hydrol. 58 (3), 283–298.

Allaire, S.E., Roulier, S., Cessna, A.J., 2009. Quantifying preferential flow in soils: areview of different techniques. J. Hydrol. 378 (1), 179–204.

Allaire-Leung, S.E., Gupta, S.C., Moncrief, J.F., 2000a. Water and solute movement insoil as influenced by macropore characteristics: 1. Macropore continuity. J.Contam. Hydrol. 41 (3), 283–301.

Allaire-Leung, S.E., Gupta, S.C., Moncrief, J.F., 2000b. Water and solute movement insoil as influenced by macropore characteristics: 2. Macropore tortuosity. J.Contam. Hydrol. 41 (3), 303–315.

Angulo-Jaramillo, R., Vandervaere, J.P., Roulier, S., Thony, J.L., Gaudet, J.P., Vauclin,M., 2000. Field measurement of soil surface hydraulic properties by disc andring infiltrometers: A review and recent developments. Soil Tillage Res. 55 (1),1–29.

Bagarello, V., Sgroi, A., 2004. Using the single-ring infiltrometer method to detecttemporal changes in surface soil field-saturated hydraulic conductivity. SoilTillage Res. 76, 13–24.

Bouma, J., Belmans, C., Dekker, L.W., 1982. Water infiltration and redistribution in asilt loam subsoil with vertical worm channels. Soil Sci. Soc. Am. J. 46 (5), 917–921.

Chowdary, V.M., Rao, M.D., Jaiswal, C.S., 2006. Study of infiltration process underdifferent experimental conditions. Agric. Water Manag. 83 (1), 69–78.

Clancy, K., Alba, V.M., 2011. Temperature and time of day influence on double-ringinfiltrometer steady-state infiltration rates. Soil Sci. Soc. Am. J. 75 (1), 241–245.

Das Gupta, S., Mohanty, B.P., Köhne, J.M., 2006. Soil hydraulic conductivities andtheir spatial and temporal variations in a vertisol. Soil Sci. Soc. Am. J. 70 (6),1872–1881.

De Vries, J., Chow, T.L., 1978. Hydrologic behavior of a forested mountain soil incoastal British Columbia. Water Resour. Res. 14 (5), 935–942.

Doerr, S.H., Ritsema, C.J., Dekker, L.W., Scott, D.F., Carter, D., 2007. Water repellenceof soils: new insights and emerging research needs. Hydrol. Process. 21 (17),2223–2228.

Edwards, W.M., Norton, L.D., Redmond, C.E., 1988. Characterizing macropores thataffect infiltration into nontilled soil. Soil Sci. Soc. Am. J. 52 (2), 483–487.

Everts, C.J., Kanwar, R.S., 1990. Estimating preferential flow to a subsurface drainwith tracers. Trans. ASAE 33 (2), 451–0457.

Flury, M., Flühler, H., 1995. Tracer characteristics of brilliant blue FCF. Soil Sci. Soc.Am. J. 59 (1), 22–27.

Germann, P., Helbling, A., Vadilonga, T., 2007. Rivulet approach to rates ofpreferential infiltration. Vadose Zone J. 6 (2), 207–220.

Greve, A., Andersen, M.S., Acworth, R.I., 2010. Investigations of soil cracking andpreferential flow in a weighing lysimeter filled with cracking clay soil. J. Hydrol.393 (1), 105–113.

Hardie, M.A. et al., 2011. Effect of antecedent soil moisture on preferential flow in atexture-contrast soil. J. Hydrol. 398 (3), 191–201.

Hillel, D., 1998. Environmental Soil Physics. Academic Press, New York.Hu, X., Li, Z., Li, X., Liu, L., 2016. Quantification of soil macropores under alpine

vegetation using computed tomography in the Qinghai Lake Watershed, NEQinghai-Tibet Plateau. Geoderma 264, 244–251.

Jabro, J.D., Lotse, E.G., Fritton, D.D., Baker, D.E., 1994. Estimation of preferentialmovement of bromide tracer under field conditions. J. Hydrol. 156 (1), 61–71.

Jarvis, N.J., 2007. A review of non-equilibrium water flow and solute transport insoil macropores: Principles, controlling factors and consequences for waterquality. Eur. J. Soil Sci. 58 (3), 523–546.

Jiang, D.S., Huang, J.G., 1986. Study on infiltration rate of soils on the Loess Plateauof China. Acta Pedol. Sin. 23 (4), 299–305.

Klute, A., Dirksen, C., 1986. Hydraulic conductivity and diffusivity: Laboratorymethods. Methods of Soil Analysis: Part 1—Physical and Mineralogical Methods(methods of soil an 1): pp. 687–734.

Kodešová, R., Šimunek, J., Nikodem, A., Jirku, V., 2010. Estimation of the dual-permeability model parameters using tension disk infiltrometer and Guelphpermeameter. Vadose Zone J. 9 (2), 213–225.

Kodešová, R., Nemecek, K., Kodeš, V., Zigová, A., 2012. Using dye tracer forvisualization of preferential flow at macro-and microscales. Vadose Zone J. 11(1).

Köhne, J.M., Mohanty, B.P., 2005. Water flow processes in a soil column with acylindrical macropore: experiment and hierarchical modeling. Water Resour.Res. 41 (3).

Lassabatere, L., Angulo-Jaramillo, R., Soria Ugalde, J.M., Cuenca, R., Braud, I.,Haverkamp, R., 2006. Beerkan estimation of soil transfer parameters throughinfiltration experiments—BEST. Soil Sci. Soc. Am. J. 70 (2), 521–532.

Legout, A., Legout, C., Nys, C., Dambrine, E., 2009. Preferential flow and slowconvective chloride transport through the soil of a forested landscape(Fougères, France). Geoderma 151 (3), 179–190.

Lei, T., Zhang, J., Wang, W., Ma, Y., 2013. Assessment on soil infiltration ratesmeasured by ring infiltrometer. Trans. Chin. Soc. Agric. Mach. 44, 99–104.

Li, Y., Ghodrati, M., 1997. Preferential transport of solute through soil columnscontaining constructed macropores. Soil Sci. Soc. Am. J. 61 (5), 1308–1317.

Liu, X., Ma, J., Zhang, Z., 2010. Spatial variability of soil infiltration characteristicsand its pseudo-transfer functions. Adv. Water Sci. 21 (214–221).

Logsdon, S.D., Jaynes, D.B., 1993. Methodology for determining hydraulicconductivity with tension infiltrometers. Soil Sci. Soc. Am. J. 57 (6), 1426–1431.

Mooney, S.J., Morris, C., 2008. A morphological approach to understandingpreferential flow using image analysis with dye tracers and X-ray computedtomography. Catena 73 (2), 204–211.

Nobles, M.M., Wilding, L.P., McInnes, K.J., 2004. Submicroscopic measurements oftracer distribution related to surface features of soil aggregates. Geoderma 123(1), 83–97.

Perroux, K.M., White, I., 1988. Designs for disc permeameters. Soil Sci. Soc. Am. J. 52(5), 1205–1215.

Reynolds, W.D., Elrick, D.E., 1991. Determination of hydraulic conductivity using atension infiltrometer. Soil Sci. Soc. Am. J. 55 (3), 633–639.

Reynolds, W.D., Zebchuk, W.D., 1996. Use of contact material in tensioninfiltrometer measurements. Soil Technol. 9 (3), 141–159.

Ronayne, M.J., Houghton, T.B., Stednick, J.D., 2012. Field characterization ofhydraulic conductivity in a heterogeneous alpine glacial till. J. Hydrol. 458,103–109.

Sander, T., Gerke, H.H., 2007. Preferential flow patterns in paddy fields using a dyetracer. Vadose Zone J. 6 (1), 105–115.

Schoen, R., Gaudet, J.P., Bariac, T., 1999. Preferential flow and solute transport in alarge lysimeter, under controlled boundary conditions. J. Hydrol. 215 (1), 70–81.

Šimunek, J., Genuchten, M.V., 1996. Estimating unsaturated soil hydraulicproperties from tension disc infiltrometer data by numerical inversion. WaterResour. Res. 32 (9), 2683–2696.

Stagnitti, F., Parlange, J.-Y., Steenhuis, T.S., Boll, J., Pivetz, B., Barry, D.A., 1995.Transport of moisture and solutes in the unsaturated zone by preferential flow.In: Singh, V.P. (Ed.), Environmental Hydrology. Kluwer Academic Publishers,Dordrecht.

Stumpp, C., Maloszewski, P., 2010. Quantification of preferential flow and flowheterogeneities in an unsaturated soil planted with different crops using theenvironmental isotope d 18O. J. Hydrol. 394 (3), 407–415.

Sun, L., 2013. Transportation and path distribution of preferential flow in citrus soilsof purple sandstone regions, Three Gorges reservoir area Dissertation forMaster’s Degree. Beijing Forestry University, Beijing, China.

Wang, D., Norman, J.M., Lowery, B., McSweeney, K., 1994. Nondestructivedetermination of hydrogeometrical characteristics of soil macropores. Soil Sci.Soc. Am. J. 58 (2), 294–303.

Wang, W., Zhang, H.J., Cheng, J.H., Wu, Y.H., Du, S.C., Wang, R., 2010. Macroporecharacteristics and its relationships with the preferential flow in broad leavedforest soils of Simian Mountains. Chin. J. Appl. Ecol. 21 (5), 1217–1223.

Weiler, M., 2005. An infiltration model based on flow variability in macropores:development, sensitivity analysis and applications. J. Hydrol. 310 (1), 294–315.

Wu, J., Zhang, J., Gao, R., 2009. Physical simulation experiments of effects ofmacropores on soil water infiltration characteristics. Trans. Chin. Soc. Agric.Eng. 25 (10), 13–18.

Yu, K., 2011. The spatial variability of soil infiltration capacity at Loess Hilly Area.Lanzhou University.

Zhang, J., Lei, T.W., Chen, T.Q., 2016. Impact of preferential and lateral flows of wateron single-ring measured infiltration process and its analysis. Soil Sci. Soc. Am. J.80 (4), 859–869.

Zhu, X., Ren, E.M., 1992. Formation process of the Loess Plateau and a strategy of itsmanagement. Soil Water Conserv. China 2, 4–11.