Characteristics of overland flow generation on steep forested hillslopes of central Japan

Journal of Hydrology (2008) 361, 275– 290

available at www.sciencedirect.com

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

Characteristics of overland flow generation on steep forested hillslopes of central Japan Takashi Gomi a,*, Roy C. Sidle b, Masayasu Ueno b, Shusuke Miyata a, Ken’ichirou Kosugi c a

Department of International Environment and Agriculture Science, Tokyo University of Agriculture and Technology, 3-5-8 Saiwai-cho, Fuchu, Tokyo 183-8509, Japan b Geohazards Division, Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji 611-0011, Japan c Laboratory of Erosion Control, Division of Forest and Biomaterial Science, Gradate School of Agriculture, Kyoto University, Kyoto 606-8502, Japan Received 9 November 2007; received in revised form 8 July 2008; accepted 30 July 2008

KEYWORDS Forest management; Hillslope scale; Hortonian overland flow; Runoff generation; Saturation overland flow; Understory vegetation

Summary Overland flow generation was monitored in large plots (8 · 25 m) on four hillslopes in a 4.9-ha catchment in Mie Prefecture, Japan. Three Japanese cypress (hinoki, Chamaecyparis obtusa) treatments (including three different understory conditions) and one deciduous forest treatment were studied. For all plots, including deciduous hillslopes, we observed overland flow even for small storm events (<10 mm in total precipitation). The mean runoff coefficients in dense Japanese cypress plots with sparse understory were highest (13.0%) followed by dense Japanese cypress with fern ground cover (6.7%), and coefficients in managed cypress and deciduous forest were 3.6% and 1.2%, respectively. The runoff coefficients tended to be higher during storms that were preceded by dry conditions. High soil water repellency initially occurred in Japanese cypress forests between the litter and mineral soil horizon and might have been partly responsible for overland flow generation. During storms with total precipitation >180 mm, runoff from Japanese cypress plots with dense fern understory exhibited a delayed and higher peak associated with return flow. The dominance of hillslope-scale flow contribution to catchment runoff was also affected by changes in the dominance of overland flow and return flow. Understory vegetation cover and the availability of a litter layer altered the amount of overland flow, which was mediated by soil water repellency and soil moisture. Observations at the hillslope scale are essential for conceptualization of runoff mechanisms and pathways in forested headwaters. ª 2008 Elsevier B.V. All rights reserved.

* Corresponding author. Tel./fax: +81 42 367 5751. E-mail address: [email protected] (T. Gomi). 0022-1694/$ - see front matter ª 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2008.07.045

276

Introduction The condition of the soil surface controls the flux of water from the atmosphere into the soil matrix. Although infiltration-excess overland flow is a major stormflow generation process that results in a quick runoff response in some catchments (Horton, 1933), most forest soils have high infiltrability that promote subsurface and saturated overland flow as the dominant runoff mechanisms (Tsukamoto, 1963; Hewlett and Hibbert, 1967; Dunne and Black, 1970). Hortonian overland flow may occur on hillslopes affected by fire (e.g., Shakesby et al., 1993), roads and skid trails (e.g., Ziegler and Giambelluca, 1997), and prolonged drought (e.g., Zehe et al., 2007). The extent and continuity of induced hydrophobicity and resultant Hortonian overland runoff from burnt forest soils depends on the severity of the fire (Doerr et al., 1998; Cannon and Reneau, 2000). Overland flow has also been observed in Japanese cypress (hinoki, Chamaecyparis obtusa) forests with sparse understory vegetation cover (Onda and Yukawa, 1994; Miura et al., 2002; Nanko et al., 2008; Fukuyama et al., 2008; Gomi et al., 2008). On hillslopes with Japanese cypress stands, soil water repellency at the 5–10 cm depth below the soil surface also promotes overland flow (Kobayashi and Shimizu, 2007; Miyata et al., 2007). Flow above the mineral soil – organic horizon interface associated with transient saturation of the surface of mineral soil may commonly provide a rapid pathway for the runoff of storm water and nutrients from forested hillslopes that have organic horizons and well-developed root networks (Burch et al., 1989; Buttle and Turcotte, 1999; Baudoux et al., 2006; Scherrer et al., 2007; Sidle et al., 2007). Overland flow from hillslopes is highly variable both spatially and temporally. Microtopographic patterns (e.g., surface depressions, roughness) cause spatial variability in ponding and preferential infiltration (Julien and Moglen, 1990; Dunne et al., 1991). Spatially variable soil moisture and soil physical properties (e.g., clay composition, buried organic matter) alter the occurrence and pathways of both overland and subsurface flow (Sharma et al., 1980; Burch et al., 1989; Sidle et al., 2000; Ziegler et al., 2001; Uchida et al., 1999; Godsey et al., 2004). Water-repellent soils may produce localized areas of high Hortonian overland flow and preferential vertical infiltration (Imerson et al., 1992; Kobayashi and Shimizu, 2007). Such effects may be obscured at the hillslope scale due to infiltration ‘hot spots’ or lack of connectivity between localized overland flow source areas (Gomi et al., 2008). The effect of hillslope position has generally not been considered in small-scale runoff studies, although runoff mechanisms likely differ as a result of such influences on soil moisture (Gascuel-Odoux et al., 1996; Hung et al., 2001). The general decreases noted in runoff coefficients at increasing hillslope lengths also suggest scaling phenomenon of overland flow generation (van de Giesen et al., 2000; Joel et al., 2002; Cerden et al., 2004; Gomi et al., 2008). Yet, the effects of these phenomena of overland flow generation on hillslope- to catchment-scale storm runoff processes are unknown (Sidle et al., 2007; Gomi et al., 2008). Current hydrological models that use empirical formulae based on small plots, averaged catchment-scale analogues, or both may overestimate the importance of overland flow contributions to streams.

T. Gomi et al. We examined hillslope-scale overland flow generation in large, unreplicated hillslope-scale plots for various conditions of understory and overstory vegetation cover in Japanese cypress forests and in an adjacent deciduous forest. Our objectives were to quantify the amount of overland flow generation on hillslopes with various forest and understory conditions. We also evaluated the role of vegetation ground cover on overland flow generation from steep forested hillslopes. We then demonstrate the role of vegetation ground cover on overland flow generation at the hillslope scale.

Study site and methods The study was conducted within a 4.9-ha catchment (catchment 1) located in central Mie Prefecture (3421 0 N, 13625 0 E; altitude: 100–260 m), south-central Japan (Fig. 1). The climate of this area is moist and temperate, with mean annual precipitation of approximately 2000 mm and mean annual air temperature of 14 C. The rainfall regime is bimodal: the Baiu season from late May through June, and the typhoon season from late August through October. Soils are Cambisols (brown forest soils in the Japanese classification) ranging in depth from 0.6 to 1.8 m. The soils are relatively shallow on lower hillslopes and thicker near mid-slope and ridgeline positions (Fig. 2). The combined A and B horizons are approximately 25–30 cm thick, underlain by a C horizon that is typically >35 cm thick. Thickness of the litter layer varies from 0 to 3.5 cm, depending on the vegetation cover (Table 1). The catchment is deeply incised with a dominant hillslope gradient of 35–45. The forest is predominantly a 40-yr-old stand of Japanese cypress (hinoki, Chamaecyparis obtusa), with a few small inclusions of Japanese cedar (sugi, Cryptomeria japonica) and broadleaf forest. The dominant understory vegetation is fern (Gleichenia japonica) and evergreen shrubs (e.g., Cleyera japonica). A nested monitoring network was installed in the catchment in spring 2004 to evaluate runoff and sediment transport from Japanese cypress stands of different density and management legacies. Each subcatchment (1.2–0.2 ha in area) was characterized based on management legacy, stand density, and understory vegetation cover (Table 1; Fig. 3). Catchment 5 had dense (4500 stems/ha) cypress cover with sparse understory vegetation (Fig. 3a), whereas catchment 4 had less dense (3500 stems/ha) cypress and more abundant fern understory (Fig. 3b). Because of thinning after plantation establishment, the stand in catchment 2 had fewer cypress stems (1500 stems/ha) and the largest average stand diameter of all cypress stands (Fig. 3c). Thinning at this area was conducted by chainsaw, and logs were manually removed by foresters. Thus, soil surface disturbance and compaction was minimized, and the litter layer remained on the hillslope. Adjacent to catchment 1, we also monitored a small catchment (catchment 8) that was covered by deciduous forest (Fig. 3d). Hillslope plots were established within the four subcatchments to monitor overland flow. Plot 1 was located within catchment 5, plot 2 in catchment 4, plot 3 in catchment 2, and plot 4 in catchment 8 (Fig. 3). Hillslopes with predominantly planar topography and little internal roughness (e.g., extensive woody debris, boulders, slope breaks)

Characteristics of overland flow generation on steep forested hillslopes of central Japan

Figure 1

277

Location and topography of study catchments and plots.

were selected for plot installation (Fig. 1). Slope widths were approximately 8 m for all runoff plots, whereas slope length ranged from 24 to 27 m from the ridge (Table 1; Fig. 1). Thus, the slope position from the valley bottom varied among plots. The sides of all plots were unbordered to avoid disrupting natural flow paths (Williams and Bonell, 1988), whereas the upper boundaries of the plots were topographic ridges. The gradients of the plots ranged from 39 to 43; thus, projected plot areas ranged from 108 to 130 m2 (Table 1). Understory vegetation cover in plot 1 was <10%. In plot 2, most (85%) of the ground surface, particularly in the upper part of the plot, was covered by ferns (Table 1). In plots 3 and 4, understory vegetation was 100%, consisting of ferns and shrubs. The soil surface was covered

by a 3- to 3.5-cm layer of leaf litter in plots 2–4. Much of the mineral soil horizon was exposed in plot 1, although patchy accumulations of needles and branches were observed. At the lower boundary of the runoff plots, plastic troughs were installed parallel to the ridges and slope contours to collect surface runoff and sediment. Flexible aluminum flashing was installed approximately 2–3 cm beneath the soil surface to facilitate the effective routing of runoff into the troughs (Fig. 3). Thus, we collected runoff occurring at and above the mineral soil – organic horizon interface and within the uppermost root-permeated soil (Sidle et al., 2007). The troughs were covered with plastic roofs to avoid the effects of direct precipitation. Water collected in the troughs was directed through a drop-box 45 V-notch weir

278

T. Gomi et al.

A transect in catchment 5 (near Plot 1)

A transect in catchment 4 (near Plot 2) Nc

Nc 20

40

60

0

50

Nc 100 0 150

Nc

30 Soil depth (cm)

20

0

0

20

40

60

50

0

20

40

60

1-c Soil depth (cm)

40 m

0

Soil depth (cm)

Soil depth (cm)

0

1-c

0

20

40

60

50

100

2-c

150 50

2-c 100

2-b 150

2-b

100

1-b

150

Nc

Nc 0

Soil depth (cm)

10 1-a

20

40

60

0

2-a

0

Soil depth (cm)

1-b

50

Valley bottom

100

1-a

0

0

20

40

60

50

100

150

2-a

150

Valley bottom

0

10

20

30

0

40 m

10

20

30

40

Figure 2 Slope profile and soil physical characteristics of catchments 4 and 5. Locations of hillslope transects are indicated in Fig. 1. Soil physical properties along the profile were measured using a portable dynamic cone penetrometer (knocking pole penetrometer). Penetration resistance (Nc) is expressed as the number of blows needed to drive the cone 10 cm into the soil. Based on Nc, the soil thickness of the hillslope ranged from 1.0 to 1.5 m.

Table 1

Outline of study plots

Plot No.

Width (m)

Length (m)

Trajection area (m2)

Slope gradient (degree) Lower half

Upper half

1

8.0

24.6

118.0

42.5

2

8.0

23.7

108.0

3

7.9

23.5

4

8.0

26.0

Average

Stand density (stems/ha)

Forest type

Mean litter thickness (cm)

43.0

43.0

4500

0.0

10

Sparse

43.0

42.0

42.5

3500

3.3

85

Fern

120.0

42.5

39.5

41.0

1500

3.4

100

130.0

44.0

38.0

41.0

na

Japanese cypress Japanese cypress Japanese cypress Deciduous

3.0

100

attached to the downslope end of the trough (Fig. 3). The water stage near the inlet of the V-notch weir was monitored every 5 min using capacitance probes (TruTrack) calibrated to discharge. Monitoring of plots 1 and 2 initiated on 5 May 2004; monitoring of plots 4 and 5 began in November 2004 and March 2005, respectively. Precipitation was measured by a tipping bucket rain gauge located in an open area 200–300 m away from the study plots.

Understory vegetation cover (%)

Understory vegetation

Fern and evergreen Deciduous and evergreen

Three undisturbed soil core samples were collected in an area adjacent to the hillslope plots at depths of 5, 12.5, 20, and 40 cm to measure saturated hydraulic conductivity (Ks), organic matter content, and particle size distribution. Cores were taken using 100-cm3 steel cylinders with cross-sectional areas of 20 cm2 and heights of 5.1 cm. In the laboratory, the soil cores were saturated for 48 h, after which Ks was measured by a constant head test (Reynolds et al.,

Characteristics of overland flow generation on steep forested hillslopes of central Japan

279

Figure 3 Characteristics of forest stands and ground cover in the four study plots: (a) dense Japanese cypress forest with sparse understory vegetation; (b) dense Japanese cypress forest with fern understory vegetation; (c) managed Japanese cypress forest with fern and evergreen shrubs; and (d) deciduous forest. Fig. 2e shows the installation of troughs to capture overland flow and the dropbox 45 V-notch weir used to measure overland flow.

2002). The samples were dried and weighed to measure bulk density, and subsamples were then ashed for 2 h at 550 C to estimate organic matter content. The samples were sieved into size classes of 8.0, 4.0, 2.0, 1.0, 0.5, 0.25, and 0.106 mm to determine particle size distribution. Soil samples were collected at depths of 0, 5, 10, and 20 cm to test for water repellency. Soils were dried at 60 C for 48 h in the laboratory before measurement (de Jonge et al., 1999). The samples were equilibrated at ambient laboratory conditions (air temperature: 25o; air moisture: 40%) for 2 days because soil water repellency is affected by air temperature and relative humidity (Doerr et al., 2002). The sensitivity was examined using the critical

surface tension (CST) test, in which drops of ethanol solutions with different concentrations were placed on the soil surface and the time required for the drops to infiltrate was measured (Watson and Letey, 1970; Miyata et al., 2007). Ethanol solutions with volumetric ethanol concentrations of 0%, 1%, 3%, 5%, 8.5%, 16% and 30% were used to examine hydrophilic to strong hydrophobic conditions (Doerr, 1998). Five drops of solution were placed on the surface of soil samples using a micropipette. If all drops did not infiltrate the samples within 5 s, solutions with successively higher concentrations were applied. When all drops infiltrated within 5 s, the ethanol concentration was taken as the resultant score.

280

T. Gomi et al.

Transects of wells were installed along the sides of plots 1 and 2 in spring 2005 to identify the location and extent of water tables related to the occurrence of overland flow. Wells were augured to bedrock (solid layer) at depths of 0.8–1.5 m below the surface. Water level loggers were installed in selected wells along the lower side of these plots adjacent to the weir. Crest gauges were placed in the other wells to measure the maximum water table between sampling intervals. Soil depths along the sides of these plots were measured using a portable dynamic cone penetrometer (knocking pole penetrometer) with a 25 mm diameter cone and a weight of 5 kg. Penetration resistance was expressed as the number of ‘‘knocks’’ needed to drive the cone to depth intervals of 10 cm. Soil moisture was measured at the 5 cm depth using a capacitance probe (Easy AG, Sentek Pty. Ltd., Stepney, Australia) in catchment 5 (Fig. 1). Storm events were defined as precipitation that produced overland flow. An inter-storm period (i.e., that would separate individual events) is defined as an interval of at least 6 h with no rainfall. Because the amount of overland flow decreased quickly after precipitation ceased, a 6-h period without precipitation was sufficient to distinguish between individual storm events. All hill-

slope overland runoff was assumed to flow perpendicular to the contours of the planar hillslope plots. The total runoff volume (mm) was divided by the projected area of hillslope plots. Runoff coefficients were calculated as total runoff depth divided by total storm precipitation. For each storm event, a threshold precipitation for initiating overland flow (the initial increase in stage) was also estimated. Relations between total storm overland flow and both precipitation and antecedent soil moisture were examined to identify factors affecting runoff from the plots. Total precipitation; maximum 5-, 20-, and 60-min storm intensities; 7- and 30-day antecedent precipitation indices (API7 and API30); and initial soil moisture at the 5 cm depth were used in these analyses. API7 and API30 were defined as the sum of precipitation during the preceding 7 and 30 days, respectively. We assumed API7 represents surface soil moisture conditions, whereas API30 represents soil moisture deeper in the profile (Sidle et al., 2000). Threshold precipitation was estimated as the total amount of precipitation required to obtain the first response of overland flow in each event. Correlation analyses were conducted for overland flow and soil and precipitation indices. All variables were log transformed.

0

100

10

80

40

40 Extreme

Slight

30

60 Very strong

20

Moderate to strong

Plot1

5 cm 12.5 cm 20 cm 40 cm

20 0 100

0

Plot2

80 10 60

Percentage finer (%)

Plot3

Soil depth (cm)

20 30 40 0 10 20

40 20 0 100 80 60 40

30

20 40 0

0 100

10

80

20

60

Plot4

40

30

20 40 0 0

10

20

30

Ethanol concentration (%)

500

1000

Ks (mm/hr)

5000 0.6

0.8

1.0

Density (g/cm)

1.2 6

7

8

9

10

Organic matter content (g/100ml)

11

0

5

10

15

Particle size (mm)

Figure 4 Soil hydrophobicity, saturated hydraulic conductivity, bulk density, organic matter content, and soil particle size distribution in plots with various soil depths.

Summary table for observed strom event and overland flow

Date

2004/05/31 2004/06/05 2004/06/10 2004/06/19 2004/06/25 2004/06/30 2004/07/04 2004/07/10 2004/07/30 2004/08/01 2004/08/03 2004/08/07 2004/08/15 2004/08/23 2004/08/30 2004/09/04 2004/09/05 2004/09/24 2004/09/27 2004/09/28 2004/10/05 2004/10/08 2004/10/19 2004/10/30 2004/11/11 2004/12/04 2004/12/30 2005/02/18 2005/03/03 2005/04/20 Sub-Total

Max rain intensity 5 min (mm)

20 min (mm)

60 min (mm)

API7 (mm)

API30

Initial soil water content at 5 cm depth (%)

Plot 1

Plot 2

Plot 3

Plot 4

Plot 1

Plot 2

Plot 3

Plot 4

12.4 82.4 67.2 290 6 4.6 32.8 12.8 195.2 55.2 194.4 8.4 14.4 48.6 102.2 17.8 39 39.6 9 345 44.8 178.2 237.8 67.4 54.2 122.6 11.6 20.2 16.6 26

1.8 3.6 2.4 2.6 0.4 0.8 2.2 3.6 8.2 2.2 2.2 2.8 3.8 3.2 7.8 4.2 2.6 4.0 2.2 7.6 1.0 3.4 3.6 1.4 3.4 4.6 0.4 0.2 0.4 1.4

3.8 5.4 10.0 23.6 1.0 1.4 3.8 9.4 29.4 5.4 7.4 6.0 8.0 11.6 24.2 12.4 4.8 9.6 4.2 25.8 3.4 8.6 12.8 4.2 9.4 14.2 1.6 0.8 1.0 3.6

7.4 9.6 22.6 61.2 1.4 1.4 4.4 12.8 32.6 6.4 9.6 7.8 8.2 17.2 28.8 8.6 5.4 12.4 5.2 29.0 7.4 20.6 32.0 7.8 16.0 26.8 4.4 1.0 2.4 7.6

0.0 12.4 87.2 0.4 260.4 34.4 6.6 203.0 3.0 203.3 257.2 259.2 0.4 10.4 75.8 113.4 44.8 2.8 44.8 56.6 341.6 56.8 0.2 66.2 0.8 0.2 9.8 20.2 0.0 0.0

– – – 245.4 460.6 467.2 459.4 406.2 56.4 251.8 308.6 474.2 468.2 493.0 565.8 310.0 243.6 186.4 226.8 227.4 422.2 420.2 594.8 511.8 315.4 80.2 138.4 32.0 64.8 43.6

– – – – 24.6 25.2 23.9 26.4 15.8 24.2 – 26.7 22.2 23.3 28.2 28.7 30.7 21.7 28.1 30.0 29.6 .825 24.7 27.4 23 22.4 21.7 24.6 25.1 20.0

0.85 9.85 7.21 18.31 0.01 0.09 3.89 1.18 39.61 4.58 15.20 0.16 2.47 4.90 7.72 1.15 1.25 4.36 0.62 19.23 1.16 8.82 12.88 – 3.81 10.28 1.47 3.72 1.33 5.48

– – – 26.15 0.00 0.03 1.71 0.78 23.74 3.11 12.50 0.09 1.18 2.61 6.67 0.27 0.87 2.04 0.16 28.14 0.77 5.26 24.39 6.92 1.86 7.18 0.49 1.83 0.13 2.36

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

– – – – – – – – – – – – – – – – – – – – – – – 0.11 0.07 0.39 0.09 0.02 0.01 0.11

5.43 9.21 7.92 4.66 0.25 1.51 11.84 9.19 20.29 8.30 7.83 1.95 17.13 10.08 7.56 6.44 3.22 11.01 6.93 5.56 2.58 4.95 5.42 11.13 7.03 8.38 12.66 18.42 8.03 21.09

– – – 9.02 0.25 0.46 5.20 6.06 12.16 5.64 6.46 1.11 8.18 5.36 6.52 1.56 2.24 5.14 1.75 8.14 1.78 2.95 10.16 10.26 3.44 5.85 4.24 9.06 0.79 9.09

– – – – – – – – – – – – – – – – – – – – – – – – – – – – – –

– – – – – – – – – – – – – – – – – – – – – – – 0.16 0.13 0.32 0.73 0.73 0.03 0.43

191.59

161.24

–

0.80

8.13

7.35

–

0.25

1.53 1.56 9.53 2.84 2.25 5.88 2.37

0.32 1.00 2.83 0.93 1.47 2.08 2.35

0.19 0.11 1.54 0.22 0.25 1.23 0.38

0.04 0.03 0.20 0.10 0.02 0.54 0.28

2356.4 9.6 6.0 21.0 12.8 10.4 25.0 22.0

0.6 0.4 0.4 1.6 0.4 5.6 2.6

3.6 1.6 1.6 2.4 1.6 13.6 6.2

3.4 2.8 2.6 8.6 3.4 19.6 7.8

1.2 2.2 6.8 28.0 12.8 0.6 27.4

46.4 46.6 43.4 63.6 48.8 104.6 110.2

17.6 13.8 17.4 15.6 16.2 15.4 19.3

Total runoff (mm)

Runoff coefficient (%)

15.91 3.28 1.97 0.43 26.08 16.66 1.80 0.53 27.90 4.60 1.90 0.96 19.00 7.00 1.60 0.74 22.50 14.70 2.49 0.22 23.55 8.30 2.90 1.11 10.80 10.70 1.72 1.28 (Continued on next page)

281

2005/05/01 2005/05/30 2005/06/03 2005/06/05 2005/06/11 2005/07/01 2005/07/03

Total rain (mm)

Characteristics of overland flow generation on steep forested hillslopes of central Japan

Table 2

1.19 0.87 3.62 3.62

Results and discussion Soil surface conditions and soil physical properties

617.4 2973.8 Sub-Total Total

Initial soil moisture content at 5 cm soil depth was measured in catchment 5. Detail description of soil moisture measurement was shown in Miyata et al., 2007.

40.62 201.87 79.96 271.55

22.34 22.34

7.31 8.11

12.95 9.13

6.58 7.18

0.54 0.53 2.30 1.60 1.13 0.17 0.80 0.30 0.78 1.04 0.23 0.97 7.60 6.30 2.40 1.50 3.90 1.40 3.70 1.97 0.37 0.97 7.88 6.85 0.82 0.28 0.62 0.26 3.29 8.31 133.0 133.0 87.0 97.2 155.0 174.2 178.4 142.4 160.0 227.0 5.6 22.0 87.2 52.6 19.0 23.6 27.2 16.2 62.0 195.2 2005/07/04 2005/07/09 2005/07/26 2005/08/18 2005/08/19 2005/08/22 2005/08/25 2005/08/31 2005/09/05 2005/09/07

0.6 1.8 1.6 6.8 6.0 1.2 1.6 1.0 3.6 6.2

1.6 4.6 5.4 16.4 13.0 3.0 2.4 2.0 8.6 13.6

3.2 7.8 13.6 28.2 15.0 3.8 6.0 3.6 17.4 19.2

57.8 39.6 0.0 4.8 60.6 79.8 51.4 34.2 17.6 87.2

– 14.5 15.6 15.0 17.5 18.6 20.2 19.2 20.2 –

0.09 1.21 15.95 12.26 1.06 0.82 2.23 0.35 6.83 13.20

0.13 0.21 6.25 3.33 0.46 0.64 1.06 0.23 2.28 3.84

0.03 0.12 2.05 0.86 0.22 0.04 0.22 0.05 0.49 2.04

1.63 5.50 18.20 23.30 8.50 3.40 8.20 2.20 11.00 6.80

0.63 4.39 9.00 13.00 4.30 1.10 2.30 1.60 5.30 4.30

Plot 4 Plot 3 Plot 2 Plot 1 API30 5 min (mm) Total rain (mm)

Max rain intensity Date

Table 2 (Continued)

20 min (mm)

60 min (mm)

API7 (mm)

Initial soil water content at 5 cm depth (%)

Total runoff (mm)

Plot 4

Plot 1

Plot 2

Plot 3

T. Gomi et al.

Runoff coefficient (%)

282

Saturated hydraulic conductivity (Ks) ranged from 570 to 6920 mm h1 (Fig. 4). There were no significant differences in Ks among the four soil depths sampled in plots 1 and 2, whereas relatively high Ks occurred near the soil surface in plots 3 and 4. Miyata et al. (2007) also reported Ks values ranging from 540 to 7090 mm h1 in the same study site. Nevertheless spatially representative values of saturated hydraulic conductivity for the hillslope scale were difficult to obtain because sample dimension (i.e., core size) affects saturated hydraulic conductivity (De Gruijter et al., 2006). Soil bulk density ranged from 0.62 to 1.11 g cm3, with relatively low values in surface soils compared to deeper soil (Fig. 4). Soil organic matter content decreased with depth in all plots. Although plot 1 had sparse understory and litter cover, the organic matter content near the soil surface (5 cm depth) was similar to that of the other plots. However, we observed that the structure and form (leaf, root, and particulate size) of organic matter near the soil surface differed among plots. Soil near the surface contained more sand than the deeper soil (Fig. 4). The percentages of ethanol concentration needed to induce infiltration of applied solution droplets within 5 s indicate the relative strength of soil hydrophobicity (Fig. 4). For all hillslopes with Japanese cypress forest cover, surface and near-surface soil had ‘‘very strong’’ hydrophobicity. Surface soil was slightly hydrophobic in the deciduous forest. Soil samples from the 10 cm depth were not hydrophobic in all sites. These findings agree with more detailed studies of hydrophobicity in Japanese cypress forests (Miyata et al., 2007). Such hydrophobic soil conditions related to depth were also reported in temperate mountain soils (Barrett and Slaymaker, 1989).

Rainfall and runoff characteristics Thirty-five rainfall events occurred during the study period in 2004 and 2005; 80% of these events occurred during the typhoon season of 2004. After all plots were installed in mid-April 2005, 17 storm events were observed until 7 September 2005. Annual precipitation in 2004 was 3201 mm, the highest since 1976 according to records from the Kayumi climate station located 9 km northwest of the study site. Indeed, a record number of typhoon storms struck Japan in 2004. In contrast, annual precipitation in 2005 (1383 mm) was the fifth lowest in the 20-year climate record. The maximum total precipitation was 345 mm on 28 September 2004, and the maximum 1 h precipitation was 61.2 mm on 19 June 2004 (Table 2). The Baiu rainy season from May to June 2005 was relatively dry compared to the long-term average (Fig. 5a and 5b). The cumulative runoff from plot 1, which had sparse understory vegetation, was higher than that from plot 2 (Fig. 5c). The mean runoff coefficients after mid-April 2005 in plots 1 and 2 were 13.0% and 6.7%, respectively, compared to 3.6% and 1.2% for plots 3 and 4, respectively. Relatively high runoff coefficients were observed during the dry season from May to September 2005 (Fig. 5d and e).

Characteristics of overland flow generation on steep forested hillslopes of central Japan

API7 (mm)

Typhoon season

Baiu season

Typhoon season

0

0

1000

20

2000

40 3000

60

a

400

b

4000

Cumulative precipitation (mm)

1 hour precipitation (mm)

Baiu season

283

200

Total Storm runoff (mm)

300 40

Plot 1 Plot 2

c

200

20 100 0 30

0

Cumulative runoff height (mm)

0

Storm runoff coefficient

d Plot 1 20 10 0 30

e Plot 2

20 10

20 05 /0 8

20 05 /0 6

20 05 /0 4

20 05 /0 2

20 04 /1 2

20 04 /1 0

20 04 /0 8

20 04 /0 6

0

Figure 5 One-hour precipitation, cumulative precipitation, and antecedent precipitation index (7 days) during the monitoring period from June 2004 to September 2005 (Fig. 5a and 5b). Total runoff amount (mm), cumulative runoff amount, and runoff coefficients for plots 1 and 2 during the monitoring period are also shown (Fig. 5c–e).

The total amount of runoff was significantly correlated with the duration of precipitation for all plots (correlation coefficient: 0.65–0.77, p < 0.01), whereas the runoff coefficients were not significantly correlated with the duration of precipitation. The total storm precipitation and maximum 20- and 60-min storm intensities were positively correlated with total overland volume for all plots (Table 3), whereas the correlation coefficients for maximum 5-min precipitation were not significant in plots 1 and 3. The initial soil moisture and 7- and 30-day antecedent precipitation indices (API7 and API30, respectively) were not significantly correlated with total runoff in all plots (Table 3). Similarly, Buttle and Turcotte (1999) showed that runoff through the organic

horizon was positively correlated with rainfall intensity on a forested hillslope in Canada.

Processes of overland flow generation Characteristics of precipitation controlled the dominant mechanisms of runoff generation on the hillslopes. During small storms, the runoff coefficients of overland flow from plot 1 were significantly higher than those from plot 2. Amount of overland flow increased linearly with increasing precipitation for storms with <50 mm of total rainfall (Fig. 6). The rate of increase in overland flow tapered down marginally when total precipitation increased from 50 to

284 Table 3

T. Gomi et al. Summary of correlation analysis between runoff amounts and both rainfall and soil moisture characteristics Plot 1

Plot 2

0.74

Total precipitation

0.83 p < 0.01

Max 5-min intensity

0.08

Max 20-min intensity

0.5

Max 60-min intensity

0.64

API7

0.23

p = 0.027 p = 0.01

p = 0.07

p = 0.002

p = 0.01

p = 0.45

p = 0.009 0.2

p = 0.60

p = 0.47

0.12 p = 0.75

0.06 p = 0.65

0.17 p = 0.90

p = 0.03 0.61

0.12

0.08

0.03

p = 0.048 0.52

0.61

0.26

p = 0.38 API30

p = 0.118 0.45

0.70

p = 0.39

p < 0.01 0.05

0.12

0.58

p = 0.006 0.24

0.89 p < 0.01

0.03

p = 0.04

Plot 4

0.74 p < 0.01

p = 0.075

Initial soil moisture

Plot 3

p = 0.82

0.18 p = 0.52

0.31 p = 0.49

p = 0.23

Analysis were conducted based on data form May and September 2005. Significant correlations are shown as bold type.

50

Plot 2

Plot 1 40 30

Total storm runoff (mm)

20 10 0 50

Plot 3

Plot 4

API7 < 50 mm > 50 mm

40 30 20 10 0

0

100

200

300

400 0

100

200

300

400

Total storm precipitation (mm) Figure 6 Total storm precipitation and total overland flow. Open circles show antecedent precipitation index values <50 mm (relatively dry periods); solid circles indicate a high antecedent precipitation index (>50 mm; relatively wet conditions). Dashed lines indicate 10% values for runoff coefficients.

180 mm in plots 1 and 2 (Fig. 6). It is likely that more infiltration occurred and the groundwater table rose, especially in plot 2, during storms with precipitation ranging from 50 to 180 mm. The total runoff in plot 2 surpassed that in plot 1 when total precipitation exceeded 180 mm. Overland flow responded very quickly to precipitation inputs for all plots (Figs. 7 and 8). The threshold cumulative

precipitation required to initiate overland flow was <5 mm in plots 1, 2, and 3, whereas the threshold precipitation was nearly 5 mm in plot 4. Overland flow through the litter layer requires relatively low initial precipitation to produce quickflow response (<10 mm; Buttle and Turcotte, 1999). The threshold precipitation tended to be higher in plots 1, 2, and 3 during storm events with high API7 than storm

Characteristics of overland flow generation on steep forested hillslopes of central Japan

Threshold precipitation (mm)

10

8

6

4

2

0 Plot 1

Plot 2

Plot 3

Plot 4

Runoff coefficient (/5min)

Storm runoff (mm/5min)

Precipitation (mm/5min)

Figure 7 Threshold precipitation to initiate overland flow in plots with different antecedent precipitation indices. White boxes indicate threshold precipitation for dry conditions (API7: <50 mm); shaded boxes indicate threshold precipitation for wet conditions (API7: >50 mm). Lines within and adjacent to the box plots indicate the mean and standard deviation, respectively.

0

A: September 28-29, 2004

285

events with low API7. Although our measurement interval for discharge was 5 min, there was no lag time between peak rainfall inputs and peak discharges from plots 1, 2, and 4 during a storm event in December 2004 (Fig. 8b). Runoff recession curves were also very steep after rainfall ceased. This runoff pattern, which occurred for all storms <180 mm, is representative of the predominant occurrence of infiltration excess overland flow. The occurrence of overland flow on hillslopes was associated with differences in runoff and infiltration mechanisms on mineral soil surfaces. Overland flow that occurred from all plots, but especially plots 3 and 4, may have been partly associated with flow above the mineral soil in the thin organic horizon and surface litter (biomat flow; Sidle et al., 2007). Soil hydrophobicity in the upper 5–15 cm of forest soils may also affect the occurrence of overland flow, even in deciduous forests that have relatively shallow organic horizons (1–3 cm; Fig. 4; Barrett and Slaymaker, 1989; Imerson et al., 1992; Buttle and Turcotte, 1999; Ellerbrock et al., 2005; Miyata et al., 2007). Because the organic horizon in plot 1 was thin compared to other plots, the lower roughness associated with the lack of litter in plot 1 contributed to the highest overland flow observed amongst all plots. The low volume of overland flow captured in plots 3 and 4 may be attributed to greater infiltration into

B: December 4-5, 2004

5 Total precipitaion:345.6mm Max 1 hr precipitation: 26.6mm API7 : 9.8mm: 56.6

Total precipitation:122.6mm Max 1 hour precipitation: 26.8mm API7 : 9.8mm

10 0.6 Plot 1 Plot 2

Plot 1 Plot 2 Plot 4

0.4

0.2

0.0 0.4 0.88

0.3

0.49

0.43

0.2 0.1 0.0 09/28 19:00 09/28 23:00 09/29 03:00

09/29 07:00 09/29 11:00 09/29 15:00

12/04 12:00

12/04 16:00

12/04 20:00

12/05 00:00

12/05 04:00

Figure 8 Overland flow responses and changes in runoff coefficients (5-min sampling interval) during storms on 28–29 September and 4–5 December 2004. The storm event on 28–29 September 2004 represented relatively wet conditions (high API7), whereas the storm event on 28–29 December 2004 represented relatively dry conditions. Storm runoff volumes and coefficients were highest in plot 1 during the December event; storm runoff volumes and coefficients were highest in plot 2 during the September event.

286

T. Gomi et al.

Storm runoff (mm/5min)

Precipitation (mm/5min)

September 4-7, 2005 0

2

4

6

Total precipitaion:195.2 mm Max 1 hr precipitation: 19.2mm API7 : 17.6 mm

a

0.6

b

1.11

0.78

0.4

0.2

Soil water table (m)

0.0 0.4

Influence of hydrophobicity and soil moisture c

0.3

0.2

0.1

Fig. 9). Although the groundwater table did not reach the soil surface at the well site (0.6 m; Fig. 10), it approached the surface and may have instigated return flow or even saturated overland flow in shallower soil profiles near the bottom of plot 2, thus delaying peak runoff. The maximum groundwater tables measured in wells along plot 2 were highest near the bottom of the plot and gradually decreased upslope during this event (Fig. 10). The higher groundwater table downslope persisted for only a few hours and decreased after rainfall stopped (Fig. 9). Differences in runoff mechanisms among plots appear to be associated with plot position within the catchments. The distance from the valley bottom to the gutter in plot 1 was 12 m, whereas the gutter in plot 2 was about 3 m from the valley bottom. These differences in hillslope position and soil depth may result in greater soil moisture accretion and return flow in plot 2 than in plot 1 (Tanaka et al., 1988; Tsukamoto and Ohta, 1988; Sidle et al., 2000).

Plot 1 Plot 2

0.0 09/06 12:00 09/06 18:00 09/07 0:00 09/07 6:00 Note: Storm event started on September 4, 2005

Figure 9 Response of overland flow and soil water tables in wells during a typhoon period from 6 to 7 September 2005. The water table developed more significantly in plot 2 than in other plots because of the sequence of rainfall. Arrows show periods when runoff was greater in plot 2 than in plot 1.

preferential flow pathways (e.g., pockets of decomposed organic matter, soil cracks) when runoff occurred in the shallow subsurface layer. Organic horizons and roughness elements (branches and exposed roots) of the litter layer slow flow velocity and promote greater water storage in the litter layer and infiltration into mineral soils (Abraham et al., 1994). A different runoff pattern was observed in plot 2 during the latter portions of large storms (>180 mm precipitation). During the 28–29 September 2004 storm, the peak discharge and total runoff volume from plot 2 were higher than from plot 1 and were delayed 20–30 min after the peak in rainfall (Fig. 8a). The duration of runoff recession (time from peak runoff to end of runoff) in plot 2 was approximately 1–1.5 h, which was slower than in plot 1 (10– 20 min). Delayed peak flow, longer recession, and larger runoff were also observed in plot 2 compared to plot 1 during the final burst of the large typhoon on 7 September 2005 (Fig. 9). A groundwater table developed in plot 2 during the 7 September event (>180 mm in total precipitation) and likely contributed to the delayed peak runoff from this plot, especially during the later period of the storm (arrows in

The runoff coefficients of overland flow varied seasonally and were related to changes in soil moisture (Fig. 5d and e). Storm runoff coefficients increased marginally throughout the relatively dry period from early spring to early summer in 2005 when API7 consistently decreased and soil moisture was lower (Table 2, Fig. 5). Seasonal changes in runoff coefficients were more pronounced on hillslopes with sparse understory vegetation (plot 1) than on fern-covered hillslopes (plot 2). Similar seasonal patterns were also reported by Miyata et al. (2007) in a small-plot (1 · 2 m) study. Changes in runoff coefficients throughout seasons may be partly related to the development and depletion of soil water repellency. The increases in runoff coefficients observed from April to June 2005 (Fig. 4) may be related to the reestablishment of soil water repellency under low soil moisture (<22%, Table 2). Although the hydrophobicity in plot 1 was relatively smaller than in plots 2 and 3 (Fig. 4), seasonal changes in hydrophobicity may be greater on hillslopes that have sparse understory vegetation because of greater ranges of soil moisture. Thus, changes in runoff coefficients were much greater in plot 1, possibly because of sparse vegetation cover and excessive dryness at the beginning of storm events. Progressive decreases in runoff coefficients (calculated at 5-min intervals) were also noted during a single storm on 4–5 December 2004, despite increases in rainfall intensity (Fig. 8). Higher runoff in the earlier part of the precipitation event may be associated with soil water repellency. Buttle and Turcotte (1999) showed that flow over and through organic matter occurred under dry conditions associated with repellency. Scherrer et al. (2007) noted that initial hydrophobicity produced temporary Hortonian overland flow and some macropore flow. It is likely that soil suction in combination with a dry, dense, and fine root mat in the topsoil initially prevented infiltration, especially during the early portion of storms. Similar hydrophobicity and overland flow generation in the early stage of runoff were also reported by Burch et al. (1989) and Ellerbrock et al. (2005). For small to moderate-sized storms (<180 mm total rainfall) preceded by dry conditions, the maximum runoff coef-

Characteristics of overland flow generation on steep forested hillslopes of central Japan 10 m

Plot 1

287

Plot 2

8

Soil surface

6

4 Bedrock 2

Box weir

Box w eir

Maximum Soil water tab le

0

0

2

4

6

8

10 m

0

2

4

6

8

10

Figure 10 Profile of the maximum water tables that developed during the 4–7 September 2005 storm based on crest gauges in wells along plot transects. The mean highest water table in plot 1 was 0.05 m and it was discontinuous. The mean highest water table in plot 2 was 0.18 m; the highest water table occurred near the bottom of the plot (0.41 m).

30

Plot 2

Plot 1

Storm runoff coefficient (mm)

20

10

0 0 10

100

200

300

400

0

100

200

300

Total precipitation

Plot 3

400

Plot 4

< 50mm 50 to 180 mm > 180 mm 5

0

0

20

40

60

80

100

0

20

40

60

80

100

API7 (mm) Figure 11 Storm runoff coefficients plotted against the antecedent precipitation index (API7) for small (total precipitation <50 mm), moderate (50–180 mm), and large (>180 mm) storm events.

ficients tended to be higher compared to wetter antecedent conditions (Fig. 11), although the coefficients were highly

variable for dry conditions (Figs. 4 and 10). Because more storms with dry antecedent conditions were sampled than

288 storms with wet conditions, inferences related to the possible effects of API7 on storm runoff coefficients and runoff volumes must be interpreted carefully. Similar rainfall inputs were required to initiate storm runoff from plots 1 and 2; the ranges of these values were lower compared to those for plots 3 and 4 (Fig. 11). It might be expected that soils with greater water storage capacity in the organic-rich layer and greater surface roughness (i.e., plots 3 and 4) require more rainfall to initiate overland flow.

Roles of forest floor vegetation in overland flow generation Our comparison among different conditions of groundcover in the plots highlighted the role of forest floor vegetation on overland flow generation on temperate forested hillslopes. Despite the presence of a litter layer and forest floor vegetation, overland flow occurred above the mineral horizon in the surface litter layer partly because of soil hydrophobicity. Differences in the amount of litter appeared to affect the potential for water storage within the layer and promoted infiltration at the hillslope scale. Surface roughness slowed flow through the surficial litter. The higher precipitation threshold for overland flow generation in plots 3 and 4 (Fig. 7) suggests that both water storage capacity and roughness contributed to the delayed runoff response in plots with high ground cover. Water storage capacity and infiltration from the litter layer to mineral soil tended to be higher during wet than dry conditions because a greater precipitation threshold is required to initiate overland flow (Fig. 7). Such water storage capacity in the litter layer prevents changes in the soil surface moisture, which affects the development of soil water repellency. Our finding of a positive correlation between rainfall intensity and the amount of overland flow (Table 3) suggests that short and intense storms produce more overland flow despite differences in vegetation cover. The amount of overland flow was controlled by both excess precipitation subject to the water storage potential of litter and the infiltration capacity of the soil. Surface runoff response at the hillslope scale can be expressed in the following scenario based on our field investigation. During early portions of storms, the amount of overland flow generation is affected by antecedent moisture conditions and soil water repellency. Overland flow occurs with respect to precipitation inputs. The amount of overland flow generation depends on the water storage and roughness of the surface litter layer. During the later periods of storms, the control of water repellency on overland flow generation disappears. Vertical infiltration associated with roots and macropores becomes more pronounced in this phase. More infiltration potentially occurs on hillslope surfaces that have higher litter and vegetation ground cover. During large and intense storms, which produce high water potential at the surface, overland flow is produced. When rainfall continued, saturation-excess overland flow (return flow) occurs rather than overland flow. Hillslope runoff including the occurrence of return flow is controlled by soil depth and hillslope position within a catchment (Sidle et al., 2000). The effect of vegetation cover on the runoff generation of hillslopes may be minimal.

T. Gomi et al. The estimation of infiltration, overland flow generation, and resultant contributions to stream discharge is best conducted at the forested hillslope scale because soil surface conditions and vegetation ground cover vary spatially (Julien and Moglen, 1990). Overland flow generation and transfer associated with infiltration capacity vary spatially, and this has implications for studies with different sizes of plots and types of vegetation ground cover. Runoff coefficients in small plots tend to be greater than in large plots in Japanese cypress hillslopes (Gomi et al., 2008). Miyata et al. (2007) showed that runoff coefficients from a small plot (1 · 2 m) in the same Japanese cypress stands with sparse understory and with fern understory vegetation were >40% during most storms. These contrasting results between small and large plots suggest that overland flow indeed occurred on small portions of the hillslope, but the flow was more or less discontinuous over longer slopes. Such discontinuities may be associated with vertical preferential infiltration (via macropores) caused by vegetation cover and root networks (Julien and Moglen, 1990; Sidle et al., 2000; Kobayashi and Shimizu, 2007; Gomi et al., 2008). In addition to the effects of soil surface hydrophobicity on overland flow, the hydrophobic lining of macropore walls may play a role in reducing lateral water abstraction from macropores to the surrounding soil matrix (Buttle and Turcotte, 1999; Kobayashi and Shimizu, 2007). Because overland flow generation and routing on forest hillslopes are complex and heterogeneous, observations at the hillslope scale provide more realistic estimates of water fluxes via overland flow generation/infiltration relationships than do measurements from small-scale plots. These hillslope-scale observations aid in understanding catchment-scale linkages between hillslopes and channels that can be used to advance runoff modeling.

Summary and conclusions We investigated overland flow generation under different forest conditions, including stand density, understory vegetation, overstory species, and management history, during 2004 and 2005. Although our hillslope-scale runoff plots could not be satisfactorily replicated in the various catchments because of their size, the selected hillslopes were representative of vegetation types in the study catchments and the adjacent area. Our main findings are: (1) despite various forest cover conditions, overland flow was observed in all hillslope-scale plots; (2) the amount of overland flow was greatest on hillslopes with Japanese cypress stands containing sparse understory vegetation (except for a few large storms); (3) soil water repellency contributed to localized overland flow on hillslopes; (4) despite changes in runoff coefficients potentially associated with soil water repellency, dominant seasonal changes in runoff at the hillslope scale were associated with a shift from overland flow to return flow; and (5) soil ground cover conditions associated with understory vegetation affect overland flow generation and flow through the organic layer above the mineral soil. Japanese cypress stands that have sparse understory vegetation are common in Japan. Our findings suggest that localized overland flow can occur during storms in these forests. The effect of such rapid pathways on catchmentscale peak flow has not yet been established. Overland flow

Characteristics of overland flow generation on steep forested hillslopes of central Japan transfer from hillslopes to channels depends on various factors, including the timing and size of rainfall events, soil physical and chemical conditions, and vegetation cover. Monitoring at the hillslope scale facilitates the understanding of linkages between hillslopes and streams related to overland flow and material dynamics (e.g., sediment and nutrient transport; Gomi et al., 2002; Sidle, 2006). For management concerns, partial cutting and commercial thinning support understory vegetation growth by allowing light to penetrate the forest canopy. These practices may be helpful in reducing overland flow and surface erosion providing that excessive soil disturbance and compaction does not occur. Further comparisons using different sizes of plots are necessary to understand scaling effects and the continuity of overland flow paths at the hillslope scale. Such approaches will support the development of improved numerical and process-based models for hillslope hydrology.

Acknowledgements This study was supported by the CREST project of the Japan Science and Technology Agency. Additional funding was provided through a JSPS grant (#16380102) provided to RCS. We are grateful to Yuichi Onda, University of Tsukuba, for his support and comments on the study. We thank Tewodros A. Taddesse, Sohei Kobayashi, Aurelian C. Trandafir, Fumitoshi Imaizumi, Mika Yamao, and Yotaro Nishi for their assistance with fieldwork. Appreciation is extended to Alan D. Ziegler and two anonymous reviewers for helpful and insightful comments on an earlier draft of the manuscript.

References Abraham, A.D., Parsons, A.J., Wainwright, J., 1994. Resistance to overland flow on semiarid grassland and shrubland hillslopes, Walnut Gulch, southern Arizona. Journal of Hydrology 156, 431– 446. Barrett, G., Slaymaker, O., 1989. Identification, characterization and hydrological implication of water repellency in mountain soils, southern British Columbia. Catena 16, 477–489. Baudoux, A., Witzig, J., Germann, P.F., Kienholz, H., Luscher, P., Weingartner, R., Hegg, C., 2006. Investigations on the runoff generation at the profile and plot scales, Swiss Emmental. Hydrological Processes 20, 377–394. Burch, G.J., Moore, I.D., Burns, J., 1989. Soil hydrophobic effects on infiltration and catchment runoff. Hydrological Processes 3, 211–222. Buttle, J.M., Turcotte, D.S., 1999. Runoff processes on a forested slope on the Canadian Shield. Nordic Hydrology 30, 1–20. Cannon, S.H., Reneau, S.L., 2000. Condition for generation of firerelated debris flows, Capulin Canyon, New Mexico. Earth Surface Processes and Landforms 25, 1103–1121. Cerden, O., Bissonnais, Y., Govers, G., Lecomte, V., Oost, K., Countourier, A., King, C., Dubreuil, N., 2004. Scale effect on runoff from experimental plots to catchment in agricultural areas in Normandy. Journal of Hydrology 299, 4–14. de Gruijter, J.J., Brus, D.J., Bierkens, M.F.P., Knotters, M., 2006. Sampling for Natural Resource Monitoring. Springer, Heidelberg. de Jonge, L.W., Jacobsen, O.H., Moldrup, P., 1999. Soil water repellency: effects of water content, temperature, and particle size. Soil Science Society of America Journal 63, 437–442.

289

Dunne, T., Black, R.D., 1970. Partial area contributions to storm runoff in a small New England watershed. Water Resources Research 6, 1296–1311. Dunne, T., Xhang, W., Aubry, B.F., 1991. Effects of rainfall, vegetation, and microtopography on infiltration and runoff. Water Resources Research 27, 2271–2295. Doerr, S.H., 1998. On standardizing the ‘water drop penetration time’ and the ‘morality of an ethanol droplet’ technique to classify soil hydrophobicity: a case study using medium textured soils. Earth Surface Processes and Landform 23, 663–668. Doerr, S.H., Dekker, L.W., Rotsema, C.J., Shakesby, R.A., Bryant, R., 2002. Water repellency of soils: the influence of ambient relative humidity. Soil Science Society of America Journal 66, 401–405. Doerr, S.H., Shakesby, R.A., Walsh, P.D.R., 1998. Spatial variability of soil hydrophobicity in fire-probe eucalyptus and pine forest, Portugal. Soil Science 163, 313–324. Ellerbrock, R.H., Gerke, H.H., Bachmann, J., Goebel, M.O., 2005. Composition of organic matter fractions for explaining wettability of three forest soils. J. Am. Soc. Soil Sci. 66, 57–66. Fukuyama, T., Onda, Y., Takenaka, T., Walling, D.E., 2008. Investigating erosion rates within a Japanese cypress plantation using Cs-137 and Pb-210ex measurements. J Geophysical Research, Earth Surface 113, F02007. doi:10.1029/2006JF00065. Gascuel-Odoux, C., Cros-Cayot, S., Durand, P., 1996. Spatial variations of sheet flow and sediment on an agricultural field. Earth Surface Processes and Landforms 21, 843–851. Godsey, S., Elsenbeer, H., Stallard, R.F., 2004. Overland flow generation in two lithologically distinct rainforest catchments. Journal of Hydrology 295 (1–4), 276–290. Gomi, T., Sidle, R.C., Richardson, J.S., 2002. Understanding processes and downstream linkages of headwater systems. BioScience 52, 905–916. Gomi, T., Sidle, R.C., Miyata, S., Kosugi, K., Onda, Y., 2008. Dynamic runoff connectivity of overland flow generation on steep forested hillslopes: scale effects and runoff transfer. Water Resources Research. doi:10.1029/2007WR00589. Hewlett, J.D., Hibbert, A.R., 1967. Factors affecting the response of small watershed to precipitation in humid areas. In: Sopper, W.E., Lull, H.W. (Eds.), International Symposium on Forest Hydrology. Pergamon, Oxford, UK, pp. 75–290. Horton, R.E., 1933. The role of infiltration in the hydrological cycle. Transactions, American Geophysical Union 14, 460–466. Hung, C., Gascuel-Odoux, C., Cros-Cayot, S., 2001. Hillslope topography and hydrologic effects on overland flow and erosion. Catena 46, 177–188. Imerson, A.C., Verstraten, J.M., van Mulligen, E.J., Sevink, J., 1992. The effects of fire and water repellency in infiltration and runoff under Mediterranean type forest. Catena 19, 345– 361. Joel, A., Nessing, I., Seguel, O., Casanova, M., 2002. Measurement of surface water runoff from plots of two different sizes. Hydrological Processes 16, 1467–1478. Julien, P.Y., Moglen, G.E., 1990. Similarity and length scale for spatially varied overland flow. Water Resources Research 26 (8), 1819–1832. Kobayashi, M., Shimizu, T., 2007. Restricted increases of water storage during storm evens owing to soil water repellency in a Japanese cypress plantation. Hydrological Processes 21, 2356– 2364. Miyata, S., Kosugi, K., Gomi, T., Onda, Y., Mizuyama, T., 2007. Occurrence of surface runoff affected by soil water repellency at a hillslope covered with Japanese cypress (Chamaecyparis obtusa). Hydrological Processes 21, 2365–2376. Miura, S., Hirai, K., Yamada, T., 2002. Transport rates of surface materials steep forested slopes induced by raindrop splash erosion. Journal of Forest Research 7, 201–211.

290 Nanko, K., Mizugaki, S., Onda, Y., 2008. Estimation of soil splash detachment rates on the forest floor of an unmanaged Japanese cypress plantation based on field measurements of throughfall drop sizes and velocities. Catena 72, 348–361. Onda, Y., Yukawa, N., 1994. The influence of understories and litter layer on the infiltration of forested hillslopes. In: Proceedings of the International Symposium of Forest Hydrology, pp. 107–114. Reynolds, W.D., Elrick, D.E., Youngs, E.G., Booltink, H.W.G., Bouma, J., 2002. Saturated and field saturated water flow parameters: laboratory methods. In: Methods of soil Analysis Part 4. SSSA Book series 5. Soil Science Society of America, Madison, pp. 88–690. Scherrer, S., Naef, F., Faeh, A.O., Cordery, I., 2007. Formation of runoff at the hillslope scale during intense precipitation. Hydrology and Earth System Science 11, 907–922. Sharma, M.L., Gander, G.A., Hunt, C.G., 1980. Spatial variability of infiltration in a watershed. Journal of Hydrology 45, 101–122. Shakesby, R.A., Coelho, C.D.A., Ferreira, A.D., Terry, J.P., Walsh, R.P.D., 1993. Wildfire impacts on soil-erosion and hydrology in wet Mediterranean forest, Portugal. International Journal of Wildland Fire 3 (2), 95–110. Sidle, R.C., 2006. Field observation and process understanding in hydrology: essential components in scaling. Hydrological Processes 20, 1439–1445. Sidle, R.C., Hirano, T., Gomi, T., Terajima, T., 2007. Hortonian overland flow from Japanese forest plantations - an aberration, the real thing, or something in between. Hydrological Processes 21, 3237–3247. Sidle, R.C., Tsuboyama, Y., Noguchi, S., Hosoda, I., Fujita, M., Shimizu, T., 2000. Stormflow generation in steep forested headwaters: a linked hydrogeomorphic paradigm. Hydrological Processes 14, 369–385. Tanaka, T., Yasuhara, M., Sakai, H., Marui, A., 1988. The Hachioji experimental basin study – storm runoff processes and the

T. Gomi et al. mechanism of its generation. Journal of Hydrology 102, 139– 164. Tsukamoto, Y., 1963. Storm discharge from an experimental watershed. Journal of Japanese Society of Forestry 45, 186–190. Tsukamoto, Y., Ohta, T., 1988. Runoff process on a steep forested slope. Journal of Hydrology 102, 165–178. Uchida, T., Kosugi, K., Mizuyama, T., 1999. Runoff characteristics of pipeflow and effects of pipeflow on rainfall–runoff phenomenon in a mountainous watershed. Journal of Hydrology 222, 18–36. van de Giesen, N.C., Stomph, T.J., de Ridder, N., 2000. Scale effects of Hortonian overland flow and rainfall runoff dynamics in a west African catena landscape. Hydrological Processes 14, 165–175. Watson, C.L., Letey, J., 1970. Indices for characterizing soil-water repellency based upon contact angle-surface tension relationships. Soil Science Society of America Proceedings 34, 841–844. Williams, J., Bonell, M., 1988. The influence of scale of measurement on the spatial and temporal variability of the Phili infiltration parameters – an experimental study in an Australian savanna woodland. Journal of Hydrology 104, 33–51. Zehe, E., Elsenbeer, H., Lindenmaier, F., Schulz, K., Blo ¨schl, G., 2007. Patterns of predictability in hydrological; threshold systems. Water Resources Research 43, W07434. doi:10.1029/ 2006WR00558. Ziegler, A.D., Giambelluca, T.W., Sutherland, R.A., Vana, T.T., Nullet, M.A., 2001. Contribution of Horton overland flow to runoff on unpaved mountain roads in northern Thailand. Hydrological Processes 15, 3203–3208. Ziegler, A.D., Giambelluca, T.W., 1997. Importance of rural roads as source areas for runoff in mountainous areas of northern Thailand. Journal of Hydrology 196, 204–229.