Investigating erosion rates within a Japanese cypress plantation using Cs‐137 and Pb‐210_ex measurements

Abstract

[1] In Japan, soil erosion represents an important threat to the longer‐term sustainability of Japanese cypress plantations. Although it has been demonstrated that raindrop impact, rather than sheet or rill/gully erosion, is the dominant process in both sediment detachment and sediment transport at the point or hillslope scale in such plantations, the role of overland flow at the catchment scale has received little attention to date. To confirm the dominant erosion process operating in a Japanese cypress plantation catchment, we predicted the sediment transport capacities of both overland flow, expressed as a function of the upslope contributing area and the local slope gradient, and rainsplash, expressed as a topographic curvature, for 45 sampling points across the catchment. In addition, we compared the predicted values of sediment transport capacity and rainsplash with the erosion rates estimated using Cs‐137 and Pb‐210_ex measurements. The results indicated a close correspondence between the local transport capacity of overland flow and the local erosion rates estimated using the radionuclide measurements. In contrast, there was no significant correlation between the topographic curvature and the erosion rates. Furthermore, the average radionuclide inventories for the catchment were much smaller than the local reference inventories, indicating net loss of eroded soil and radionuclides from the catchment. These results suggest that overland flow makes a significant contribution to surface soil erosion even on forested hillslopes. In addition, the results provide a useful demonstration of the potential for using radionuclide measurements to produce independent information on long‐term erosion rates, which can be used for investigating erosion processes and validating erosion models.

1. Introduction

[2] In Japan, the Japanese cypress (Hinoki; Chamaecyparis obtusa Sieb. et Zucc.) is a major commercial tree species. However, soil erosion represents an important threat to the longer‐term sustainability of Japanese cypress plantations, particularly those located in mountainous regions [e.g., Kawana et al., 1963; Miura et al., 2002]. Without thinning of the high‐density stands, canopy closure prevents the development of understory vegetation and the soil remains bare. The leaves of the Japanese cypress are scaly and readily disintegrate after falling to the ground. They are then readily dispersed and little leaf litter accumulates on the forest floor [Hattori et al., 1992]. Because of a minimal surface cover, the structure of the surface soil is rapidly destroyed by raindrop impact, resulting in the formation of a structural crust. Where a crust is formed, the infiltration rate decreases and overland flow occurs more frequently [Yukawa and Onda, 1995]. The occurrence of surface runoff increases the likelihood of soil erosion within the cypress plantations. This situation is in strong contrast to that found more generally on undisturbed forested slopes in Japan, where the soils are characterized by a substantial litter layer, a well‐developed structure and relatively large pores. Because of the relatively high infiltration rates and the protection afforded by the litter layer, overland flow rarely occurs on these slopes [Tani and Ohta, 1992].

[3] Several previous studies have reported the occurrence of overland flow on forest slopes within Japanese cypress plantations [e.g., Hattori et al., 1992; Nishiyama, 2003; Gomi et al., 2005]. Inoue et al. [1987] and Tsukamoto [1989] reported that large volumes of eroded sediment could be associated with rilling during heavy rainfall events. However, most previous studies have concluded that raindrop impact, rather than sheet erosion or rill/gully erosion, is the dominant driver of both sediment detachment and sediment transfer at the point or slope scale [e.g., Miura et al., 2002]. The potential contribution of overland flow to sediment production at the catchment scale has received little attention to date.

[4] In recent years, physically based soil erosion models, based on the sediment transport capacity of overland flow, have been used to describe the potential sediment flux and also for scaling actual erosion rates [e.g., Prosser and Rustomji, 2000]. However, the prediction of surface runoff and soil erosion requires detailed information on the rainfall distribution and kinetic energy, overland flow hydraulics, and soil properties, As a result, a process‐based model may not be a realistic tool for predicting soil erosion, although it may be useful for developing an improved understanding of the erosion processes involved [Parsons et al., 1997]. A simple means of obtaining approximate estimates of soil erosion rates is needed for land use planning and developing appropriate land management strategies.

[5] Empirically based erosion prediction procedures, such as the Universal Soil Loss Equation (USLE), have been developed to estimate soil loss on the basis of soil‐erosion plot studies. However, these equations do not provide the information needed to understand the spatial distribution of erosion and the erosion rates involved. As a result, the use of fallout radionuclides as a complement to existing methods of estimating erosion rates has attracted increasing attention in recent years [see Zapata et al., 2002; Walling, 2006].

[6] The literature now contains many examples of the successful use of fallout radionuclides, particularly Cs‐137 [e.g., Ritchie and McHenry, 1990; Walling and He, 1999], Pb‐210_ex [e.g., Walling et al., 2003], and Be‐7 [e.g., Blake et al., 1999], as sediment tracers to document soil redistribution by surface erosion. These techniques are based on both the high affinity of fallout radionuclides for soil particles at the ground surface and the relatively uniform distribution of the fallout input at a local scale. On the basis of the deviation of measured fallout radionuclide inventories at the sampling point from the local reference inventory, detailed information on both the magnitude and the spatial distribution of soil mobilization rates can be obtained [Zapata et al., 2002].

[7] In a recent study, Porto et al. [2001] provided an empirical validation of the estimates of the soil erosion rates obtained for a small (1.38 ha) uncultivated catchment in southern Italy using Cs‐137 measurements and fallout radionuclide measurements have also been successfully used as an independent data source for validating erosion models. For example, Warren et al. [2005] validated the unit stream power erosion and deposition (USPED) model and the USLE, using Cs‐137‐derived erosion/deposition measurements; He and Walling [2003] used the spatial patterns of both the Cs‐137 inventories and the resulting estimates of soil redistribution rates to test four spatially distributed soil erosion and sediment delivery models, Ferro [1998] validated the application of the sediment delivery distributed (SEDD) model based on the revised Universal Soil Loss Equation (RUSLE) to a small Sicilian basin by comparing the model‐predicted sediment yield with the Cs‐137 loss relative to the local reference inventory; and Chappell [1996] derived the stream power indices to provide a surrogate measure of water erosion and compared them with the measured reduction in the Cs‐137 inventory, relative to the reference inventory, to determine the effect of topographic parameters on soil redistribution. These studies demonstrate that the Cs‐137 technique is being increasingly used as both an alternative and complement to existing methods of erosion assessment, such as modeling and direct measurement.

[8] The current inventory of a fallout radionuclide in the soil will be influenced by soil redistribution and thus by topographic factors (i.e., slope gradient, slope length, and upslope area), ground cover, and previous disturbance. The current inventory therefore has the potential to provide a retrospective assessment of antecedent erosion processes. In this context, fallout radionuclides have also been used as a means of establishing the dominant erosion processes, for example by assessing the strength of the relationship between the Cs‐137 inventory and slope gradient [e.g., Fulajtar, 2003; Kashiwaya, 2003; Nearing et al., 2005; Kaste et al., 2006]. The relationship between the current radionuclide inventory and the physical factors that contribute to erosion (i.e., topographic curvature, slope length, upslope area, and shear stress of overland flow) merits further investigation, particularly in terms of establishing the importance of different erosion processes. Recent advances in the availability of detailed digital elevation models (DEM) derived using airborne laser devices could provide an important data source in such analysis.

[9] In this study, fallout radionuclide measurements were used as a means of confirming the dominant erosion process operating in a representative Japanese cypress plantation catchment. Within the study, the spatial patterns associated with the sediment transport capacities of both overland flow, expressed as a function of the upslope contributing area and local slope, and rainsplash, expressed as a function of topographic curvature, were predicted using a detailed DEM, and the resulting patterns were compared with the spatial patterns of erosion rates established using both Cs‐137 and Pb‐210_ex measurements on a network of bulk cores collected within the catchment.

2. Materials and Methods

2.1. Study Area

[10] The study focused on a small headwater catchment dominated by Japanese cypress plantations, located near the town of Taiki, in Mie prefecture, central Japan (136°25′E, 34°21′N) (Figure 1). The catchment has an area of 0.36 ha; its altitude ranges from 146 to 222 m and its slopes are steep with an average angle of 42.9°. The climate is typically humid subtropical, with a mean annual rainfall of 2069 mm at the nearest meteorological station (Kayumi, Matsusaka city, Mie prefecture). The soil type is a brown forest soil formed in situ from crystalline schist of the Sambagawa metamorphic belt. Japanese cypress trees were planted in the early 1960s after clear‐cutting of the preexisting forest. The stand density is 4000 stems ha⁻¹ and the canopy is now closed. As a result the understory vegetation is very sparse. In this catchment, a large number of landslide scars have been reported (T. A. Taddese, personal communication, 2006). There are at least two landslide scars on the left‐side hillslope in the study catchment (Figure 2).

Figure 1

Location and topographic map of study catchment. The a and b locations represent landslide scars shown in Figure 2.

Figure 2

Landslide scars in the catchment.

[11] To date, most studies involving the use of fallout radionuclides to estimate soil redistribution rates have been undertaken in agricultural areas, where the land use involves cultivation, grassland or rangeland. Forest areas are commonly viewed as introducing complications into such studies, by virtue of, first, the increased small‐scale spatial variability of fallout inputs caused by canopy interception and the concentration of rainfall inputs by stemflow and, secondly, the effects of litter in preventing the rapid adsorption of the radionuclide input by the surface soil. However, these potential complications were judged to be of limited importance in the study catchment. The clear‐cutting and replanting of the forest cover in the early 1960s means that much of the Cs‐137 input associated with bomb fallout occurred while the trees were immature and with a limited canopy, with the result that the potential influence of canopy interception and stemflow in increasing the spatial heterogeneity of the fallout input will have been greatly reduced. Equally, much of the Pb‐210_ex fallout input will again have occurred while the forest canopy was incomplete. Thus, although the effects of the forest canopy in increasing the spatial heterogeneity of fallout inputs cannot be discounted, they are judged to be much reduced in importance. Furthermore, the results presented by Wallbrink and Murray [1996] for Be‐7 inputs to areas under different land cover, including forest, in Australia, demonstrated that surface cover did not exert a significant influence on the local spatial variability of Be‐7 inventories, suggesting that the importance of a forest cover in increasing the spatial variability of fallout inputs may be overestimated. The absence of a significant litter cover under the Japanese cypress forest, and the existence of large areas of bare soil also mean that the potential effects of a litter cover in reducing direct adsorption of the fallout by the surface soil are again minimized. Fukuyama et al. [2005] reported that Cs‐137 loss with litter represented up to only 2.7% of the total erosional loss of Cs‐137 from Japanese cypress plantations.

2.2. Sample Collection and Analysis Procedure

[12] To establish the local reference inventories for Cs‐137 and Pb‐210_ex, a scraper plate (450 cm²) was used to collect samples from a depth incremental profile, located on the ridge top of the study catchment, to a depth of 30 cm (0.5‐, 1‐, and 2‐cm increments). To determine the spatial distribution of Cs‐137 and Pb‐210_ex inventories across the study catchment, 45 bulk soil cores were collected from representative locations with a spacing of ∼5–10 m in each direction. These cores were obtained using a steel core sampler (5.5 cm i.d.) inserted to a depth of 30 cm. The soil samples were transported to the Environmental Modeling and Creation Laboratory of the Department of Integrative Environmental Sciences, within the Graduate School of Life and Environmental Sciences of the University of Tsukuba, for processing and analysis. The Environmental Modeling and Creation Laboratory has been certificated for the determination of Cs‐137 and Pb‐210 in soil through participation in an IAEA Co‐ordinated Research Project Proficiency Test (IAEA CRP DI.50.08, 2005–2006). The samples were air‐dried for a few days before being oven‐dried at 110°C for 24 h and then disaggregated and passed through a 2‐mm sieve. A representative fraction of each sample (<2 mm) was then placed into a U‐8 plastic container (100 mL) for determination of its Cs‐137 and Pb‐210_ex activity.

[13] The activities in all the samples were determined directly by gamma spectrometry using an n‐type coaxial high‐purity Ge‐detector (Eurisys Mesures, EGC30‐200‐R) with an efficiency of 30% and a full width at half maximum (FWHM) of 1.85 keV for Co‐60 at 1332 keV and a multichannel analyzer. The gamma emission peaks at 46.5, 351, and 661 keV for Pb‐210, Pb‐214, and Cs‐137, respectively were used for the measurements. Ra‐226, which provides the supported Pb‐210, was measured indirectly, using its daughter radionuclide Pb‐214, after sealing and storage for three weeks, in order to allow radioactive equilibration between Ra‐226 and Rn‐222. The count time was typically ∼43,200 s, yielding results with an analytical precision of < ∼10% at the 95% confidence level. The system was calibrated and the absolute full‐energy peak efficiency was determined using a Cs‐137 and Pb‐210 spiked soil of known activity, provided by the IAEA within the framework of Proficiency Test IAEA‐CU‐2006‐02. To correct for self‐absorption, we determined the peak efficiency for standard volume sources with different thicknesses. The measured weight concentration of radionuclides (Bq kg⁻¹) was converted to a inventory (Bq m⁻²) using the weight of <2 mm fraction and the cross section of the sample.

2.3. Estimating Erosion and Deposition Rates From the Cs‐137 and Pb‐210_ex Inventories

[14] To estimate the long‐term water‐induced soil redistribution rates for the points within the study catchment from which the soil cores were collected for Cs‐137 and Pb‐210_ex analysis, the Diffusion and Migration conversion model proposed by Walling and He [1999] and included in the software produced by Walling et al. [2006] was employed to convert the measured Cs‐137 and Pb‐210_ex inventories to estimates of the net soil redistribution rate at the sampling points. This research tool has been developed for converting Cs‐137, Pb‐210_ex, and Be‐7 inventories to estimates of soil erosion and deposition rates. The software uses VBA (Visual Basic Application) and is operated as a standard add‐in within Microsoft Excel. Models applicable to both cultivated and undisturbed (e.g., rangeland and permanent pasture) soils are included. The Diffusion and Migration conversion model considers the time‐dependent behavior of both the fallout input and its subsequent redistribution in the soil profile. The net erosion and deposition rates were calculated by comparing the Cs‐137 and Pb‐210_ex inventories of the soil cores collected within the catchment with the local Cs‐137 and Pb‐210_ex reference inventories, based on samples collected from an uneroded and undisturbed reference site on the ridge top. In the absence of direct measurement of the annual atmospheric Cs‐137 deposition, the record of annual fallout inputs to a study site can be synthesized from those recorded at other monitoring stations [Walling et al., 2002]. For example, for study sites in the northern hemisphere, the temporal distribution of annual fallout inputs can be assumed to be similar to that reported for total fallout inputs to the northern hemisphere by Cambray et al. [1989], based on data recorded at a number of stations located in that hemisphere. Representative data for the annual Cs‐137 deposition flux in the northern hemisphere is included in the software. These data are scaled to match the local reference inventory, in order to synthesize the record of annual Cs‐137 deposition flux for the study area.

2.3.1. Cs‐137‐Derived Erosion/Deposition Rates

[15] Redistribution of Cs‐137 within the soil profile represents the result of a complex set of mechanisms including physical, physico‐chemical and biological processes. Their net effect is represented in the conversion model using a one‐dimensional transport model characterized by an effective diffusion coefficient D (kg² m⁻⁴ a⁻¹) and a migration rate V (kg m⁻² a⁻¹) for the soil profile [He and Walling, 1997]. The two lumped parameters D and V reflect all redistribution processes, including physico‐chemical processes involving the adsorption and desorption of Cs‐137 by soil particles. The variation in the Cs‐137 inventory C_u(t) (Bq kg⁻¹) for the surface soil with time t (year) can be represented as:

where

t

time elapsed since the Cs‐137 deposition started, years;

I(t)

annual Cs‐137 deposition flux, Bq m⁻² a⁻¹;

H

relaxation mass depth of the initial distribution where the activity decreases to 1/e that of the surface activity, kg m⁻²;

R

erosion rate, kg m⁻² a⁻¹;

λ

decay constant, a⁻¹;

D

diffusion coefficient, kg² m⁻⁴ a⁻¹;

V

downward migration rate of Cs‐137 in the soil profile, kg m⁻² a⁻¹.

[16] The parameter values D and C were determined from the vertical distribution of Cs‐137 in the soil profile at the uneroded ridge top within the study catchment. Although more precise values of these two parameters can be obtained through solving the one‐dimensional transport equation [cf. He and Walling, 1997], they may be approximated using the following equations:

where:

t

the year when the soil core was collected (years);

W_p

mass depth of the maximum Cs‐137 concentration (kg m⁻²);

N_p

distance between the depth of the maximum Cs‐137 concentration and the depth where the Cs‐137 concentration reduces to 1/e of the maximum concentration (kg m⁻²).

[17] Values of D and V of 5.4 and 0.14 were obtained for the study site. The relaxation depth H for the initial distribution of fallout Cs‐137 in surface soil is defined as the mass depth (kg m⁻²) at which the Cs‐137 concentration reduces to 1/e of the surface concentration. The value can be determined experimentally by application of Cs solution using a rainfall simulator. Although the value could be expected to vary according to soil properties, we adopted the reported value of H of 4.0 kg m⁻² for soils in Devon, United Kingdom [He and Walling, 1997].

[18] Erosion rates were calculated from the reduction in the Cs‐137 inventory A_ls (t) (Bq m⁻²) at the sites where the measured Cs‐137 inventory was less than the reference inventory. A_ls (t) is defined as the local reference Cs‐137 inventory A_ref less the measured total Cs‐137 inventory A_u. The Cs‐137 inventory in the surface soil, C_u (t′), was calculated on the basis of equation (1), and A_ls (t) was calculated as

where P is a particle size correction factor to take account of differences between the grain size composition of the mobilized sediment and the original soil. Net deposition rates (R′) were calculated as

where A_ex(t) (Bq m⁻²) is the increase in the Cs‐137 inventory at the sampling points where the measured total Cs‐137 inventory was greater than the reference inventory, and C_d(t′) is the weighted average of the Cs‐137 concentrations of the sediment generated from the upslope contribution area S (m²). C_d (t′) can be calculated from

2.3.2. Pb‐210_ex‐Derived Erosion/Deposition Rates

[19] A modified version of the Diffusion and Migration conversion model developed for Cs‐137 measurements and described above was employed to derive estimates of the longer‐term water‐induced soil redistribution rates from the information on measured Pb‐210_ex inventories at the sampling points [see Walling et al., 2003]. This model assumes a constant annual Pb‐210_ex fallout flux and takes into account postdepositional redistribution processes and their influence on the Pb‐210_ex depth distribution.

[20] A diffusion coefficient D_Pb (kg² m⁻⁴ a⁻¹) and a migration rate V_Pb (kg m⁻² a⁻¹) are used to represent the net effect of the slow vertical redistribution of unsupported Pb‐210_ex by physicochemical and biological processes. These parameters account for the evolution of the shape of the unsupported Pb‐210_ex depth profile over time. Assuming a constant deposition flux of Pb‐210_ex over time, a stable Pb‐210_ex inventory will be achieved in an undisturbed soil profile, with the fallout input being balanced by radioactive decay. Taking into account the radionuclide decay effect, the variation of the Pb‐210_ex concentration C_{u, Pb} (z) (Bq kg⁻¹) of the soil with cumulative mass depth z in the soil profile can be represented by the following equation:

where C_{u, Pb} (0) (mBq g⁻¹) is the Pb‐210_ex concentration in surface soil, and the constant β (g⁻¹ cm²) is defined as:

[21] Equation (7) can be used to simulate the Pb‐210_ex profile in stable undisturbed soils. The concentration of Pb‐210_ex in the surface soil remains constant [He and Walling, 1997]. Routines are available within the software for estimating β, D_Pb, and V_Pb for reference sites where values of depth incremental sample mass and corresponding values of mass concentration are available.

[22] At a location that experiences water‐induced erosion, the rate of soil loss R (kg m⁻² a⁻¹) can be estimated from the reduction in the Pb‐210_ex inventory (A_{u, ls} (t)) when compared with the local reference value along with information on the Pb‐210_ex content of the surface soil (C_u (t′)), namely,

Alternatively, at a location that experiences water‐induced deposition, the rate of sedimentation R′ (kg m⁻² a⁻¹) can be estimated from the increase in the Pb‐210_ex inventory (A_u,ex(t)) when compared with the local reference value along with information on the Pb‐210_ex content of the deposited sediment (C_d(t′)), namely,

where C_d(t′) can be calculated as the product of:

[23] The spatial distributions of radionuclide inventories and the estimated soil redistribution rates across the study catchment were mapped using the contouring software Surfer 8 (Golden Software, Inc. Golden, Colorado, U.S.A.). The point estimates of soil redistribution rates obtained for the individual sampling points were used to calculate the mean net soil erosion rate for the study catchment.

2.4. Estimating the Sediment Transport Capacity of Overland Flow

[24] Topography exerts an important control on soil erosion through its influence on the hydraulics of overland flow. The overland flow volume increases with increasing hillslope length or upslope contributing area, thereby increasing the sediment transport capacity [Prosser and Abernethy, 1999]. The effect of topography on sediment transport has been represented using arguments based on the boundary shear stress of the flow and the stream power per unit weight of water. In this study, we calculated the sediment transport capacity of overland flow Q_f across the catchment on the basis of the local topography, to simulate the hydrologic potential for erosion. For this purpose, Q_f can be predicted from a 5‐m grid DEM as:

where k can be assumed constant for small catchments [Prosser and Abernethy, 1999] and a denotes upslope contributing area per unit width of slope. In a DEM, this can be calculated as the upslope contributing area (A) divided by the element width (a = A/b), in meters. Here c is the typical value of the empirical exponent for shallow flows over rough surfaces (c = − 0.7) [Prosser and Abernethy, 1999], and S is sine of the slope gradient.

[25] The DEM for the study catchment was generated from airborne laser scanning data obtained in May 2004. The Laser pulse frequency and the footprint diameter were 50 kHz and 0.15 m, respectively. The values of a and local gradient for the soil sampling points were calculated from the DEM using ESRI ArcMap 9.0.

[26] Determining the appropriate scale of the topographic grid is significant when calculating morphometric parameters such as curvature. Heimsath et al. [1999] explored the effects of various sizes of grids ranging from 1.5 m to 20 m on calculated curvature and reported that the curvature becomes relatively scale‐independent at grid scales greater than 5 m. In this study, we calculated upslope area and curvature using different sizes of grid (1 m, 2 m, 5 m and 10 m, respectively) and a 5 m‐grid size was finally employed in the analysis to be consistent with the spatial resolution of the erosion rates (5–10 m) derived from the radionuclide inventories.

2.5. Splash Transport

[27] Topographic curvature has been used to represent the key control on diffusion‐like downslope soil transport processes [Heimsath et al., 1999; O'Farrell et al., 2006]. Downslope sediment transport flux, which is controlled by linearly slope‐dependent processes, Q_d (l³ l⁻¹ t⁻¹), can be expressed in one dimension as:

where K_d is the constant of proportionality known as the diffusion coefficient (l² t⁻¹); z is land surface elevation, in meters; and x is downslope distance. The divergence of sediment transport flux is equivalent to the erosion, or landscape‐lowering rate, dz/dt such that differentiating equation (13) leads to:

where E is land surface lowering and ∇²z is topographic curvature.

[28] This equation shows that in areas where linearly slope‐dependent processes are dominant, land surface lowering (E) is proportional to topographic curvature [O'Farrell et al., 2006]. To test the significance of rainsplash in erosional processes, we plotted the erosion rate against topographic curvature. Furthermore, we calculated the topographic curvature from the DEM using the contouring software Surfer 8.

3. Results

3.1. Cs‐137 and Pb‐210_ex Inventories

[29] The Cs‐137 and Pb‐210_ex reference inventories obtained for the study were 2948 and 8454 Bq m⁻² (Table 1). The local Cs‐137 reference inventory is about 15% greater than the cumulative deposition decay corrected to the sampling year (2003) of 2560 Bq m⁻² reported for the local region by Aoyama et al. [2006]. To confirm the magnitude of the Pb‐210_ex reference inventory, we compared the annual Pb‐210_ex deposition flux estimated from the local reference inventory, with the annual deposition flux extrapolated from the data available for the same climatic zone. Assuming that Pb‐210_ex can be detected for five half‐lives (111 years), a value of annual Pb‐210_ex deposition of 267 Bq m⁻² a⁻¹ can be estimated from the local reference inventory (8454 Bq m⁻²). Values of annual Pb‐210_ex deposition of 200 Bq m⁻² and 135 Bq m⁻², respectively, have been reported for Tokyo and Osaka, both located on the Pacific coast of Japan, within the same climatic zone [Yamamoto et al., 2006]. The 30‐year mean annual precipitation values for Tokyo and Osaka are 1466 mm and 1306 mm, respectively. If it is assumed that the mean annual deposition flux will be proportional to the mean annual precipitation, the value of mean annual rainfall for the study area of 2069 mm can be used to estimate an annual Pb‐210_ex deposition flux of 252 Bq m⁻² a⁻¹. This value is consistent with the estimate of 267 Bq m⁻² a⁻¹ derived from the measured Pb‐210_ex inventory.

Table 1. Cs‐137 and Pb‐210_ex Inventories for the Reference Sites and for the Sampling Points

	Cs‐137		Pb‐210ex
	Reference Site^aa Soil sample was collected using scraper plate with area of 450 cm^2.	Sampling Points^bb Soil samples were collected using a steel core sampler with area of 23.7 cm^2.	Reference Site	Sampling Points
Mean, Bq m^−2	2948	1710	8453	4354
Maximum, Bq m^−2		4477		13,352
Minimum, Bq m^−2		79		0
Standard deviation, Bq m^−2		1051		2972
Coefficient of variation, %		61		68
Number of samples	1	45	1	45

• a Soil sample was collected using scraper plate with area of 450 cm².
• b Soil samples were collected using a steel core sampler with area of 23.7 cm².

[30] The Cs‐137 and Pb‐210_ex inventories measured for the individual sampling points were lower than the local reference inventories at 87% and 91% of the sampling points, respectively. Table 1 summarizes the Cs‐137 and Pb‐210_ex inventories for the study catchment. The highest Cs‐137 and Pb‐210_ex inventories were 4477 and 13,352 Bq m⁻², respectively. For the 45 samples, the mean Cs‐137 and Pb‐210_ex inventories were 1710 and 4354 Bq m⁻², with coefficient of variation (CV) values of 61.4% and 68.3%, respectively. The spatial distributions of the Cs‐137 and Pb‐210_ex inventories at the sampling points in the study catchment are shown in Figure 3. The open circles in this figure represent the sampling points. The Cs‐137 and Pb‐210_ex inventories exhibited a similar spatial distribution. On the ridge top and upper slope, both the Cs‐137 and Pb‐210_ex inventories were greater than those on the lower slope or in the vicinity of the valley bottom. On the left‐side hillslope, there were some areas with relatively high inventories.

Figure 3

Spatial patterns of Cs‐137 and Pb‐210_ex inventories.

3.2. Rates and Spatial Patterns of Soil Erosion and Deposition Derived from Cs‐137 and Pb‐210_ex Inventories

[31] The erosion or deposition rates were estimated derived using the Diffusion and Migration conversion model described previously. The parameter values used for H, D, and V were 15 kg m⁻², 10 kg² m⁻⁴ a⁻¹, and 0.14 kg m⁻² a⁻¹, respectively. The spatial distribution of the erosion and deposition rates derived from the Cs‐137 and Pb‐210_ex inventories measured at the sampling points in the study catchment are shown in Figure 4. Most of the sampling points were characterized by net soil loss. The mean, minimum, and maximum rates of soil erosion and deposition for the sampling points are shown in Table 2. The values indicated in the table are based on the erosion and deposition rates estimated for the individual sampling points and not the interpolated results. The highest erosion rates estimated from the Cs‐137 and Pb‐210_ex inventories were 4.8 and 6.8 t ha⁻¹ a⁻¹, respectively. The mean erosion rates were 2.1 and 3.3 t ha⁻¹ a⁻¹, with CV values of 85.2% and 73.2%, respectively.

Figure 4

Spatial patterns of rates of soil erosion and deposition for the sampling points derived from Cs‐137 and Pb‐210_ex inventories.

Table 2. Mean, Minimum, and Maximum Rates of Calculated Net Soil Loss for the Sampling Points

	Cs‐137	Pb‐210ex
Mean, t ha^−1 a ^−1	2.1	3.3
Maximum, t ha^−1 a ^−1	4.8	6.8
Minimum,^aa Negative values represent deposition rate. t ha^−1 a ^−1	−2.5	−3.9
Standard deviation, t ha^−1 a ^−1	1.7	2.4
Coefficient of variation, %	85.2	73.2
Number of samples	45	45

• a Negative values represent deposition rate.

3.3. Erosion Prediction

[32] We calculated the sediment transport capacity of the overland flow for individual sampling points on the basis of the local gradient and the upslope contributing area to simulate the hydrologic erosion potential. The spatial pattern of the sediment transport capacity over the study catchment is shown in Figure 5. Large values indicate areas that are more susceptible to water erosion, where the number of flow lines and the gradient within the DEM cell are large. As with the local upslope contributing area, the power law product of contributing upslope area and local gradient was relatively greater along the valley bottom and lower on the ridge (Figure 5). The spatial pattern of the calculated topographic curvature across the catchment is shown in Figure 6.

Figure 5

Spatial patterns of the power law product of upslope contributing area and local slope a^1.3/3*S^2/3. Open circles and the crosses represent sampling points. Crosses represent the sampling points excluded from analysis.

Figure 6

Spatial patterns of topographic curvature. Open circles and the crosses represent sampling points. Crosses represent the sampling points excluded from analysis.

[33] The relationships between erosion and deposition rates estimated from the radionuclide measurements and upslope area for the individual sampling points is shown in Figure 7. Erosion rates can be seen to be positively related to the upslope contributing area. The equivalent relationships between the estimated erosion and deposition rates and the calculated values of sediment transport capacity of overland flow for the individual sampling points are presented in Figure 8. Soil redistribution rates can again be seen to be positively related to the sediment transport capacity. Figure 9 presents the relationships between the soil redistribution rates estimated from the radionuclide measurements and the measure of topographic curvature. In this case there is little evidence of any relationship between the two variables.

Figure 7

Relationships between contributing upslope area and erosion rate. Open circles represent data used for regression. The crosses represent excluded values.

Figure 8

Relationships between the power law product of upslope contributing area and local slope a^1.3/3*S^2/3 and erosion rate. Open circles represent data used for regression. The crosses represent excluded values.

Figure 9

Relationships between topographic curvature and erosion rate. Open circles represent data used for regression. The crosses represent excluded values.

4. Discussion

4.1. Predicted Sediment Transport Capacity and the Soil Redistribution Rates

[34] In this study, we estimated the erosion and deposition rates for the 45 sampling points, on the basis of the measured radionuclide inventories. The sediment transport capacity of overland flow and rainsplash was also calculated for each sampling point. If a significant correlation between the erosion rates estimated from the radionuclide measurements and the predicted sediment transport capacity of the overland flow and the estimated erosion rates is found, overland flow is inferred to be the dominant process contributing to sediment production.

[35] In this catchment, a large number of landslide scars have been reported (T. A. Taddese, personal communication, 2006). There are at least two landslide scars on the left‐side hillslope in the study catchment (Figures 1 and 2). Landsliding will also induce both mass movement and subsequent sediment movement from above the landslide scar into the hollow [Dietrich et al., 1986; Shimokawa and Jitousono, 1989]. Erosional loss of radionuclides from the upper sideslope and their accumulation in the hollow can therefore be expected to occur [Onda et al., 2003]. To exclude the direct and indirect influence of landsliding on the redistribution of surface soil and radionuclides, and to focus on assessing the relative importance of overland flow and rainsplash in accounting for soil redistribution in the study area, the results obtained from the seven sampling sites located on the landslide scar, the lower side hollow, and the upper sideslope were therefore excluded from the final analysis (Figures 5 and 6). The exclusion of these seven sites as being potentially influenced by landslide activity was based on consideration of both the local topography (Figure 1) and the spatial distribution of the radionuclide inventories (Figure 4). Excluded data points have been designated by crosses on Figures 5 and 6. The following discussion focuses on the results from the remaining 38 sampling points (Figures 3, 4, 5, and 6).

[36] All of the data are shown in the plots of the relationships between the estimates of soil redistribution rates obtained from the radionuclide measurements and the measures of sediment transport capacity and contour curvature presented in Figures 7, 8 and 9, but the points representing the seven points on the left‐hand slope excluded from the analysis are distinguished by use of a different symbol. Some scatter can be seen in the relationships between the soil redistribution rates and the measures of sediment transport capacity. The scatter in the relationship can be attributed to the following potential causes. First, the sediment transport capacity at a given point is used to indicate the hydrologic potential for erosion at that point. However, the local radionuclide inventories represent the cumulative net soil redistribution, and therefore reflect both the erosion and redeposition processes. Second, although we assumed a uniform infiltration rate over the study catchment to derive the sediment transport capacity, the heterogeneity of the physical properties of the hillslope, including surface cover, micromorphology, and infiltration rate could be expected to cause spatial variability in overland flow generation and soil erosion.

[37] In the case of the erosion rates estimated from the measured Pb‐210_ex inventories, the r² value for the relationship with the sediment transport capacity is lower than that for the erosion rates estimated from the measured Cs‐137 inventories (Figure 8). This contrast can be attributed to the difference in the temporal pattern of the fallout between Cs‐137 and Pb‐210_ex. Cs‐137 fallout was produced by bomb tests and did not begin to occur until the late 1950s. Current Cs‐137 fallout is effectively zero. In contrast the annual atmospheric deposition of Pb‐210_ex (a natural fallout radionuclide) has been continuous and occurs almost at a steady state in Japan [Yamamoto et al., 2006]. The current Pb‐210_ex inventory will therefore reflect the longer‐term erosional history of the catchment, although with emphasis on recent years, whereas Cs‐137 inventories will reflect the general pattern of erosion during the post 1960s period. Further downslope, the surface material can accumulate additional Pb‐210_ex fallout by direct deposition [Wallbrink et al., 2002]. Nevertheless, a positive correlation exists between the erosion rates estimated from the measured Pb‐210_ex inventories and the sediment transport capacity and the importance of overland flow in accounting for the long‐term soil loss can thus be inferred. Kaste et al. [2006] reported a significant inverse correlation between upslope contributing area and Cs‐137 inventories for undisturbed grasslands in northeastern Kansas. Here, the sediment transport capacity is also expressed as a function of the upslope contributing area and local slope. The Cs‐137 inventory represents the cumulative net soil redistribution over the period since the onset of Cs‐137 global fallout (∼45 years). The existence of positive relationships between the erosion rates estimated from the Cs‐137 inventories and the calculated sediment transport capacities for the sampling points (Figures 7 and 8) suggests that overland flow has exerted an important influence on soil erosion in the Japanese cypress plantation catchment over the past 45 years.

4.2. Topographic Curvature and the Soil Redistribution Rates

[38] Topographic curvature is one of the key factors that control diffusion‐like downslope soil transport processes. In this study, we compared the estimates of soil redistribution rate derived from the radionuclide measurements with the corresponding values of local topographic curvature for the sampling points. The soil redistribution rates exhibit no significant correlation with the topographic curvature (Figure 9). Because the estimates of soil redistribution rate showed no significant relationship with topographic curvature, it is inferred that soil erosion induced by the overland flow, which reflects both the upslope contributing area and local gradient, dominates long‐term soil redistribution in Japanese cypress plantations.

4.3. Erosion Processes in Japanese Cypress Plantations

[39] Generally, it is known that overland flow rarely occurs on undisturbed Japanese forested slopes [Tani and Ohta, 1992], because of the relatively high infiltration rate and the protection afforded by the litter layer. It has therefore been suggested that raindrop impact, rather than sheet erosion or rill/gully erosion is the dominant driving force for both surface soil erosion and sediment transport in forested areas at the point or slope scale [e.g., Miura et al., 2002]. The occurrence of overland flow has, however, been reported in Japanese cypress plantations [e.g., Hattori et al., 1992; Nishiyama, 2003; Gomi et al., 2005]. Tsujimura et al. [2005] conducted a hydrological and tracer investigation at the present study site. They reported a coincidence of the peak discharge of overland flow and the peak rainfall intensity. In addition, they reported that the rainfall component of overland flow was dominant during heavy rainfall intensities exceeding 2 mm/5 min, by performing an end‐member mixing analysis using electrical conductivity and Cl⁻ concentration as the tracers. These results were seen as demonstrating the importance of Hortonian overland flow in a catchment covered by unmanaged Japanese cypress plantations.

[40] If diffusion‐like processes are responsible for redistributing fallout nuclides in this catchment, the losses and the gains of fallout radionuclides should tend to balance at the landscape scale, with little net loss from the system. However, if advection is the dominant sediment (and radionuclide) transport mechanism, a net loss of radionuclides might be expected. The arithmetic means of the inventories for all 45 sampling points were 1710 Bq m⁻² for Cs‐137 and 4354 Bq m⁻² for Pb‐210_ex, respectively. These values are substantially lower than the reference inventories, suggesting that advection processes have caused a net loss of radionuclides from the catchment.

[41] Our study suggests that overland flow makes a significant contribution to long‐term surface soil erosion, even on forested hillslopes. Although Miura et al. [2002] indicated that raindrop impact may contribute to surface soil erosion at a relatively small spatial scale and over short timescales, our results suggest that overland flow makes a significant contribution to surface soil redistribution over wider spatial scale and over longer timescale, even on undisturbed forested hillslopes (Figures 7 and 8). The results presented clearly demonstrate the potential for using Cs‐137 and Pb‐210_ex measurements to estimate soil redistribution rates. Such data can be used both in the validation of erosion models and in documenting erosion processes in mountainous‐forested catchments.

5. Conclusion

[42] To identify the dominant erosion process operating in a Japanese cypress plantation catchment, we predicted the sediment transport capacities of both overland flow and rainsplash. These results were compared with the erosion and deposition rates estimated from measured Cs‐137 and Pb‐210_ex inventories. Both the upslope contributing area and the magnitude of the local sediment transport capacity showed significant correlations with the erosion rates. In contrast, there was no significant relationship between the estimated erosion and deposition rates and topographic curvature. The current mean Cs‐137 and Pb‐210_ex inventories in the catchment are substantially smaller than those of the reference site. This net loss of the radionuclide input to the catchment provides further confirmation of the importance of advection processes in controlling erosion and sediment redistribution in the study catchment. Overall, these results suggest that overland flow can make a significant contribution to surface soil erosion processes on timescales of decades, even on hillslopes covered by forest plantations. Thus, without proper management of planted forest to prevent on‐site surface soil loss, both soil degradation and offsite delivery of sediment and associated nutrients can occur. Furthermore, these results confirm that Cs‐137 and Pb‐210_ex measurements can be used to obtain independent estimates of longer‐term soil redistribution rates, which can be used for the validation of erosion models and elucidation of erosion processes.

Acknowledgments

[43] This work was partially supported by the Japan Science and Technology Agency (JST). The authors acknowledge the helpful comments of the referees and an Associate Editor. The assistance of Yusheng Zhang (Department of Geography, University of Exeter, UK) in implementing the conversion model software for estimating soil redistribution rates from Cs‐137 and Pb‐210_ex measurements is gratefully acknowledged. The study represents a contribution to the International Atomic Energy Agency Co‐ordinated Research Project (CRP) “Assessing the effectiveness of soil conservation techniques for sustainable watershed management using fallout radionuclides” and was partially supported by technical contract UK‐12094.