Published online 20 November 2007
Published in Vadose Zone J 6:879-889 (2007)
DOI: 10.2136/vzj2007.0029
© 2007 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
ORIGINAL RESEARCH
Heterogeneous Soil Water Dynamics around a Tree Growing on a Steep Hillslope
Wei-Li Liang*,
Ken'ichirou Kosugi and
Takahisa Mizuyama
Laboratory of Erosion Control, Dep. of Forest Science, Graduate School of Agriculture, Kyoto Univ., Kyoto 606-8502, Japan
* Corresponding author (iritu{at}kais.kyoto-u.ac.jp).
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.
Received 3 February 2007.
 |
ABSTRACT
|
|---|
Precipitation in a forest is intercepted by the canopy and partitioned into throughfall and stemflow, leading to heterogeneous water inputs that affect soil water dynamics. To clarify the effects of a tree stand on rainfall infiltration processes on a steep forested hillslope, we conducted detailed observations of throughfall, stemflow, soil water content, and pore water pressure at high spatial resolution for many storm events. The results showed that the soil water content increased rapidly and greatly in the region downslope from the tree stem, especially at points close to the tree stem. At these points, maximal soil water storage was >100 to 200% of the cumulative open-area rainfall, and occurrences of bypass flow were recognized. Moreover, the pore water pressure at the soil–bedrock interface increased more rapidly and to a greater degree in the region downslope from the tree stem than in the upslope region. For a heavy storm event, the cumulative stemflow per infiltration area along the downslope sides of the tree trunk was 18.9 times the cumulative open-area rainfall. Locally concentrated rainwater input attributable to the stemflow on the downslope side of the tree trunk probably caused the large and rapid increases in water content and pore water pressure in the downslope region, resulting in the development of an asymmetric saturated zone around the tree.
|
INTRODUCTION
|
|---|
In forested landscapes, trees have a major impact on water movement in soil because of the combined effects of canopy interception and stemflow. This has major implications for catchment hydrology, particularly on steep slopes where the impacts on runoff generation and slope stability can have major consequences. Ford and Deans (1978) and Durocher (1990) suggested that very rapid water movement underneath trees was primarily controlled by small-scale spatial variability in the water input to the soil surface rather than by variability in soil physical properties (e.g., macropore distribution). Precipitation intercepted by the canopy is partitioned into throughfall and stemflow, as diffuse input and point input, respectively, so that the water reaching the forest floor is not uniform. Tanaka et al. (1996) emphasized the importance of stemflow as a point source input because it can induce preferential flow. In general, fractures and root channels are primary pathways for preferential flow (Dasgupta et al., 2006). As a principal point of entry to the network of root channels transporting water vertically and horizontally, stemflow tends to follow larger roots into the soil (Lutz and Chandler, 1946, p. 255; Voigt, 1960).
The irregular water input by stemflow may affect not only the spatial and temporal distributions of soil water (Voigt, 1960) but also the physical and chemical properties of the soil (Gersper and Holowaychuk, 1970a,b) and root development (Herwitz and Levia, 1997). Ford and Deans (1977) found a greater concentration of fine roots close to tree stems and noted that this might reflect the spatial pattern of water flux in forest soil. Herwitz (1986) and Tanaka et al. (1991) observed soil erosion scars around the bases of trees and attributed these to infiltration excess produced by concentrated stemflow inputs to the forest floor. The influences of stemflow have been found not only on a small scale at the tree base but also in catchment hydrological processes such as runoff generation (Neave and Abrahams, 2002), groundwater recharge (Taniguchi et al., 1996), and solution chemical responses (Chang and Matzner, 2000).
Thus previous studies have provided information on soil water dynamics around trees and have suggested the influence of stemflow in catchment hydrologic processes. Owing to the low spatial resolution of water content and pore water pressure observations around trees and the limited number of investigated storm events, however, many of these studies have stressed the need for further investigation to generalize the effects of trees. Moreover, most previous studies have been conducted on relatively flat topography and have not adequately addressed the effects of stemflow on steep topography. Generally, trees on steep hillslopes incline toward the slope and have an "S" shape (Schweingruber, 1996, p. 276), unlike trees growing on flat land. Thus the stemflow may differ between trees on hillslopes and those on flat lands. An understanding of stemflow distribution and its effects on soil water dynamics for trees on steep hillslopes is necessary for modeling runoff generation from steep headwater catchments and for predicting shallow landslides.
The purpose of this study was to clarify the effects of a tree stand on rainfall infiltration processes on a steep forested hillslope based on high spatial resolution observations of throughfall, stemflow, soil water content, and pore water pressure for many storm events. Using observations of 141 storm events, we analyzed the preferential flow, heterogeneous soil water storage, and local generation of a saturated zone around a tree.
|
Materials and Methods
|
|---|
Observations began in August 2005 on a hillslope at the Kamigamo experimental station of Kyoto University, located in southern Kyoto Prefecture, central Japan (35°04' N, 135°46' E). The climate is warm temperate. The mean annual air temperature for 1971 to 2000 was 14.6°C, with highest and lowest monthly averages of 27.8°C (August) and 4.6°C (January), respectively. The mean annual precipitation was 1582 mm. Rainfall was distributed year-round, with a peak in summer and just a few centimeters of snow in winter.
The hillslope has a mean gradient of 28°, with brown forest soil classified as a Cambisol underlain by sandstone and slate (Fig. 1
). It is predominantly covered with tall stewartia (Stewartia monadelpha Siebold & Zucc.) planted in 1956. Tall stewartia, which is widespread in natural forests in the western and southern parts of Japan, is a deciduous tree with smooth, exfoliated bark that exhibits leaf fall in November and regrowth in April.
View larger version (53K):
[in this window]
[in a new window]
|
FIG. 1. (a) Topography of the observation area showing the locations of tree stems, tree canopy areas, soil water content measurement points, and throughfall measurement points; (b) side view of tree stands; (c) longitudinal section along the observation line for tree S1 with soil water content and pore water pressure measurement points P1 through P10.
|
|
To monitor the soil water dynamics around a tree, we selected a tall stewartia (S1 in Fig. 1a and 1b: height, 17.47 m; diameter at breast height, 22.3 cm; projected canopy area, 17.39 m2) and delineated a longitudinal observation line from upslope to downslope of this tree. There was no understory vegetation or other trees along the observation line, so that only the effects of tree S1 would be identified (Fig. 1a). We installed capacitance meters (EasyAG-5p, Sentek Sensor Technologies, Stepney, SA, Australia) at each of 10 points: 250 (P1), 200 (P2), 150 (P3), 100 (P4), and 50 cm (P5) upslope from the tree stem, and 25 (P6), 50 (P7), 100 (P8), 150 (P9), and 200 cm (P10) downslope from the tree stem (Fig. 1a and 1c). Each capacitance meter consisted of five sensors to measure soil water content at depths of 10, 20, 30, 40, and 50 cm. Thus we used 50 sensors in total, each of which was assumed to represent the soil water content for each element, defined by the mesh system shown in Fig. 1c. The same type of capacitance meter has been applied in several recent field studies (Mattos et al., 2003; Nachabe et al., 2005).
In June 2006, we installed tensiometers at the soil–bedrock interface at the same 10 points used for surface water content measurements (Fig. 1c). The soil depth to bedrock at each point was determined by penetration tests using a knocking-type cone penetrometer with a 60° bit, a cone diameter of 20 mm, a weight of 2 kg, and a fall distance of 50 cm. From the results of the penetration test, we computed the penetration resistance value, Nh, as the number of blows required for a 10-cm penetration. Previous investigations have proposed that an Nh value of 100 can be an indicator of the boundary between soil and bedrock (Okimura and Tanaka, 1980; Yoshimatsu et al., 2002). The soil depths of all points were estimated to be between 104 and 190 cm (Fig. 1c).
For throughfall and stemflow measurements, we selected another tall stewartia (S2 in Fig. 1a: height, 13.83 m; diameter at breast height, 21.7 cm; projected canopy area, 12.84 m2), located at a similar point on the slope and having a similar tree shape as tree S1. To measure throughfall distribution, we installed a tipping bucket rain gauge (7852M, Davis Instruments, Hayward, CA; 0.2 mm/tip; water collection area, 211 cm2) at each of six points: 200 (TF1), 150 (TF2), and 50 cm (TF3) upslope from the tree stem, and 50 (TF4), 100 (TF5), and 200 cm (TF6) downslope from the tree stem (Fig. 1a). Gross precipitation (open-area rainfall) was measured at an open site 112 m from the observation slope.
To separately collect the stemflow along the upslope (SF-up) and downslope (SF-down) sides of the trunk of tree S2, we used two tubes cut longitudinally and wrapped spirally around the upslope and downslope sides of the trunk (Fig. 2
). The flow rates of SF-up and SF-down were measured using tipping-bucket gauges that tipped at 4.2 mL (7852M, Davis Instruments) and 500 mL (TQX-500, Ikeda, Keiki, Japan), respectively.
All the measurements were conducted until September 2006, with all data simultaneously and automatically recorded at 5-min intervals by a datalogger (CR-1000, Campbell Scientific, Logan, UT).
|
Results and Discussion
|
|---|
General Trends in Change of Soil Water Content and Pore Water Pressure
During the entire observation period of 1 Aug. 2005 through 31 Sept. 2006, 141 storm events occurred. An individual storm event was defined as being separated by 6 consecutive h of no rain. The accumulated rainfall for each event ranged between 0.5 to 194 mm, with the heaviest storm event from 17 to 19 July 2006 being the fourth-largest accumulated rainfall (194 mm) for 1996 to 2006. Figure 3
shows results observed for 4 to 20 July 2006. This time period had eight storm events and was divided into three periods: a light-precipitation period (Period I, 4–7 July), a no-precipitation period (Period II, 7–15 July), and a heavy-precipitation period (Period III, 15–20 July). Figure 3 does not show the results for locations P1, P2, or P4, which were similar to those for P3, or the results for P8 or P10, which were similar to those for P9.
In Period I, values of soil water content (
) at P3 and P5 increased at depths of 10, 20, and 30 cm, but no obvious changes were measured at 40 and 50 cm. At P6 and P7, values increased at all depths up to 50 cm. The changes in at P9 were similar to those at P3 and P5, with the changes confined to only the upper layers. Values of pore water pressure head (
) rose slightly in the upslope region (P3 and P5) and markedly in the downslope region (P6, P7, and P9). In particular, at P6 and P7 became positive during storm events on 5 July, indicating the transient generation of a saturated zone.
In Period II, values gradually decreased at all points. The values showed greater decreases at points close to the tree (P5, P6, and P7) than at points away from the tree (P3 and P9) and exhibited diurnal changes at the end of the period, which might be the result of water uptake by the tree.
In Period III, the values changed greatly at every point, corresponding to the heavier rainfall intensities during this period than during Period I. In particular, dramatic increases and decreases in were measured at depths of 30 and 50 cm at P6, and at a depth of 30 cm at P7. The values changed more gradually at P9 than at P6 and P7, so that P9 had a similar trend to those at P3 and P5. The values increased greatly at every point. Before the heavy storm event of 17 July, the value at P3 was still low, whereas the value at P5 had already begun a gradual rise. During the heavy storm event from 17 to 19 July, the maximum values at P3 and P5, which were located upslope from the tree stem, were 2.3 and 0.2 cm, respectively. In the downslope region, the values showed very spiked responses. In this region, a saturated zone occurred frequently, with maximal values of 30.0 and 16.3 cm at P6 and P7, respectively; P9 had a lower maximal value (4.3 cm).
The results for Period I suggest that, during small storm events, remarkable increases in and occurred only at the points closest to the tree stem in the downslope region and that occurrences of positive pore water pressure were limited to those points. In contrast, the results for Period III suggest that obvious increases in and occurred at every point during a heavy storm event. A saturated zone expanded widely around the tree stem. At the same time, the largest positive pore water pressure was observed at the points closest to the tree stem in the region downslope from the tree. At those points, at certain depths showed very spiked responses on the hyetograph.
To clarify the dynamics of and during small and heavy storm events, the data for two storm events (Fig. 3, blue-shaded areas) were analyzed.
Change in Soil Water Content and Pore Water Pressure during a Small Storm Event
Figure 4
shows results for the small storm event on 5 July 2006 (accumulated rainfall, 14.5 mm; average wind speed, 0.04 m/s; average wind direction, southwest), where soil water content change (
) is defined as the difference between the water content at the time of measurement and the initial water content observed at the start of the storm event. The distribution of clearly indicates where the water brought by the storm event was stored. The value of the hydraulic head (
) was computed as the sum of the observed and the height for the measurement, which corresponds to the height of the soil–bedrock interface (see Fig. 1c). Each lower panel in Fig. 4 shows the soil–bedrock interface, with the value being the difference between the value and the height of the soil–bedrock interface. For values greater than the soil–bedrock interface, the generation of positive pore water pressure is indicated. In addition, each panel shows the initial distribution observed at the start of the storm event.
At 170 min (Fig. 4a), the accumulated rainfall was 3.5 mm, and obviously increased at P6, especially at the 50-cm depth, where was >0.2. At the same time, at P6 became higher than the height of the soil–bedrock interface, indicating the generation of positive pore water pressure. No obvious changes in and were observed at the other points.
At 220 min (Fig. 4b), the accumulated rainfall was 10 mm, and the rainfall intensity reached the maximum (2 mm in 10 min) for this storm event, as did the average (0.029) for all points and depths (total of 50). At P6 and P7, increased greatly, especially at depths of 30 and 50 cm at P6 and 30 cm at P7; however, the increases in at other points were limited to the upper soil layers, from 0 to 10 cm. At P6, reached a maximum in this event (corresponding value of 25.3 cm), and increases in were also found at P7 and P9. There were no changes in in the upslope region.
When the rainfall intensity decreased to 0.5 mm in 10 min at 230 min (Fig. 4c), decreased greatly at P6 and P7 compared with the at 220 min (Fig. 4b). The increase in reached a maximum at P7 in this event (corresponding value of 9.8 cm), and positive pore water pressures were measured at P6 and P7. No changes in the values were seen in the region upslope from the tree stem, including at P5 (50 cm from the tree stem).
When the storm event ended at 380 min (Fig. 4d), increases from the initial values were observed for at all points in the downslope region except P10. At P6, the generation of positive continued. In the upslope region, the subsurface layer had smaller values than the surface layer, suggesting that the wetting front was gradually expanding downward. In the upslope region, no obvious change in was observed with this event, except for an increase of only 3.5 cm at P5.
In summary, increased rapidly and greatly at P6 and P7, especially at depths of 30 and 50 cm at P6 and 30 cm at P7. The increases in at other points were dominated by a slow expansion of the wetting front. The rapid propagation of infiltrated water at P6 and P7 caused a significant rise of at the points closest to the tree stem in the downslope region, resulting in the generation of a saturated zone; however, no obvious change in was observed in the upslope region, even at the points closest to the tree stem.
Change in Soil Water Content and Pore Water Pressure during a Heavy Storm Event
Figure 5
shows the results for a heavy storm event that occurred from 17 to 19 July 2006 (accumulated rainfall, 194 mm; average wind speed, 0.07 m/s; average wind direction, southeast). At 40 min, when the accumulated rainfall was still just 1 mm (Fig. 5a), remarkable increases in were measured at P6, and at P6 rose over the bedrock surface, indicating the generation of a saturated zone.
At 200 min (Fig. 5b), the accumulated rainfall was 23.5 mm. Increased values were measured at all points, with the greatest increases occurring at 30 and 50 cm at P6 and 30 cm at P7. This trend was similar to that observed for the small storm event (see Fig. 4b). At the other points, the surface layer generally had a greater increase than the subsurface layers. At all points, values in the downslope region rose to near or above the bedrock surface. Especially at P6 and P7, saturated zones were formed, and the maximum values for this event were recorded at P6 (30.0 cm) and P7 (16.3 cm). There were no obvious changes in at any of the points in the upslope region. The values at P5 and P6 indicated the probable occurrence of water flow in the upslope direction.
At 520 min (Fig. 5c), the accumulated rainfall was 66.5 mm, and rainfall intensity had reached a maximum (5 mm in 10 min) for the event. The distribution was similar to that observed at 200 min (Fig. 5b). In the upslope region, an increase in was observed at all points except P4.
At 2770 min (Fig. 5d), the accumulated rainfall was 175 mm, and the average at all points and depths became the largest (0.056) for this storm event. The distribution of was very similar to that observed at 200 min (Fig. 5b). In the region upslope from the tree stem, the values were further increased compared with the values at 520 min (Fig. 5c).
In summary, the distributions observed during the heavy storm event were similar to those observed during the small storm event (Fig. 4). The heavy storm event caused increases in the pore water pressure at the soil–bedrock interface in the upslope region, where no such increases were observed during the small event. The downslope region, however, showed much more rapid and greater pore pressure increases than the upslope region. During the first high rainfall intensity period, we observed an expansion of a saturated zone from the downslope to the upslope region around the tree.
Throughfall and Stemflow Distribution Measured for a Heavy Storm Event
Figure 6
shows the accumulated heights of the open-area rainfall (OR), throughfall (TF), and stemflow for the heavy storm event illustrated in Fig. 5. The stemflow was measured separately along the upslope (SF-up) and downslope (SF-down) sides of the trunk (see Fig. 2). Stemflow height is shown in two ways: the observed total stemflow volume divided by the canopy projected area (Ac) and by the stemflow infiltration area (As). The Ac values in the regions upslope and downslope of tree S2 were 1.67 and 11.17 m2, respectively (Fig. 1a). To determine As, we used a model proposed by Tanaka et al. (1991, 1996), which assumes a ring infiltration area and the following relationship between the radius of the infiltration area, y (cm), and the diameter of the tree base, x (cm):
 | [1] |
Then As is calculated by
 | [2] |
Using the measured value of x = 21.7 cm, y and As were calculated to be 42.2 cm and 5220 cm2, respectively. We used half the value of As (2610 cm2) to calculate the heights of SF-up and SF-down.
View larger version (16K):
[in this window]
[in a new window]
|
FIG. 6. Accumulated heights of the open area rainfall (OR), throughfall (TF), and stemflow along upslope (SF-up) and downslope (SF-down) sides of the trunk for the heavy storm event on 17 to 19 July 2006. Stemflow height is shown in two ways: the observed total stemflow volume divided by the canopy projected area (Ac) or by the stemflow infiltration area (As).
|
|
As shown in Fig. 6, throughfall ranged between 51.4 and 121.8 mm and tended to be larger in the upslope region and near the tree stem (i.e., TF3 and TF4). The average throughfall for the six points was 91.7 mm, corresponding to 47.3% of the cumulative OR (194 mm). This result indicates that more than half of the precipitation intercepted by leaves and branches was partitioned into stemflow and evaporation. Stemflow per the canopy projected area was greater in the downslope region than in the upslope region, and the weighted average was 75.5 mm, corresponding to 38.9% of OR. The stemflow ratio was very high in comparison with that reported for other tree species in previous studies. For example, Huber and Iroumé (2001) conducted long-term observations of precipitation, throughfall, and stemflow at 29 research plots covering a wide range of rainfall zones and forest types, species, ages, and densities, and reported a ratio of throughfall to total precipitation of 55 to 86%, with a greater ratio in coniferous stands than in broadleaved forests. In their study, the ratio of stemflow to total precipitation ranged from 1 to 13% in coniferous stands and from 1 to 8% in broadleaved forests.
As shown in Fig. 6, the stemflow per infiltration area on the downslope side of the tree stem was 3670 mm, which was 18.9 times the open-area rainfall. In contrast, the stemflow on the upslope side was only 45.6 mm, or 23.5% of the open-area rainfall. During a storm event on 11 Apr. 2006, an intense water flow developed on the downslope side of the trunk of tree S1, at a rainfall intensity of 3 mm/h (Fig. 7
). At the same time, no obvious flow was observable on the upslope side of the trunk. A similar trend was observed for all tall stewartia trees in the study area (Fig. 1a), including tree S2 for which throughfall and stemflow were measured.
We presume that such large differences in stemflow amount between the downslope and upslope sides are attributable to the uneven distribution of the canopy (Fig. 1a) and the tilt of the whole tree shape toward the downslope direction (Fig. 1b). Trees S1 and S2 tilted approximately 6 and 4° in the downslope direction, respectively. Although many studies have reported funneled water input caused by stemflow (Ford and Deans, 1978; Herwitz, 1986; Durocher, 1990; Taniguchi et al., 1996), such a large difference in stemflow between the upslope and downslope sides of a tree trunk has not been previously reported. A comparison between the soil water dynamics (Fig. 5) and the throughfall and stemflow distribution (Fig. 6) shows the significant difference in soil water dynamics between the upslope and downslope sides of a tree resulting from uneven distribution of the stemflow.
Spatial Distribution of Water Content Increase
Figures 4b and 5d show the spatial distribution of when the average of all points and depths (total of 50) reached a maximum for the small and heavy storm events, respectively. For all 118 storm events that caused increases in soil water content during the entire observation period, we computed the distribution when the average value reached its maximum. To remove the factor of rainfall magnitude, at each observation point i (i = 1, 2, ..., 50) for the jth storm event (ij) was rescaled by
 | [3] |
where µj and 
j are the mean and standard deviation of ij (i = 1, 2, ..., 50) for the jth storm event.
The spatial distribution of the rescaled water content increase, ij*, is shown as a box plot in Fig. 8
. Points P6 and P7, which were close to the downslope side of the tree stem, showed water content increases greater than the average (that is, ij* > 0) for most storm events. Point P6, which was located 25 cm downslope, showed an increase much greater than the average at every depth. At points other than P6 and P7, ij* was small and tended to decrease with increasing depth, suggesting that the rainwater stayed within the surface layers and that the wetting front had not yet reached farther depths. Such dependence of ij* on depth was not observed at P6 or P7, indicating that the rainwater rapidly expanded to the deeper layers at these points. Figure 8 shows that the spatial distribution of the water content increase observed for the small and heavy storm events shown in Fig. 4b and 5d was commonly found for all storm events.
Soil Water Storage
For the quantitative analysis of rainwater storage, the maximum increase in soil water storage from the surface through the 55-cm depth, Smax, at each observation point and for each storm event was compared with the cumulative open-area rainfall from the start of the event to the time when Smax was recorded (Fig. 9
).
At P6 and P7, which were located 25 and 50 cm downslope from the tree stem, the Smax values were more than 200 and 100%, respectively, of the cumulative open-area rainfall for most of the storm events. These results indicate that a large amount of rainwater converged at the downslope side of the tree stem. Except at P6 and P7, the Smax was less than the cumulative open-area rainfall. Although P7 and P5 were both located 50 cm from the tree stem, the Smax was much larger at P7 than at P5, clearly showing that the amount of water input was very different between the downslope and upslope sides of the tree stem.
For every point other than P6, the initial soil moisture condition showed no clear effect on the Smax (Fig. 9). At P6, the Smax under the wet antecedent condition tended to be smaller than that under the dry antecedent condition for storm events with cumulative rainfall >20 mm. When a heavy storm supplied a large amount of rainwater at P6 under the wet antecedent condition, the soil water storage at P6 easily reached the maximum water-holding capacity determined by the soil hydraulic properties. Thus, under the wet antecedent condition, the Smax did not increase at P6 when the cumulative rainfall was greater than about 20 mm, resulting in a smaller Smax than that under the dry antecedent condition.
Generation of Bypass Flow around the Tree
In Fig. 8, * at P6 tended to be large at depths of 10, 30, and 50 cm, and * at P7 was extremely large at 30 cm for some storm events. These irregular vertical distributions suggest the existence of bypass flows affected by tree root networks at P6 and P7; however, the irregular vertical distributions might be explained by the heterogeneous distribution of soil porosity that is effective for water storage rather than bypass flow.
In general, the occurrence of bypass flow is recognized when water content and pore water pressure respond to rainwater input earlier in deeper soil layers than in shallower layers (Kobayashi et al., 2000). For example, van Stiphout et al. (1987) demonstrated that water infiltration induced by bypass flow reached to depths of 60 and 135 cm, where macropores became discontinuous, whereas the wetting front in the soil matrix expanded only to depths of 7 to 12 cm. To evaluate bypass flow, we investigated occurrences of earlier water content response in a deeper layer (ERDL) at each point for each storm event. Here, we assumed that a water content increase of >0.005 cm3 cm–3 indicated a response to the rainwater input.
In Fig. 10
, each raw value corresponds to a specific storm event, and each column corresponds to a single sampling point (P1–P10) for the water content measurement. The white columns indicate points at which the water content did not respond to the storm event. The red and blue columns indicate points at which ERDL did and did not occur, respectively. Storm events are sorted primarily by total event rainfall (Ra) and secondarily by the average water content (Wi) at the start of the storm event for the region covered by the 10 monitoring points.
View larger version (29K):
[in this window]
[in a new window]
|
FIG. 10. Occurrences of earlier water content response in the deeper layer (ERDL) at pore water pressure measurement points P1 through P10 in relation to the event total rainfall (Ra) and the average water content at the start of the storm event (Wi). The green line shows the location of the tree.
|
|
At P6 and P7, the water content responded to 83 and 63% of all storm events, respectively. The ratio was much smaller at the other points, ranging from 44 to 57%. At P6, ERDL occurred for 42% of the storm events that caused a response in soil water content at P6. This ratio was high at P7 (62%), medium at P8 through P10 (32–34%), and small at P1 through P5 (1–18%). These results indicate that bypass flow frequently occurred in the region downslope of the tree stem, especially in the areas 25 and 50 cm downslope. Conversely, bypass flow hardly happened in the upslope region, even at P5, which was adjacent to the tree.
Gaiser (1952) found at least one vertical channel per square meter of forest soil; the channels were formed by decayed root systems and contained materials more permeable to water than the surrounding soils. Furthermore, Noguchi et al. (1997, 1999) pointed out that infiltrated flow impeded by living roots can be diverted around the perimeter of the roots, as shown by white paint staining. Hiramatsu and Kumazawa (2002) reported that the equivalent hydraulic conductivity in areas surrounding tree roots was 1.1 to 3.2 times the soil matrix conductivity and that rainwater can be expected to concentrate around roots. These studies indicate that tree roots can form channels that serve as pathways for rapid water movement. Thus, we presume that, when a large amount of rainwater was supplied by stemflow to the downslope side of a tree stem, not all of the water infiltrated into the soil matrix, causing infiltration excess and activating these pathways. Thus, rainwater flowed along these pathways as bypass flow. This hypothesis was supported by the observations of the root system of tree S3 (Fig. 11
), which was located just below the monitoring region for tree S1 (Fig. 1a). Figure 11 shows that, within the surface soil layer 40 cm thick, several thick roots 2 to 10 cm in diameter elongated in the downslope direction from the stem bottom, which could form the flow pathways. We also found many fine roots from the soil surface down to 50-cm depth. Below the depth of 50 cm, the number of roots was small.
As shown in Fig. 10, bypass flow was frequently observed for heavy storm events with large amounts of total precipitation. For small storm events with <6 mm of total precipitation, bypass flow tended to occur under the dry antecedent condition. Kobayashi et al. (2000) reported that less water was needed to generate bypass flow under dry conditions, and water repellency of the surface soil was thought to explain this tendency. Several recent studies have pointed out that water repellency can be a major factor inducing bypass flow (Bauters et al., 1998; Täumer et al., 2006). Therefore, it is possible that soil water repellency affected the occurrence of bypass flow at our study site for small storm events when the soil was initially dry.
Generation of a Saturated Zone around the Tree
As for the small and heavy storm events shown in Fig. 4 and 5, respectively, the response of pore water pressure, , at the soil–bedrock interface was greatly affected by the location relative to the tree stem. Similar results were obtained for all the storm events studied, as illustrated in Fig. 12
, which presents the raw value corresponding to each storm event from June to September 2006, with a column corresponding to each point (P1–P10) for the measurement. The white columns indicate points where did not respond to the storm event; the other column colors correspond to the maximum pore water pressure, max, observed for the event. The numbers in each column represent the order of the response time; for example, the number 1 for P6 indicates that, among the 10 points, P6 showed the earliest increase in . The storm events in Fig. 12 are sorted primarily by total event rainfall (Ra) and secondarily by the average values at the 10 monitoring points at the start of the storm event.
The values responded to rainwater input more frequently in the downslope region than in the upslope region. In particular, a response was measured most frequently at P6, even for small storm events with total rainfall of <4 mm. For every event, increased earliest at P6. Comparing P5 and P7, both of which were 50 cm from the tree stem, always responded earlier at P7 than at P5. Thus, the values tended to increase earlier at downslope points than at upslope points. Moreover, max was larger downslope from the tree stem than in the upslope region; max was largest and frequently exceeded 20 cm at P6, which was 25 cm downslope from the tree stem. In the downslope region, max was frequently positive, indicating the generation of a saturated zone, whereas the upslope region remained unsaturated, except for one storm event. In the upslope region, P5 had the most frequent responses. We presume that the large increases in at P6 (e.g., Fig. 5b) caused a total head gradient in the upslope direction, expanding the saturated zone at P6 to the upslope region and causing responses at P5.
An extremely rapid response in the downslope region and an asymmetrically distributed saturated zone around the tree can greatly affect the waveform of a storm hydrograph, as well as the location and timing of shallow landslide occurrences because the formation of perched groundwater in the soil mantle generally controls rainwater discharge and slope instability on steep landscapes (Anderson and Sitar, 1995; Wang and Sassa, 2003). For a single storm event at a gentle hillslope inclined at 3°, Durocher (1990) observed that soil water pressure tended to show more rapid and greater changes underneath the trees than in areas apart from the trees, and attributed this to the concentration of rainwater by stemflow; however, Durocher (1990) did not find any difference in the pressure responses between regions upslope and downslope from the tree stem. Our study demonstrates significantly different responses of pore water pressure at the soil–bedrock interface between upslope and downslope regions for various storm events, thereby clarifying the effects of a tree stand on rainfall infiltration processes on a steep forested hillslope.
|
Conclusions
|
|---|
In this study, we conducted detailed observations of soil water dynamics for many storm events around stands of tall stewartia growing on a steep hillslope. Topography and stemflow had a large influence on soil water content, which increased rapidly and greatly in the region downslope from the stem. Upslope from the stem, however, the increase in soil water content was small and dominated by a slow expansion of the wetting front. Bypass flow, indicated by an irregular vertical distribution of water, was much greater downslope from the stem than upslope.
As a consequence of stemflow, topography, and bypass flow, the pore water pressure at the soil–bedrock interface increased rapidly and greatly in the downslope region, especially at points close to the tree stem. In the upslope region, the increases in the pore water pressure were slow and small. This produced an asymmetric saturated zone around the tree. In comparison to studies on other tree species, tall stewartia produced smaller throughfall and greater stemflow, with respective percentages of 47.3 and 38.9%.
This study clarified the significant difference in soil water dynamics between the upslope and downslope sides of a tree stand growing on a steep hillslope. Investigation of soil water dynamics around other tree species on steep hillslopes could be one direction for future studies. In addition, the phenomenon of the irregular water input caused by tree stands should be combined with a hydrologic hillslope model for accurately estimating the soil water dynamics on a steep hillslope.
|
ACKNOWLEDGMENTS
|
|---|
We thank S. Miyata, Y. Yamakawa, and other colleagues at the Erosion Control Laboratory of Kyoto University for their support in field observations. The open area rainfall and study site environmental data in this study were provided by Kamigamo Experimental Station, Kyoto University. This work was partly supported by a CREST project of Japan Science and Technology Agency.
|
REFERENCES
|
|---|
- Anderson, S.A., and N. Sitar. 1995. Analysis of rainfall-induced debris flows. J. Geotech. Eng. ASCE 121:544–552.[CrossRef]
- Bauters, T.W.J., D.A. DiCarlo, T.S. Steenhuis, and J.-Y. Parlange. 1998. Preferential flow in water-repellent sands. Soil Sci. Soc. Am. J. 62:1185–1190.[Abstract/Free Full Text]
- Chang, S.-C., and E. Matzner. 2000. The effect of beech stemflow on spatial patterns of soil solution chemistry and seepage fluxes in a mixed beech/oak stand. Hydrol. Processes 14:135–144.[CrossRef]
- Dasgupta, S., B.P. Mohanty, and J.M. Köhne. 2006. Impacts of juniper vegetation and karst geology on subsurface flow processes in the Edwards Plateau, Texas. Vadose Zone J. 5:1076–1085.[Abstract/Free Full Text]
- Durocher, M.G. 1990. Monitoring spatial variability of forest interception. Hydrol. Processes 4:215–229.[CrossRef]
- Ford, E.D., and J.D. Deans. 1977. Growth of a Sitka spruce plantation: Spatial distribution and seasonal fluctuations of lengths, weights, and carbohydrate concentrations of fine roots. Plant Soil 47:463–485.[CrossRef][Web of Science]
- Ford, E.D., and J.D. Deans. 1978. The effects of canopy structure on stemflow, throughfall, and interception loss in a young Sitka spruce plantation. J. Appl. Ecol. 15:905–917.[CrossRef]
- Gaiser, R.N. 1952. Root channels and roots in forest soils. Soil Sci. Soc. Am. Proc. 16:62–65.
- Gersper, P.L., and N. Holowaychuk. 1970a. Effects of stemflow water on a Miami soil under a beech tree: I. Morphological and physical properties. Soil Sci. Soc. Am. Proc. 34:779–786.
- Gersper, P.L., and N. Holowaychuk. 1970b. Effects of stemflow water on a Miami soil under a beech tree: II. Chemical properties. Soil Sci. Soc. Am. Proc. 34:786–794.
- Herwitz, S.R. 1986. Infiltration-excess caused by stemflow in a cyclone-prone tropical rainforest. Earth Surf. Processes Landforms 11:401–412.[CrossRef]
- Herwitz, S.R., and D.F. Levia. 1997. Mid-winter stemflow drainage from bigtooth aspen (Populus grandidentata Michx.) in central Massachusetts. Hydrol. Processes 11:169–175.[CrossRef]
- Hiramatsu, S., and Y. Kumazawa. 2002. Influence of forest roots on hydrological processes in forest soil. (In Japanese, with English abstract.) J. Jpn. Soc. Erosion Control Eng. 55(4):12–22.
- Huber, A., and A. Iroumé. 2001. Variability of annual rainfall partitioning for different sites and forest covers in Chile. J. Hydrol. 248:78–92.[CrossRef]
- Kobayashi, M., S. Onodera, and M. Kato. 2000. Effect of a tree on water movement in a forest soil. (In Japanese, with English abstract.) J. Jpn. For. Soc. 82:287–294.
- Lutz, H.J., and R.F. Chandler. 1946. General physical properties of forest soils. p. 223–289. In H.J. Lutz and R.F. Chandler (ed.) Forest soils. John Wiley & Sons, New York.
- Mattos, D.J., D.A. Graetz, and A.K. Alva. 2003. Biomass distribution and nitrogen-15 partitioning in citrus trees on a sandy Entisol. Soil Sci. Soc. Am. J. 67:555–563.[Abstract/Free Full Text]
- Nachabe, M., N. Shah, M. Ross, and J. Vomacka. 2005. Evapotranspiration of two vegetation covers in a shallow water table environment. Soil Sci. Soc. Am. J. 69:492–499.[Abstract/Free Full Text]
- Neave, M., and A.D. Abrahams. 2002. Vegetation influences on water yields from grassland and shrubland ecosystems in the Chihuahuan Desert. Earth Surf. Processes Landforms 27:1011–1020.[CrossRef]
- Noguchi, S., A.R. Nik, B. Kasran, M. Tani, T. Sammori, and K. Morisada. 1997. Soil physical properties and preferential flow pathways in tropical rain forest, Bukit Tarek, peninsular Malaysia. J. For. Res. 2:115–120.[CrossRef]
- Noguchi, S., Y. Tsuboyama, R.C. Sidle, and I. Hosoda. 1999. Morphological characteristics of macropores and the distribution of preferential flow pathways in a forested slope segment. Soil Sci. Soc. Am. J. 63:1413–1423.[Abstract/Free Full Text]
- Okimura, T., and S. Tanaka. 1980. Researches on soil horizon of weathered granite mountain slope and failured surface depth in a test field. (In Japanese, with English abstract.) J. Jpn. Soc. Erosion Control Eng. 33(1):7–16.
- Schweingruber, F.H. 1996. Influence of mass movement. p. 271–287. In F.H. Schweingruber (ed.) Tree rings and environment dendroecology. Paul Haupt Publ., Bern, Switzerland.
- Tanaka, T., M. Taniguchi, and M. Tsujimura. 1996. Significance of stemflow in groundwater recharge: 2. A cylindrical infiltration model for evaluating the stemflow contribution to groundwater. Hydrol. Processes 10:81–88.[CrossRef]
- Tanaka, T., M. Tsujimura, and M. Taniguchi. 1991. Infiltration area of stemflow-induced water. Annu. Rep. Inst. Geosci. Univ. Tsukuba 17:30–32.
- Taniguchi, M., M. Tsujimura, and T. Tanaka. 1996. Significance of stemflow in groundwater recharge: 1. Evaluation of the stemflow contribution to recharge using a mass balance approach. Hydrol. Processes 10:71–80.[CrossRef]
- Täumer, K., H. Stoffregen, and G. Wessolek. 2006. Seasonal dynamics of preferential flow in a water repellent soil. Vadose Zone J. 5:405–411.[Abstract/Free Full Text]
- van Stiphout, T.P.J., H.A.J. van Lanen, O.H. Boerma, and J. Bouma. 1987. The effect of bypass flow and internal catchment of rain on the water regime in a clay loam grassland soil. J. Hydrol. 95:1–11.[CrossRef]
- Voigt, G.K. 1960. Distribution of rainfall under forest stands. For. Sci. 6:2–10.
- Wang, G., and K. Sassa. 2003. Pore-pressure generation and movement of rainfall-induced landslide: Effect of grain size and fine-particle content. Eng. Geol. 69:109–125.[CrossRef]
- Yoshimatsu, H., K. Kawamitsu, K. Seo, S. Hasegawa, and S. Muranaka. 2002. Simplified penetrometer for surface structure survey in hillslopes. (In Japanese.) Trans. Jpn. Soc. Erosion Control Eng. 2002:392–393.
This article has been cited by other articles:
|
|
|
|
 
K. Kosugi, Y. Yamakawa, N. Masaoka, and T. Mizuyama
A Combined Penetrometer-Moisture Probe for Surveying Soil Properties of Natural Hillslopes
Vadose Zone J.,
February 10, 2009;
8(1):
52 - 63.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
