- Email: [email protected]
Changes in peak flow with decreased forestry practices: Analysis using watershed runoff data
Journal of Environmental Management 92 (2011) 1528e1536
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Changes in peak flow with decreased forestry practices: Analysis using watershed runoff data Hikaru Komatsu a, b, *, Yoshinori Shinohara a, Tomonori Kume b, Kyoichi Otsuki a a b
Kasuya Research Forest, Kyushu University, 394 Tsubakuro, Sasaguri, Kasuya, Fukuoka 811-2415, Japan School of Forestry and Resource Conservation, National Taiwan University, 1, Sce. 4, Roosevelt Road, Taipei 106, Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 April 2010 Received in revised form 3 December 2010 Accepted 5 January 2011 Available online 1 February 2011
The prevalence of forestry practices such as thinning and pruning have gradually decreased since the 1980s. Researchers have noted an increased flood risk with decreased forestry practices for coniferous plantations in Japan on the basis of infiltration and overland flow measurements at a plot scale (typically several square meters). However, no studies have examined changes in peak flow with decreased forestry practices at a watershed scale (typically several tens or hundreds of square kilometers) even though flood disasters generally occur at this scale in Japan. We examined changes in frequency distributions of daily precipitation (P) and runoff (Q) during the period 1979e2007 at the Terauchi watershed, where forestry practices are known to have decreased. For this purpose, we divided P and Q data into 14 and 15 classes according to the magnitude, respectively, and examined changes in the frequency for each class during the period. We observed no significant increasing trend for any P or Q class. Even when taking into account the effect of interannual variations in precipitation on the frequency for each Q class, there was no significant increasing trend in the frequencies except for two Q classes with moderate Q values. These results suggest that the increase in flood risk due to decreased forestry practices might be less than expected. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: Flood Forestry practices Peak flow Plantation Runoff Watershed
1. Introduction Watershed runoff is influenced by many factors, such as climate, geology, topography, and vegetation (Ward and Robinson, 2000; Komatsu et al., 2008a; Shinohara et al., 2009). Human activities that influence vegetation cover could affect the runoff regime (e.g., Yue and Hashino, 2005; Komatsu et al., 2008b, 2009b; Yao et al., 2009), which in turn could alter flood and drought risks. Forestry clear-felling (or conversion from forests to pasture) and development of forest cover following clear-felling (or conversion from pasture to forests) are the most significant examples of vegetation change. Many studies (e.g., Bosch and Hewlett, 1982; Brown et al., 2005; Farley et al., 2005; Grant et al., 2008) have examined the effect of clear-felling or development of forest cover on the runoff regime. Changes in the runoff regime with clear-felling and development of forest cover are not critical issues in Japan, because such land-use changes have recently been uncommon, and therefore,
* Corresponding author. Kasuya Research Forest, Kyushu University, 394 Tsubakuro, Sasaguri, Kasuya, Fukuoka 811-2415, Japan. Tel.: þ81 92 948 3109; fax: þ81 92 948 3119. E-mail address: [email protected] (H. Komatsu). 0301-4797/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2011.01.010
the percentage of forested area has been nearly constant over the past 40 years (National Astronomical Observatory, 2009). Instead, forestry practices and their effect on runoff have recently been raising as a concern in Japan (e.g., Kuraji and Hoyano, 2004; Calder, 2005; Komatsu et al., 2007b,c, 2009a; Onda, 2008; Onda et al., 2010). Japan developed large areas of coniferous plantations for timber production from the 1950s to 1970s, mainly by converting natural broadleaved forests (Fujimori, 2000). Coniferous plantations occupy approximately 40% of forested areas and approximately 25% of Japan’s land surface (National Astronomical Observatory, 2009). To develop these plantations, seedlings are planted typically at approximately 3000 stems ha1. Trees are thinned and pruned several times in the decades after planting and are harvested at an age of 40e50 years (e.g., Kikuzawa, 1999). By this stage, the stem density is typically 500 stems ha1. Forestry practices such as thinning and pruning were conducted in these plantations during the 1950e1980 period. However, the prevalence of these practices has gradually decreased since the 1980s, because of low timber prices and increased labor costs (Fujimori, 2000; Japan Forestry Agency, 2007; Onda, 2008). Thus, many plantations in Japan have become unmanaged. Researchers (Kuraji and Hoyano, 2004; Kuraji, 2007; Onda, 2008; Onda et al., 2010) have noted increased flood risk with
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
decreased forestry practices in coniferous plantations in Japan. The leaf area index (LAI) of unmanaged coniferous plantations is high owing to the absence of forestry practices such as thinning and pruning, which results in the sparse distribution of understory vegetation (Kiyono, 1988; Fukada et al., 2005, 2006; Japan Forestry Agency, 2007). When understory vegetation is sparse, throughfall strikes the ground surface directly, resulting in compaction of the ground surface and therefore lower rates of infiltration of the soil surface and greater overland flow (e.g., Yukawa and Onda, 1995; Onda and Yukawa, 1995; Gomi et al., 2008a; Kato et al., 2008; Hirano et al., 2009; Hiraoka et al., 2010). The infiltration rate for unmanaged coniferous plantations has been found to be lower than that for bare soil in some cases (Onda, 2008). This contrasts to having dense understory vegetation, in which case the infiltration
1529
rates for the ground surface are relatively high. In addition, the presence of understory vegetation reduces the persistence of overland flow even when flow does occur, owing to the roughness that the vegetation provides (Gomi et al., 2008a,b, 2010a; Wakiyama et al., 2010). Researchers (Kuraji and Hoyano, 2004; Kuraji, 2007; Onda, 2008; Onda et al., 2010) speculated that these results imply increased flood risk in watersheds where coniferous plantations are unmanaged. On the basis of this speculation, many local governments in Japan have introduced local taxes to stimulate forestry practices (Imawaka and Sato, 2008). However, the increase in flood risk might be less than expected for two reasons. First, the speculation is mainly based on plot-scale measurements (typically within several square meters) of infiltration and overland flow (e.g., Hattori et al., 1992; Tsujimura et al.,
Fig. 1. (a) Location of the study area and (b) topographical map of the area.
1530
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
overstory, and high decomposition rates of litters. Although unmanaged C. obtusa plantations satisfy both these conditions, unmanaged C. japonica plantations only satisfy the former condition. Most watersheds in Japan include large areas of C. japonica plantations as well as large areas of C. obtusa plantations (National Astronomical Observatory, 2009). Thus, the increase in flood risk might not be considerable because of the species composition. No studies have examined changes in flood risk with decreased forestry practices in coniferous plantations in Japan at a watershed scale. Flood disasters generally occur at this scale in Japan, and the necessity of studies at spatial scales larger than a plot scale has been pointed out (Kuraji and Hoyano, 2004; Onda, 2008). Thus, we examine changes in the runoff regime using daily runoff data obtained for a watershed where forestry practices are known to have decreased.
2. Study area and data Fig. 2. Frequency distribution of coniferous plantation tree ages for the Terauchi watershed. The vertical axis indicates the area of each age class divided by the whole area of coniferous plantations.
2006; Miyata et al., 2007; Gomi et al., 2008a,b; Hirano et al., 2009). Recent studies in Japan (e.g., Sidle et al., 2007; Gomi et al., 2010a,b) performed overland flow measurements in many plots of various size (several square meters to several tens of square meters). They reported distinct differences in overland flow occurrence between unmanaged and managed coniferous plantations and between unmanaged coniferous plantations and natural broadleaved forests in small plots. However, the differences were less significant in large plots. This phenomenon is referred to as the scale effect (Gomi et al., 2008b, 2010a,b) and has also been reported in other countries (e.g., Duley and Ackerman, 1980; Joel et al., 2002; Parsons et al., 2006). Furthermore, Hirano et al. (2009) examined differences in runoff between unmanaged coniferous plantations and natural broadleaved forests at a small-catchment scale (w1 ha) and reported no clear differences in catchment runoff during large storms, although they reported higher runoff for the unmanaged coniferous plantations during small storms. These results imply that flood risk does not considerably increase at a watershed scale (typically several tens or hundreds of square kilometers) in the case of decreased forestry practices because of the scale effect even though flood disasters generally occur at this scale in Japan (e.g., Sidle et al., 2007; Gomi et al., 2010a,b). Second, plantations in Japan mainly comprise Cryptomeria japonica and Chamaecyparis obtusa. These two species account for approximately 70% of the plantation area in Japan (National Astronomical Observatory, 2009). Although many studies have reported low infiltration rates and significant overland flow for unmanaged C. obtusa plantations (e.g., Miyata et al., 2007; Sidle et al., 2007; Gomi et al., 2008a,b, 2010b), no studies have reported the same results for unmanaged C. japonica plantations. Litters of C. japonica have lower decomposition rates than those of C. obtusa (Miura, 2000; Ichikawa et al., 2005, 2006). Thus, the soil surface of C. japonica plantations is generally covered with a litter layer, which contrasts with the soil surface of C. obtusa plantations (Miura, 2000; Ichikawa et al., 2003). This suggests that throughfall would not directly strike the soil surface in unmanaged C. japonica plantations despite sparse understory vegetation, due to a high LAI of the overstory, resulting in relatively high infiltration rates and less significant overland flow. In addition, the litter layer might reduce the persistence of overland flow owing to the roughness that the litter layer provides (Onda, 2008). Thus, two conditions should be satisfied for the occurrence of overland flow: a sparse distribution of understory vegetation due to a high LAI of the
2.1. Study area A complete description of the study area was given by Komatsu et al. (2010b). Here we give a brief description of the area. The study area was the Terauchi watershed located in northern Kyushu, Japan (Fig. 1). This watershed has an area of 51 km2 and ranges 120e720 m in elevation. The mean annual precipitation in the watershed during the 1979e2007 period, measured at the Ohshiro station (approximately 460 m in
Fig. 3. (a) Time series of annual precipitation and runoff in the period 1979e2007. (b) Seasonal changes in precipitation and runoff averaged for the period 1979e2007. Vertical bars indicate the standard deviation.
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
1531
Fig. 5. Seasonal changes in the occurrence of (a) daily precipitation P ¼ 100e200 mm, P ¼ 50e60 mm, and P ¼ 10e20 mm and (b) daily runoff Q ¼ 50Me60M, Q ¼ 10Me20M, and Q ¼ 5Me6M during the 1979e2007 period. M is the median of all Q data (i.e., 1.76 mm). The vertical axis indicates the frequency for each month relative to the total frequency.
Fig. 4. Frequency distributions of (a) daily precipitation (P) and (b) daily runoff (Q) during the 1979e2007 period.
elevation), was 2176 mm. Although temperature measurements were not conducted in the watershed, data recorded at the Asakura station located w10 km to the southwest were available from the Japan Meteorological Agency (2010). The mean annual temperature during the 1978e2007 period at the Asakura station was 15.5 C. The geology of the watershed was mainly older-Sangun metamorphic schist, but partly late-Miocene to Pliocenenon alkaline mafic volcanic rock (Geological Survey of Japan, 2007). The forested area occupied 76% of the watershed area (Komatsu et al., 2010b). The remaining 24% comprised other land uses, such as water bodies and orchards. Coniferous plantations occupied 93% of the forested area while natural broadleaved forests and natural pine forests occupied 5%. The major coniferous plantation species were C. japonica and C. obtusa, occupying 61% and 32% of the area, respectively. Thus, although C. japonica was the major species in the watershed, C. obtusa plantations still occupied a large area of the watershed. This species composition is typical for watersheds in Japan (National Astronomical Observatory, 2009). Forestry practices in Japan’s coniferous plantations have decreased (Komatsu et al., 2010b). Forestry practices in the Terauchi watershed have been undertaken mainly by the Fukuoka Water Source Forest Foundation since 1979. The area in which the foundation conducted forestry practices was w1000 ha in 1980, but decreased gradually to w200 ha by 2000 (Kawasaki, 2005). Annual investment in forestry operations was
360 million yen in 1979e1983 but decreased gradually to 57 million yen in 2003e2007. Furthermore, the tree age frequency distribution versus coniferous plantation area in 2003 (Fig. 2) also suggests decreased forestry practices. Frequencies were lower for trees less than 35 years old, higher for trees 35e55 years old, and lower again for trees more than 55 years old. Coniferous plantations in Japan are usually harvested for timber when trees are older than 40 years (Kikuzawa, 1999). If forestry activity had been continuous, the age frequency for trees less than 40 years old would be nearly constant and the age frequency for trees more than 40 years old would decrease with age. Thus, the frequency distribution suggests decreased forestry practices in the Terauchi watershed over recent decades.
Table 1 Slope of linear regression, Spearman’s rho, and Kendall’s tau for the frequency for each daily precipitation (P) class. No significant trends were observed for any P class. P class
Slope
Rho
Tau
P < 1.0 1.0 P < 5.0 5.0 P < 10 10 P < 20 20 P < 30 30 P < 40 40 P < 50 50 P < 60 60 P < 70 70 P < 80 80 P < 90 90 P < 100 100 P < 200 200 P < 300
0.323 0.00345 0.0754 0.135 0.00936 0.0256 0.0670 0.0227 0.0172 0.0241 0.00936 0.0123 0.0246 0.00148
0.236 0.0311 0.130 0.244 0.0493 0.0123 0.306 0.158 0.0862 0.172 0.0345 0.0665 0.175 0.0148
0.118 0.0222 0.0714 0.172 0.0694 0.00482 0.219 0.187 0.147 0.116 0.0951 0.103 0.106 0.0406
1532
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
2.2. Data We obtained precipitation and runoff data for the period 1979e2007 from the Chikugo Regional Bureau, Japan Water Agency. We used daily precipitation (P) data recorded at the Ohshiro station in the Terauchi watershed and daily inflow data recorded at the Terauchi Dam reservoir. The dam reservoir captured all runoff from the Terauchi watershed. Besides precipitation data for the Ohshiro station, we obtained P data for the period 1979e2007 recorded at two other stations in the Terauchi watershed. P at these two stations correlated well with that at the Ohshiro station for the period 1979e2007. Using these stations’ data did not qualitatively alter our results and conclusions.
Table 2 Slope of linear regression, Spearman’s rho, and Kendall’s tau for the frequency for each daily runoff (Q) class. Numerals in bold indicate the trend was significant at the 5% level. Q class
Slope
Rho
Tau
Q < 1Ma 1M Q < 2M 2M Q < 3M 3M Q < 4M 4M Q < 5M 5M Q < 6M 6M Q < 7M 7M Q < 8M 8M Q < 9M 9M Q < 10M 10M Q < 20M 20M Q < 30M 30M Q < 40M 40M Q < 50M 50M Q
1.28 0.516 0.548 0.142 0.0719 0.00640 0.0606 0.0118 0.00690 0.0399 0.0828 0.0128 0.0424 0.0153 0.0276
0.236 0.129 0.327 0.148 0.0616 0.0222 0.232 0.0591 0.0222 0.363 0.101 0.103 0.189 0.0887 0.244
0.165 0.128 0.219 0.216 0.0542 0.0630 0.150 0.0830 0.000910 L0.246 0.144 0.166 0.118 0.145 0.145
a
M is the median of all Q data (i.e., 1.76 mm).
3. Methods of analysis 3.1. Methodology We employed the method developed and used by Archer et al (Archer and Newson, 2002; Archer, 2003, 2007; Archer et al., 2010) to examine changes in the runoff regime with changes in vegetation. We applied the method to detect changes in daily runoff (Q). The method examines changes in the whole frequency distribution of Q data. For this purpose, the method classifies Q data according to the magnitude and examines temporal changes in the data frequency of each Q class. Numerous studies have examined changes in the runoff regime (Jones and Grant, 1996; Beschta et al., 2000; Jones, 2000; Ashagrie et al., 2006; Lin and Wei, 2008). Almost all examined changes in the runoff magnitude (e.g., annual maximum Q), but did not examine changes in the whole frequency distribution of data according to the Q magnitude. However, it is possible that changes in the runoff regime could vary among different Q magnitude classes. Most previous studies did not consider this possibility, and therefore their results are uncertain (Alila et al., 2009, 2010). Alternative methods were proposed by Archer et al (Archer and Newson, 2002; Archer, 2003, 2007; Archer et al., 2010) and by Alila et al. (2009), who considered changes in the whole frequency distribution of data according to the Q magnitude. Archer’s method is more suited to examining changes in the runoff regime for watersheds with gradual vegetation changes, and Alila’s method to changes in runoff with abrupt vegetation changes (e.g., forest clear-felling). We used Archer’s method because we examined changes in the runoff regime with a gradual vegetation change due to decreased forestry practices. 3.2. Procedures
Fig. 6. Time series of frequencies for (a) daily precipitation P ¼ 100e200 mm, (b) P ¼ 50e60 mm, and (c) P ¼ 10e20 mm. Solid lines are regression lines determined using the least-squares method.
First, we examined general characteristics of precipitation and runoff, such as the mean annual precipitation, runoff, evapotranspiration, frequency distributions of P and Q for the whole period, and typical seasonal changes in precipitation and runoff. Second, we examined changes in the precipitation and runoff regime during the period 1979e2007. P and Q data were classified into 14 and 15 classes according to their magnitudes, respectively. We examined changes in the frequencies for each P and Q class during the period 1979e2007 using the statistical method described below. Third, we examined changes in Q considering the relationship between annual precipitation and the frequency for each Q class. Annual precipitation and the frequency for each Q class generally
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
1533
Kundzewicz et al. (2005), Archer (2007), and Komatsu et al. (2010b). Kundzewicz and Robson (2004) stated that this method is the most appropriate for detecting trends in time-series data, because it does not include assumptions about the form of the distribution from which the data are derived. We first calculated the slope of linear regression, Spearman’s rho, and Kendall’s tau from the original time-series data. We then calculated these parameters after resampling 29 values from the original time-series data using the bootstrap method with data replacement, where the original time-series comprised data for a period of 29 years. This process was repeated 1000 times (Note that our conclusions did not change when this process was repeated 5000 and 10,000 times.). From the parameter values obtained by resampling, we produced probability density functions (PDFs) for the three parameters. We finally compared the parameter values based on the original time-series data with the PDFs. If the parameter values obtained from original time-series data were greater than the highest 2.5% threshold values of the PDFs, the increasing trend of the original time series was considered to be significant at the 5% level (Svensson et al., 2005). We used not one but all three parameters (i.e., the slope of linear regression, Spearman’s rho, and Kendall’s tau) for statistical analysis. Results of the statistical test might vary according to the parameter used (Kundzewicz and Robson, 2004). Therefore, if we obtain the same results from statistical tests using different parameters, the results are more reliable than those derived from a statistical test using only one parameter (Kundzewicz and Robson, 2004). 4. Results and discussion Fig. 3a shows time series of annual precipitation and runoff. The mean annual precipitation and runoff during the 1979e2007 period were 2176 and 1351 mm, respectively. Thus, the mean annual evapotranspiration, calculated as the difference between annual precipitation and runoff, was 825 mm. This value is comparable to values reported for other forested catchments and watersheds on Kyushu (Shimizu et al., 2003; Komatsu et al., 2007c, 2008a). Trends in annual precipitation, runoff, and evapotranspiration during the period were not found to be significant (p > 0.05), in agreement with Komatsu et al.’s (2010b) report using the same dataset. They did not find significant trends in annual evapotranspiration during the period 1979e2007. In addition, they found no significant trends in low flow during the same period. Fig. 7. Time series of frequencies for (a) daily runoff Q ¼ 50Me60M, (b) Q ¼ 10Me20M, and (c) Q ¼ 5Me6M, where M is the median of all Q data (i.e., 1.76 mm). Solid lines are regression lines determined using the least-squares method.
correlate linearly (Archer and Newson, 2002; Archer, 2003, 2007; Archer et al., 2010), indicating that interannual variations in the frequency correspond to interannual variations in precipitation. Thus, we determined the linear regression equation for the relationship between annual precipitation and the frequency for each Q class. We calculated the residual of the frequency (i.e., the observed frequency for the Q class minus the frequency predicted for the Q class using the linear regression equation) and examined temporal changes in the residual. This analysis evaluates changes in Q after excluding interannual variations in the frequencies for Q corresponding to interannual variations in precipitation. 3.3. Statistics To detect temporal trends, we used a non-parametric method described by Kundzewicz and Robson (2004) and used by
Table 3 Correlation between annual precipitation and the data frequency for different daily runoff (Q) classes. The linear slope, intercept, and correlation coefficient (R) are given. Numerals in bold indicate the correlation was positively significant at the 5% level according to Pearson’s correlation coefficient test. Q class
Slope
Intercept
R
Q < 1Ma 1M Q < 2M 2M Q < 3M 3M Q < 4M 4M Q < 5M 5M Q < 6M 6M Q < 7M 7M Q < 8M 8M Q < 9M 9M Q < 10M 10M Q < 20M 20M Q < 30M 30M Q < 40M 40M Q < 50M 50M Q
0.0757 0.0114 0.0173 0.0101 0.00818 0.00335 0.00406 0.00141 0.00197 0.00131 0.0101 0.00227 0.00234 0.000789 0.00102
348 74.6 4.07 6.88 9.56 2.08 4.80 0.728 2.60 1.60 14.8 2.96 4.13 1.03 1.71
0.729 0.234 0.524 0.703 0.679 0.470 0.680 0.398 0.576 0.530 0.865 0.580 0.704 0.419 0.529
a
M is the median of all Q data (i.e., 1.76 mm).
1534
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
Fig. 3b shows seasonal changes in precipitation and runoff averaged for the period 1979e2007. Precipitation and runoff were higher in June and July than in other months. Such seasonality is typical for most regions in Japan (Komatsu et al., 2007a, 2008a, 2010a). Fig. 4a and b show the frequency distributions for all data according to P and Q, respectively. The maximum P was 285 mm, which was recorded on July 2, 2007. When approximating the frequency distribution of annual maximum P data for the period 1979e2007 using the logarithmic normal distribution, the return period for the P value was calculated as 134 years. Q ranged between 0.0678 and 112 mm. The median value (M) for Q was 1.76 mm. High P values were generally recorded during summer, although low P values were recorded throughout the year (Fig. 5a). Consequently, high Q values were generally recorded during summer (Fig. 5b). Table 1 presents the results of statistical analysis for detecting trends in the precipitation pattern. Time series of the frequencies for three P classes are shown in Fig. 6 as examples. No significant trend was observed for any class (Table 1). Table 2 presents the results of statistical analysis for detecting trends in the runoff pattern. Time series of the frequencies for three Q classes are shown in Fig. 7 as examples. No significant increasing trend was observed for any class (Table 2). For one class, a decreasing trend was observed according to the tau value. Table 3 presents the results of correlation analysis for the relationship between annual precipitation and the frequency of each Q class. The correlation was significantly positive in most cases. Table 4 presents the results of statistical analysis for detecting trends in residuals of Q data frequencies. Time series of the frequencies for three Q classes are shown in Fig. 8 as examples. We did not observe significant trends in most classes (Table 4). In the two classes Q ¼ 4Me5M and Q ¼ 6Me7M, we observed significant increasing trends. The numbers of data included in these classes were 239 and 68 during the period 1979e2007 respectively. (Fig. 4b) This indicates that the data for these classes were commonly recorded at the watershed. Thus, the increase in frequencies for these classes indicates an increase in the frequency of moderate peak flow. On the other hand, we did not observe increasing trends for classes Q > 7M. This suggests no clear increase in the frequency of high peak flow in this watershed. The above results agree with those reported by Hirano et al. (2009), which were based on runoff measurements at a smallcatchment scale (w1 ha). Hirano et al. (2009) examined the runoff
regime during storms for two catchments; one was covered with C. obtusa plantation that was not thinned and the other was covered with natural broadleaved forest. They reported a distinct difference in peak flow intensity for a relatively small storm (Q was w10 mm according to Fig. 5b of their paper), and no clear difference in peak flow intensity for a large storm (Q was w50 mm according to Fig. 5c of their paper). One possible explanation for these results is as follows. As the soil layer is not generally saturated during small storms, the infiltration rate of the soil layer could affect the occurrence of overland flow and therefore catchment runoff. Indeed, Hirano et al. (2009) recorded a higher magnitude of overland flow for the coniferous catchment than for the broadleaved
Table 4 Slope of linear regression, Spearman’s rho, and Kendall’s tau for the residual (i.e., observed minus predicted) of the frequency for each daily runoff (Q) class. Numerals in bold indicate the trend was significant at the 5% level. Q class
Slope
Rho
Tau
Q < 1Ma 1M Q < 2M 2M Q < 3M 3M Q < 4M 4M Q < 5M 5M Q < 6M 6M Q < 7M 7M Q < 8M 8M Q < 9M 9M Q < 10M 10M Q < 20M 20M Q < 30M 30M Q < 40M 40M Q < 50M 50M Q
0.479 0.396 0.365 0.0351 0.158 0.0417 0.103 0.00307 0.0139 0.0261 0.0239 0.0367 0.0177 0.00695 0.0168
0.0443 0.0985 0.255 0.0591 0.371 0.0640 0.479 0.0197 0.0246 0.254 0.0394 0.229 0.0837 0.0640 0.177
0.0640 0.113 0.177 0.0616 0.266 0.106 0.330 0.0320 0.0635 0.187 0.0424 0.158 0.131 0.101 0.103
a
M is the median of all Q data (i.e., 1.76 mm).
Fig. 8. Time series of the residuals (observed minus predicted) for (a) daily runoff Q ¼ 50Me60M, (b) Q ¼ 10Me20M, and (c) Q ¼ 5Me6M, where M is the median of all Q data (i.e., 1.76 mm). Solid lines are regression lines determined using the least-squares method.
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
catchment during small storms. On the other hand, the soil layer is generally saturated during large storms. Overland flow would occur owing to return flow from the saturated soil layer, even when the infiltration rate is high (Gomi et al., 2008a). Thus, the influence of the infiltration rate of the soil layer on the magnitude of overland flow is less significant under these conditions. This could result in less considerable difference in peak flow between the two catchments during large storms. Our results also agree with those reported by Maita and Suzuki (2008), who examined changes in Q with forest clearcutting in Japan at a small catchment scale (w1 ha). They reported a significant increase in peak flow intensity during moderate storms (Q was w10 mm), but no significant increase in peak flow intensity during large storms (Q was w50 mm). 5. Conclusions As a countermeasure to increased flood risk due to decreased forestry practices, many local governments in Japan have introduced local taxes to stimulate forestry practices in coniferous plantation areas (Imawaka and Sato, 2008). However, we did not observe increases in frequencies of high peak flow with decreased forestry practices. Therefore, the increase in flood risk might be less than expected. This study is the first to quantitatively assess the increase in flood risk with decreased forestry practices for coniferous plantations in Japan on the basis of watershed runoff data. Considering that there have been many studies reporting low infiltration rates of the soil surface in C. obtusa plantations with decreased forestry practices (Miyata et al., 2007; Gomi et al., 2008a,b; Hirano et al., 2009), further studies are required to generalize our results. If the results of these future studies support ours, then countermeasures for decreased forestry practices might be unnecessary from the viewpoint of flood control. If different results are obtained among regions, current forest policies should be modified to accommodate the risk for each region. Additionally, we recommend studies examining changes in peak flow due to decreased forestry practices using watershed runoff data with high time resolution (e.g., hourly runoff data). Since we could not obtain data with high time resolution, we used daily watershed runoff data. However, previous studies reported that high-intensity overland flow occurs during short-term highintensity storms (Gomi et al., 2008a, 2010b; Miyata et al., 2010). This implies that changes in peak flow due to decreased forestry practices might be more apparent when using data with high time resolution. This paper focused on the relationship between forestry practices and flood risk. Researchers in Japan (Ohgaki, 2005; Onda, 2008) have pointed out other risks due to decreased forestry practices from the viewpoint of soil protection and water quality. Soil transport and water quality are related to the peak-flow regime (e.g., Fukuyama et al., 2005; Ide et al., 2007; Onda, 2008). Thus, the results of this study have implications for research of these risks. Acknowledgments We acknowledge Mr. K. Takeda and Mr. F. Yamamoto from the Japan Water Agency for providing precipitation and runoff data for the Terauchi watershed. We also thank Mr. O. Funakoshi and Ms. N. Ogawa from Fukuoka Prefecture for providing vegetation data for the Terauchi watershed. We are grateful to Dr. A. Kawasaki, Dr. T. Kajisa, and Dr. Y. Miyazawa from Kyushu University and Ms. A. Miyazawa from Fukuoka Prefecture for their assistance in collating and analyzing vegetation data. Thanks are also due to Mr. D. Notomi from Kyushu University for producing the topographical map. We
1535
acknowledge three anonymous reviewers for making critical comments. This research was supported by Core Research for Evolution Science and Technology of the Japan Science and Technology Agency. References Alila, Y., Kuras, P.K., Schnorbus, M., Hudson, R., 2009. Forests and floods: a new paradigm sheds light on age-old controversies. Water Resources Research 45, W08416. Alila, Y., Hudson, R., Kuras, P.K., Schnorbus, M., Rasouli, K., 2010. Reply to comment by Jack Lewis et al. on “forests and floods: a new paradigm sheds light on age old controversies”. Water Resources Research 46, W05802. Archer, D., 2003. Scale effects on the hydrological impact of upland afforestation and drainage using indices of flow variability: the River Irthing, England. Hydrology and Earth System Sciences 7, 325e338. Archer, D.R., 2007. The use of flow variability analysis to assess the impact of land use change on the paired Plynlimon catchments, mid-Wales. Journal of Hydrology 347, 487e496. Archer, D.R., Newson, M.D., 2002. The use of indices of flow variability in assessing the hydrological and instream habitat impacts of upland afforestation and drainage. Journal of Hydrology 268, 244e258. Archer, D.R., Climent-Soler, D., Holman, I.P., 2010. Changes in discharge rise and fall rates applied to impact assessment of catchment land use. Hydrology Research 41, 13e26. Ashagrie, A.G., de Laat, P.J.M., de Wit, M.J.M., Tu, M., Uhlenbrook, S., 2006. Detecting the influence of land use changes on discharges and floods in the Meuse River Basinethe predictive power of a ninety-year rainfall-runoff relation? Hydrology and Earth System Sciences 10, 691e701. Beschta, R.L., Pyles, M.R., Skaugset, A.E., Surfleet, C.G., 2000. Peakflow responses to forest practices in the western cascades of Oregon, USA. Journal of Hydrology 233, 102e120. Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. Journal of Hydrology 55, 3e23. Brown, A., Zhang, L., McMahon, T.A., Western, A.W., Vertessy, R.A., 2005. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. Journal of Hydrology 310, 28e61. Calder, I.R., 2005. Blue Revolution. Earthscan Publications, London. Duley, F.L., Ackerman, F.G., 1980. Run-off and erosion from plots of different length. Journal of Agricultural Research 48, 505e510. Farley, K.A., Jobbagy, E.G., Jackson, R.B., 2005. Effects of afforestation on water yield: a global synthesis with implications for policy. Global Change Biology 11, 1565e1576. Fukada, H., Watanabe, N., Kajihara, N., Tsukamoto, J., 2005. Altitudinal zoning of understory vegetation in planted Hinoki cypress (Chamaecyparis obtusa Endl.) forest from the view of surface soil erosion control. Japan Journal of Forest Environment 47, 77e84. Fukada, H., Watanabe, N., Kajihara, N., Tsukamoto, J., 2006. Dynamics of under growth and its application to vegetation control of planted Chamaecyparis obtusa Endl. forests with special reference to mitigation of surface soil loss. Journal of the Japanese Forest Society 88, 231e239. Fukuyama, T., Takenaka, C., Onda, Y., 2005. Cs-137 loss via soil erosion from a mountainous headwater catchment in central Japan. Science of the Total Environment 350, 238e247. Fujimori, T., 2000. Living with Forests. Maruzen, Tokyo. Geological Survey of Japan, 2007. Seamless Digital Geological Map of Japan. http:// riodb02.ibase.aist.go.jp/db084/index_e.html. Gomi, T., Sidle, R.C., Ueno, M., Miyata, S., Kosugi, K., 2008a. Characteristics of overland flow generation on steep forested hillslopes of central Japan. Journal of Hydrology 361, 275e290. Gomi, T., Sidle, R.C., Miyata, S., Kosugi, K., Onda, Y., 2008b. Dynamic runoff connectivity of overland flow on steep forested hillslopes: scale effects and runoff transfer. Water Resources Research 44, W08411. Gomi, T., Miyata, S., Onda, Y., 2010a. Overland flow and the rainfall-runoff relationship in a catchment covered with Japanese cedar plantations. Suiri-kagaku 311, 77e94. Gomi, T., Asano, Y., Uchida, T., Onda, Y., Sidle, R.C., Miyata, S., Kosugi, K., Mizugaki, S., Fukuyama, T., Fukushima, T., 2010b. Evaluation of storm runoff pathways in steep nested catchments draining a Japanese cypress forest in central Japan: a geochemical approach. Hydrological Processes 24, 550e566. Grant, G.E., Lewis, S.L., Swanson, F.J., Cissel, J.H., McDonnell, J.J., 2008. Effects of Forest Practices on Peak Flows and Consequent Channel Response: a State-ofscience Report for Western Oregon and Washington. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland. Hattori, S., Abe, T., Kobayashi, T., Tamai, K., 1992. Effect of forest floor coverage on reduction of soil erosion in Hinoki plantations. Bulletin of Forestry and Forest Products Research Institute 362, 1e34. Hirano, T., Terajima, T., Nakamura, T., Sakai, M., Aoki, F., Nanami, A., 2009. The differences in the short-term runoff characteristics between the coniferous catchment and the deciduous catchmentethe effects of storm size on stormflow generation processes of small forested catchments. Journal of the Japan Society of Hydrology & Water Resources 22, 24e39.
1536
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
Hiraoka, M., Onda, Y., Kato, H., Mizugaki, S., Gomi, T., Nanko, K., 2010. Effects of understory vegetation on infiltration capacity in Japanese cypress plantation. Journal of the Japanese Forest Society 92, 145e150. Ichikawa, T., Takahashi, T., Asano, Y., 2003. Effects of the conversion of forest management type from natural deciduous broad-leaved forests to artificial Japanese cypress (Chamaecyparis obtusa) and Japanese cedar (Cryptomeria japonica) forests on nutrient dynamics. Japan Journal of Forest Environment 45, 35e42. Ichikawa, T., Okabe, N., Takahashi, T., Asano, Y., 2005. Comparison of soil microbial biomass and activity among site positions on a slope planted with Japanese cypress and cedar forests. Japan Journal of Forest Environment 47, 21e27. Ichikawa, T., Takahashi, T., Asano, Y., 2006. Comparison of changes in organic matter dynamics due to stand age between artificial Japanese cedar (Cryptomeria japonica D. Don) forests and Japanese cypress (Chamaecyparis obtusa Sieb. et Zucc.) forests. Journal of the Japanese Forest Society 88, 525e533. Ide, J., Nagafuchi, O., Chiwa, M., Kume, A., Otsuki, K., Ogawa, S., 2007. Effects of discharge level on the load of dissolved and particulate components of stream nitrogen and phosphorus from a small afforested watershed of Japanese cypress (Chamaecyparis obtusa). Journal of Forest Research 12, 45e56. Imawaka, S., Sato, N., 2008. Study on new forest maintenance projects by "Forest Environmental Tax". Bulletin of Kyushu University Forest 89, 75e126. Japan Forestry Agency, 2007. Forest White Paper. Japan Forestry Agency, Tokyo. Japan Meteorological Agency, 2010. Climate Statistics. http://www.jma.go.jp/jma/ index.html. Joel, A., Nessing, I., Seguel, O., Casanova, M., 2002. Measurement of surface water runoff from plots of two different sizes. Hydrological Processes 16, 1467e1478. Jones, J.A., 2000. Hydrological processes and peak discharge response to forest removal, regrowth, and roads in 10 small experimental basins, western Cascades, Oregon. Water Resources Research 36, 2621e2642. Jones, J.A., Grant, G.E., 1996. Peak flow responses to clear-cutting and roads in small and large basins, western Cascades, Oregon. Water Resources Research 32, 959e974. Kato, H., Onda, Y., Ito, S., Nanko, K., 2008. Field measurement of infiltration rate using a oscillating nozzle rainfall simulator in devastated Hinoki plantation. Journal of the Japan Society of Hydrology and Water Resources 21, 439e448. Kawasaki, A., 2005. Corporation Between Upstream and Downstream Regions for Forestry Practices. Bachelor thesis, Kyushu University. Kikuzawa, K., 1999. Forest Ecology. Kyoritsu Shuppan, Tokyo. Kiyono, Y., 1988. Dynamics and control of understories in Chamaecyparis obtusa plantations. Bulletin of Forestry and Forest Products Research Institute 359, 1e122. Komatsu, H., Tanaka, N., Kume, T., 2007a. Do coniferous forests evaporate more water than broad-leaved forests in Japan? Journal of Hydrology 336, 361e375. Komatsu, H., Kume, T., Otsuki, K., 2007b. Contemporary role of catchment water balance data for forest evapotranspiration research. Journal of the Japanese Forest Society 89, 345e358. Komatsu, H., Ide, J., Shinohara, Y., Haga, H., Fujiyama, Y., Miyano, T., Maruno, R., Chiwa, M., Kume, T., Higashi, N., Otsuki, K., 2007c. Evapotranspiration from unmanaged coniferous plantations. Suiri-kagaku 297, 107e127. Komatsu, H., Maita, E., Otsuki, K., 2008a. A model to estimate annual forest evapotranspiration in Japan from mean annual temperature. Journal of Hydrology 348, 330e340. Komatsu, H., Kume, T., Otsuki, K., 2008b. The effect of converting a native broadleaved forest to a coniferous plantation forest on annual water yield: a pairedcatchment study in northern Japan. Forest Ecology and Management 255, 880e886. Komatsu, H., Kume, T., Otsuki, K., 2009a. Effects of coniferous plantation thinning on annual wet-canopy evaporation: model verification. Journal of the Japanese Forest Society 91, 94e103. Komatsu, H., Kume, T., Otsuki, K., 2009b. Low flow changes in runoff with converting a coniferous plantation to a broad-leaved forest in the TatsunokuchiMinami catchment, Japan. Ecohydrology 2, 164e172. Komatsu, H., Kume, T., Otsuki, K., 2010a. Water resource management in Japaneforest management or dam reservoirs? Journal of Environmental Management 91, 814e823. Komatsu, H., Kume, T., Shinohara, Y., Miyazawa, Y., Otsuki, K., 2010b. Did annual runoff and low flow decrease with reduced forestry practices in Japan? Hydrological Processes 24, 2440e2451.
Kundzewicz, Z.W., Robson, A.J., 2004. Change detection in hydrological recordsda review of the methodology. Hydrological Sciences Journal 49, 7e19. Kundzewicz, Z.W., Graczyk, D., Maurer, T., Pinskwar, I., Radziejewski, M., Svensson, C., Szwed, M., 2005. Trend detection in river flow series: 1. Annual maximum flow. Hydrological Sciences Journal 50, 797e810. Kuraji, K., 2007. Expectations from society for forest hydrological sciencesetowards establishment of forest sociohydrology. In: The Editing Committee of Forest Hydrology. Forest Hydrology. Tsukiji-shokan, Tokyo, pp. 283e308. Kuraji, K., Hoyano, H., 2004. Green Dams. Tsukiji Shokan, Tokyo. Lin, Y., Wei, X., 2008. The impact of large-scale forest harvesting on hydrology in the Willow watershed of Central British Columbia. Journal of Hydrology 359, 141e149. Maita, E., Suzuki, M., 2008. The effect of forest cutting on flow duration curve of mountainous small watershed: evaluation of the influence of vegetation on inter-watershed variation in flow duration curve. Journal of the Japanese Forest Society 90, 36e45. Miura, S., 2000. Proposal for a new definition to evaluate the status of forest floor cover and floor cover percentage (FCP) from the viewpoint of the protection against raindrop splash. Journal of the Japanese Forestry Society 82, 132e140. Miyata, S., Kosugi, K., Gomi, T., Onda, Y., Mizuyama, T., 2007. Surface runoff as affected by soil water repellency in a Japanese cypress forest. Hydrological Processes 21, 2365e2376. Miyata, S., Kosugi, K., Nishi, Y., Gomi, T., Sidle, R.C., Mizuyama, T., 2010. Spatial pattern of infiltration rate and its effect on hydrological processes in a small headwater catchment. Hydrological Processes 24, 535e549. National Astronomical Observatory, 2009. Chronological Environmental Tables 2009/2010. Maruzen, Tokyo. Ohgaki, S., 2005. Nutrients in River Water. Gihodo, Tokyo. Onda, Y., 2008. Water and Sediment Discharge in Abandoned Plantations. Iwanami, Tokyo. Onda, Y., Yukawa, N., 1995. The influences of understories on the infiltration capacities of Chamaecyparis obtusa plantations (II) Laboratory experiments. Journal of the Japanese Forestry Society 77, 399e407. Onda, Y., Gomi, T., Mizugaki, S., Nonoda, T., Sidle, R.C., 2010. An overview of the field and modelling studies on the effects of forest devastation on flooding and environmental issues. Hydrological Processes 24, 527e534. Parsons, A.J., Brazier, R.E., Wainwright, J., Powell, D.M., 2006. Scale relationships in hillslope runoff and erosion. Earth Surface Processes and Landforms 31, 1381e1393. Shimizu, A., Shimizu, T., Miyabuchi, Y., Ogawa, Y., 2003. Evapotranspiration and runoff in a forest watershed, western Japan. Hydrological Processes 17, 3125e3149. Shinohara, Y., Kumagai, T., Otsuki, K., Kume, A., Wada, N., 2009. Impact of climate change on runoff from a mid-latitude mountainous catchment in central Japan. Hydrological Processes 23, 1418e1429. Sidle, R.C., Hirano, T., Gomi, T., Terajima, T., 2007. Hortonian overland flow from Japanese forest ptantationsan aberration, the real thing, or something in between? Hydrological Processes 21, 3237e3247. Svensson, C., Kundzewics, Z.W., Maurer, T., 2005. Trend detection in river flow series: 2. Flood and low-flow index series. Hydrological Sciences Journal 50, 811e824. Tsujimura, M., Onda, Y., Harada, D., 2006. The role of Horton overland flow in rainfall-runoff process in an unchanneled catchment covered by unmanaged Hinoki plantation. Journal of the Japan Society of Hydrology & Water Resources 19, 17e24. Wakiyama, Y., Onda, Y., Nanko, K., Mizugaki, S., Kim, Y., Kitahara, H., Ono, H., 2010. The estimation of temporal variation in splash detachment in two Japanese cypress plantations of contrasting age. Earth Surface Processes and Landforms 35, 993e1005. Ward, R.C., Robinson, M., 2000. Principles of Hydrology. McGraw-Hill, London. Yao, H.X., Hashino, M., Xia, J., Chen, X.H., 2009. Runoff reduction by forest growth in Hiji River basin, Japan. Hydrological Sciences Journal 54, 556e570. Yue, S., Hashino, M., 2005. Statistical interpretation of the impact of forest growth on streamflow of the Sameura basin, Japan. Environmental Monitoring and Assessment 104, 369e384. Yukawa, N., Onda, Y., 1995. The influences of understories on the infiltration capacities of Chamaecyparis obtusa plantations (I) experimental results using mist type rainfall simulator. Journal of the Japan Forestry Society 77, 224e231.
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Changes in peak flow with decreased forestry practices: Analysis using watershed runoff data Hikaru Komatsu a, b, *, Yoshinori Shinohara a, Tomonori Kume b, Kyoichi Otsuki a a b
Kasuya Research Forest, Kyushu University, 394 Tsubakuro, Sasaguri, Kasuya, Fukuoka 811-2415, Japan School of Forestry and Resource Conservation, National Taiwan University, 1, Sce. 4, Roosevelt Road, Taipei 106, Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history: Received 22 April 2010 Received in revised form 3 December 2010 Accepted 5 January 2011 Available online 1 February 2011
The prevalence of forestry practices such as thinning and pruning have gradually decreased since the 1980s. Researchers have noted an increased flood risk with decreased forestry practices for coniferous plantations in Japan on the basis of infiltration and overland flow measurements at a plot scale (typically several square meters). However, no studies have examined changes in peak flow with decreased forestry practices at a watershed scale (typically several tens or hundreds of square kilometers) even though flood disasters generally occur at this scale in Japan. We examined changes in frequency distributions of daily precipitation (P) and runoff (Q) during the period 1979e2007 at the Terauchi watershed, where forestry practices are known to have decreased. For this purpose, we divided P and Q data into 14 and 15 classes according to the magnitude, respectively, and examined changes in the frequency for each class during the period. We observed no significant increasing trend for any P or Q class. Even when taking into account the effect of interannual variations in precipitation on the frequency for each Q class, there was no significant increasing trend in the frequencies except for two Q classes with moderate Q values. These results suggest that the increase in flood risk due to decreased forestry practices might be less than expected. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: Flood Forestry practices Peak flow Plantation Runoff Watershed
1. Introduction Watershed runoff is influenced by many factors, such as climate, geology, topography, and vegetation (Ward and Robinson, 2000; Komatsu et al., 2008a; Shinohara et al., 2009). Human activities that influence vegetation cover could affect the runoff regime (e.g., Yue and Hashino, 2005; Komatsu et al., 2008b, 2009b; Yao et al., 2009), which in turn could alter flood and drought risks. Forestry clear-felling (or conversion from forests to pasture) and development of forest cover following clear-felling (or conversion from pasture to forests) are the most significant examples of vegetation change. Many studies (e.g., Bosch and Hewlett, 1982; Brown et al., 2005; Farley et al., 2005; Grant et al., 2008) have examined the effect of clear-felling or development of forest cover on the runoff regime. Changes in the runoff regime with clear-felling and development of forest cover are not critical issues in Japan, because such land-use changes have recently been uncommon, and therefore,
* Corresponding author. Kasuya Research Forest, Kyushu University, 394 Tsubakuro, Sasaguri, Kasuya, Fukuoka 811-2415, Japan. Tel.: þ81 92 948 3109; fax: þ81 92 948 3119. E-mail address: [email protected] (H. Komatsu). 0301-4797/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2011.01.010
the percentage of forested area has been nearly constant over the past 40 years (National Astronomical Observatory, 2009). Instead, forestry practices and their effect on runoff have recently been raising as a concern in Japan (e.g., Kuraji and Hoyano, 2004; Calder, 2005; Komatsu et al., 2007b,c, 2009a; Onda, 2008; Onda et al., 2010). Japan developed large areas of coniferous plantations for timber production from the 1950s to 1970s, mainly by converting natural broadleaved forests (Fujimori, 2000). Coniferous plantations occupy approximately 40% of forested areas and approximately 25% of Japan’s land surface (National Astronomical Observatory, 2009). To develop these plantations, seedlings are planted typically at approximately 3000 stems ha1. Trees are thinned and pruned several times in the decades after planting and are harvested at an age of 40e50 years (e.g., Kikuzawa, 1999). By this stage, the stem density is typically 500 stems ha1. Forestry practices such as thinning and pruning were conducted in these plantations during the 1950e1980 period. However, the prevalence of these practices has gradually decreased since the 1980s, because of low timber prices and increased labor costs (Fujimori, 2000; Japan Forestry Agency, 2007; Onda, 2008). Thus, many plantations in Japan have become unmanaged. Researchers (Kuraji and Hoyano, 2004; Kuraji, 2007; Onda, 2008; Onda et al., 2010) have noted increased flood risk with
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
decreased forestry practices in coniferous plantations in Japan. The leaf area index (LAI) of unmanaged coniferous plantations is high owing to the absence of forestry practices such as thinning and pruning, which results in the sparse distribution of understory vegetation (Kiyono, 1988; Fukada et al., 2005, 2006; Japan Forestry Agency, 2007). When understory vegetation is sparse, throughfall strikes the ground surface directly, resulting in compaction of the ground surface and therefore lower rates of infiltration of the soil surface and greater overland flow (e.g., Yukawa and Onda, 1995; Onda and Yukawa, 1995; Gomi et al., 2008a; Kato et al., 2008; Hirano et al., 2009; Hiraoka et al., 2010). The infiltration rate for unmanaged coniferous plantations has been found to be lower than that for bare soil in some cases (Onda, 2008). This contrasts to having dense understory vegetation, in which case the infiltration
1529
rates for the ground surface are relatively high. In addition, the presence of understory vegetation reduces the persistence of overland flow even when flow does occur, owing to the roughness that the vegetation provides (Gomi et al., 2008a,b, 2010a; Wakiyama et al., 2010). Researchers (Kuraji and Hoyano, 2004; Kuraji, 2007; Onda, 2008; Onda et al., 2010) speculated that these results imply increased flood risk in watersheds where coniferous plantations are unmanaged. On the basis of this speculation, many local governments in Japan have introduced local taxes to stimulate forestry practices (Imawaka and Sato, 2008). However, the increase in flood risk might be less than expected for two reasons. First, the speculation is mainly based on plot-scale measurements (typically within several square meters) of infiltration and overland flow (e.g., Hattori et al., 1992; Tsujimura et al.,
Fig. 1. (a) Location of the study area and (b) topographical map of the area.
1530
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
overstory, and high decomposition rates of litters. Although unmanaged C. obtusa plantations satisfy both these conditions, unmanaged C. japonica plantations only satisfy the former condition. Most watersheds in Japan include large areas of C. japonica plantations as well as large areas of C. obtusa plantations (National Astronomical Observatory, 2009). Thus, the increase in flood risk might not be considerable because of the species composition. No studies have examined changes in flood risk with decreased forestry practices in coniferous plantations in Japan at a watershed scale. Flood disasters generally occur at this scale in Japan, and the necessity of studies at spatial scales larger than a plot scale has been pointed out (Kuraji and Hoyano, 2004; Onda, 2008). Thus, we examine changes in the runoff regime using daily runoff data obtained for a watershed where forestry practices are known to have decreased.
2. Study area and data Fig. 2. Frequency distribution of coniferous plantation tree ages for the Terauchi watershed. The vertical axis indicates the area of each age class divided by the whole area of coniferous plantations.
2006; Miyata et al., 2007; Gomi et al., 2008a,b; Hirano et al., 2009). Recent studies in Japan (e.g., Sidle et al., 2007; Gomi et al., 2010a,b) performed overland flow measurements in many plots of various size (several square meters to several tens of square meters). They reported distinct differences in overland flow occurrence between unmanaged and managed coniferous plantations and between unmanaged coniferous plantations and natural broadleaved forests in small plots. However, the differences were less significant in large plots. This phenomenon is referred to as the scale effect (Gomi et al., 2008b, 2010a,b) and has also been reported in other countries (e.g., Duley and Ackerman, 1980; Joel et al., 2002; Parsons et al., 2006). Furthermore, Hirano et al. (2009) examined differences in runoff between unmanaged coniferous plantations and natural broadleaved forests at a small-catchment scale (w1 ha) and reported no clear differences in catchment runoff during large storms, although they reported higher runoff for the unmanaged coniferous plantations during small storms. These results imply that flood risk does not considerably increase at a watershed scale (typically several tens or hundreds of square kilometers) in the case of decreased forestry practices because of the scale effect even though flood disasters generally occur at this scale in Japan (e.g., Sidle et al., 2007; Gomi et al., 2010a,b). Second, plantations in Japan mainly comprise Cryptomeria japonica and Chamaecyparis obtusa. These two species account for approximately 70% of the plantation area in Japan (National Astronomical Observatory, 2009). Although many studies have reported low infiltration rates and significant overland flow for unmanaged C. obtusa plantations (e.g., Miyata et al., 2007; Sidle et al., 2007; Gomi et al., 2008a,b, 2010b), no studies have reported the same results for unmanaged C. japonica plantations. Litters of C. japonica have lower decomposition rates than those of C. obtusa (Miura, 2000; Ichikawa et al., 2005, 2006). Thus, the soil surface of C. japonica plantations is generally covered with a litter layer, which contrasts with the soil surface of C. obtusa plantations (Miura, 2000; Ichikawa et al., 2003). This suggests that throughfall would not directly strike the soil surface in unmanaged C. japonica plantations despite sparse understory vegetation, due to a high LAI of the overstory, resulting in relatively high infiltration rates and less significant overland flow. In addition, the litter layer might reduce the persistence of overland flow owing to the roughness that the litter layer provides (Onda, 2008). Thus, two conditions should be satisfied for the occurrence of overland flow: a sparse distribution of understory vegetation due to a high LAI of the
2.1. Study area A complete description of the study area was given by Komatsu et al. (2010b). Here we give a brief description of the area. The study area was the Terauchi watershed located in northern Kyushu, Japan (Fig. 1). This watershed has an area of 51 km2 and ranges 120e720 m in elevation. The mean annual precipitation in the watershed during the 1979e2007 period, measured at the Ohshiro station (approximately 460 m in
Fig. 3. (a) Time series of annual precipitation and runoff in the period 1979e2007. (b) Seasonal changes in precipitation and runoff averaged for the period 1979e2007. Vertical bars indicate the standard deviation.
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
1531
Fig. 5. Seasonal changes in the occurrence of (a) daily precipitation P ¼ 100e200 mm, P ¼ 50e60 mm, and P ¼ 10e20 mm and (b) daily runoff Q ¼ 50Me60M, Q ¼ 10Me20M, and Q ¼ 5Me6M during the 1979e2007 period. M is the median of all Q data (i.e., 1.76 mm). The vertical axis indicates the frequency for each month relative to the total frequency.
Fig. 4. Frequency distributions of (a) daily precipitation (P) and (b) daily runoff (Q) during the 1979e2007 period.
elevation), was 2176 mm. Although temperature measurements were not conducted in the watershed, data recorded at the Asakura station located w10 km to the southwest were available from the Japan Meteorological Agency (2010). The mean annual temperature during the 1978e2007 period at the Asakura station was 15.5 C. The geology of the watershed was mainly older-Sangun metamorphic schist, but partly late-Miocene to Pliocenenon alkaline mafic volcanic rock (Geological Survey of Japan, 2007). The forested area occupied 76% of the watershed area (Komatsu et al., 2010b). The remaining 24% comprised other land uses, such as water bodies and orchards. Coniferous plantations occupied 93% of the forested area while natural broadleaved forests and natural pine forests occupied 5%. The major coniferous plantation species were C. japonica and C. obtusa, occupying 61% and 32% of the area, respectively. Thus, although C. japonica was the major species in the watershed, C. obtusa plantations still occupied a large area of the watershed. This species composition is typical for watersheds in Japan (National Astronomical Observatory, 2009). Forestry practices in Japan’s coniferous plantations have decreased (Komatsu et al., 2010b). Forestry practices in the Terauchi watershed have been undertaken mainly by the Fukuoka Water Source Forest Foundation since 1979. The area in which the foundation conducted forestry practices was w1000 ha in 1980, but decreased gradually to w200 ha by 2000 (Kawasaki, 2005). Annual investment in forestry operations was
360 million yen in 1979e1983 but decreased gradually to 57 million yen in 2003e2007. Furthermore, the tree age frequency distribution versus coniferous plantation area in 2003 (Fig. 2) also suggests decreased forestry practices. Frequencies were lower for trees less than 35 years old, higher for trees 35e55 years old, and lower again for trees more than 55 years old. Coniferous plantations in Japan are usually harvested for timber when trees are older than 40 years (Kikuzawa, 1999). If forestry activity had been continuous, the age frequency for trees less than 40 years old would be nearly constant and the age frequency for trees more than 40 years old would decrease with age. Thus, the frequency distribution suggests decreased forestry practices in the Terauchi watershed over recent decades.
Table 1 Slope of linear regression, Spearman’s rho, and Kendall’s tau for the frequency for each daily precipitation (P) class. No significant trends were observed for any P class. P class
Slope
Rho
Tau
P < 1.0 1.0 P < 5.0 5.0 P < 10 10 P < 20 20 P < 30 30 P < 40 40 P < 50 50 P < 60 60 P < 70 70 P < 80 80 P < 90 90 P < 100 100 P < 200 200 P < 300
0.323 0.00345 0.0754 0.135 0.00936 0.0256 0.0670 0.0227 0.0172 0.0241 0.00936 0.0123 0.0246 0.00148
0.236 0.0311 0.130 0.244 0.0493 0.0123 0.306 0.158 0.0862 0.172 0.0345 0.0665 0.175 0.0148
0.118 0.0222 0.0714 0.172 0.0694 0.00482 0.219 0.187 0.147 0.116 0.0951 0.103 0.106 0.0406
1532
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
2.2. Data We obtained precipitation and runoff data for the period 1979e2007 from the Chikugo Regional Bureau, Japan Water Agency. We used daily precipitation (P) data recorded at the Ohshiro station in the Terauchi watershed and daily inflow data recorded at the Terauchi Dam reservoir. The dam reservoir captured all runoff from the Terauchi watershed. Besides precipitation data for the Ohshiro station, we obtained P data for the period 1979e2007 recorded at two other stations in the Terauchi watershed. P at these two stations correlated well with that at the Ohshiro station for the period 1979e2007. Using these stations’ data did not qualitatively alter our results and conclusions.
Table 2 Slope of linear regression, Spearman’s rho, and Kendall’s tau for the frequency for each daily runoff (Q) class. Numerals in bold indicate the trend was significant at the 5% level. Q class
Slope
Rho
Tau
Q < 1Ma 1M Q < 2M 2M Q < 3M 3M Q < 4M 4M Q < 5M 5M Q < 6M 6M Q < 7M 7M Q < 8M 8M Q < 9M 9M Q < 10M 10M Q < 20M 20M Q < 30M 30M Q < 40M 40M Q < 50M 50M Q
1.28 0.516 0.548 0.142 0.0719 0.00640 0.0606 0.0118 0.00690 0.0399 0.0828 0.0128 0.0424 0.0153 0.0276
0.236 0.129 0.327 0.148 0.0616 0.0222 0.232 0.0591 0.0222 0.363 0.101 0.103 0.189 0.0887 0.244
0.165 0.128 0.219 0.216 0.0542 0.0630 0.150 0.0830 0.000910 L0.246 0.144 0.166 0.118 0.145 0.145
a
M is the median of all Q data (i.e., 1.76 mm).
3. Methods of analysis 3.1. Methodology We employed the method developed and used by Archer et al (Archer and Newson, 2002; Archer, 2003, 2007; Archer et al., 2010) to examine changes in the runoff regime with changes in vegetation. We applied the method to detect changes in daily runoff (Q). The method examines changes in the whole frequency distribution of Q data. For this purpose, the method classifies Q data according to the magnitude and examines temporal changes in the data frequency of each Q class. Numerous studies have examined changes in the runoff regime (Jones and Grant, 1996; Beschta et al., 2000; Jones, 2000; Ashagrie et al., 2006; Lin and Wei, 2008). Almost all examined changes in the runoff magnitude (e.g., annual maximum Q), but did not examine changes in the whole frequency distribution of data according to the Q magnitude. However, it is possible that changes in the runoff regime could vary among different Q magnitude classes. Most previous studies did not consider this possibility, and therefore their results are uncertain (Alila et al., 2009, 2010). Alternative methods were proposed by Archer et al (Archer and Newson, 2002; Archer, 2003, 2007; Archer et al., 2010) and by Alila et al. (2009), who considered changes in the whole frequency distribution of data according to the Q magnitude. Archer’s method is more suited to examining changes in the runoff regime for watersheds with gradual vegetation changes, and Alila’s method to changes in runoff with abrupt vegetation changes (e.g., forest clear-felling). We used Archer’s method because we examined changes in the runoff regime with a gradual vegetation change due to decreased forestry practices. 3.2. Procedures
Fig. 6. Time series of frequencies for (a) daily precipitation P ¼ 100e200 mm, (b) P ¼ 50e60 mm, and (c) P ¼ 10e20 mm. Solid lines are regression lines determined using the least-squares method.
First, we examined general characteristics of precipitation and runoff, such as the mean annual precipitation, runoff, evapotranspiration, frequency distributions of P and Q for the whole period, and typical seasonal changes in precipitation and runoff. Second, we examined changes in the precipitation and runoff regime during the period 1979e2007. P and Q data were classified into 14 and 15 classes according to their magnitudes, respectively. We examined changes in the frequencies for each P and Q class during the period 1979e2007 using the statistical method described below. Third, we examined changes in Q considering the relationship between annual precipitation and the frequency for each Q class. Annual precipitation and the frequency for each Q class generally
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
1533
Kundzewicz et al. (2005), Archer (2007), and Komatsu et al. (2010b). Kundzewicz and Robson (2004) stated that this method is the most appropriate for detecting trends in time-series data, because it does not include assumptions about the form of the distribution from which the data are derived. We first calculated the slope of linear regression, Spearman’s rho, and Kendall’s tau from the original time-series data. We then calculated these parameters after resampling 29 values from the original time-series data using the bootstrap method with data replacement, where the original time-series comprised data for a period of 29 years. This process was repeated 1000 times (Note that our conclusions did not change when this process was repeated 5000 and 10,000 times.). From the parameter values obtained by resampling, we produced probability density functions (PDFs) for the three parameters. We finally compared the parameter values based on the original time-series data with the PDFs. If the parameter values obtained from original time-series data were greater than the highest 2.5% threshold values of the PDFs, the increasing trend of the original time series was considered to be significant at the 5% level (Svensson et al., 2005). We used not one but all three parameters (i.e., the slope of linear regression, Spearman’s rho, and Kendall’s tau) for statistical analysis. Results of the statistical test might vary according to the parameter used (Kundzewicz and Robson, 2004). Therefore, if we obtain the same results from statistical tests using different parameters, the results are more reliable than those derived from a statistical test using only one parameter (Kundzewicz and Robson, 2004). 4. Results and discussion Fig. 3a shows time series of annual precipitation and runoff. The mean annual precipitation and runoff during the 1979e2007 period were 2176 and 1351 mm, respectively. Thus, the mean annual evapotranspiration, calculated as the difference between annual precipitation and runoff, was 825 mm. This value is comparable to values reported for other forested catchments and watersheds on Kyushu (Shimizu et al., 2003; Komatsu et al., 2007c, 2008a). Trends in annual precipitation, runoff, and evapotranspiration during the period were not found to be significant (p > 0.05), in agreement with Komatsu et al.’s (2010b) report using the same dataset. They did not find significant trends in annual evapotranspiration during the period 1979e2007. In addition, they found no significant trends in low flow during the same period. Fig. 7. Time series of frequencies for (a) daily runoff Q ¼ 50Me60M, (b) Q ¼ 10Me20M, and (c) Q ¼ 5Me6M, where M is the median of all Q data (i.e., 1.76 mm). Solid lines are regression lines determined using the least-squares method.
correlate linearly (Archer and Newson, 2002; Archer, 2003, 2007; Archer et al., 2010), indicating that interannual variations in the frequency correspond to interannual variations in precipitation. Thus, we determined the linear regression equation for the relationship between annual precipitation and the frequency for each Q class. We calculated the residual of the frequency (i.e., the observed frequency for the Q class minus the frequency predicted for the Q class using the linear regression equation) and examined temporal changes in the residual. This analysis evaluates changes in Q after excluding interannual variations in the frequencies for Q corresponding to interannual variations in precipitation. 3.3. Statistics To detect temporal trends, we used a non-parametric method described by Kundzewicz and Robson (2004) and used by
Table 3 Correlation between annual precipitation and the data frequency for different daily runoff (Q) classes. The linear slope, intercept, and correlation coefficient (R) are given. Numerals in bold indicate the correlation was positively significant at the 5% level according to Pearson’s correlation coefficient test. Q class
Slope
Intercept
R
Q < 1Ma 1M Q < 2M 2M Q < 3M 3M Q < 4M 4M Q < 5M 5M Q < 6M 6M Q < 7M 7M Q < 8M 8M Q < 9M 9M Q < 10M 10M Q < 20M 20M Q < 30M 30M Q < 40M 40M Q < 50M 50M Q
0.0757 0.0114 0.0173 0.0101 0.00818 0.00335 0.00406 0.00141 0.00197 0.00131 0.0101 0.00227 0.00234 0.000789 0.00102
348 74.6 4.07 6.88 9.56 2.08 4.80 0.728 2.60 1.60 14.8 2.96 4.13 1.03 1.71
0.729 0.234 0.524 0.703 0.679 0.470 0.680 0.398 0.576 0.530 0.865 0.580 0.704 0.419 0.529
a
M is the median of all Q data (i.e., 1.76 mm).
1534
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
Fig. 3b shows seasonal changes in precipitation and runoff averaged for the period 1979e2007. Precipitation and runoff were higher in June and July than in other months. Such seasonality is typical for most regions in Japan (Komatsu et al., 2007a, 2008a, 2010a). Fig. 4a and b show the frequency distributions for all data according to P and Q, respectively. The maximum P was 285 mm, which was recorded on July 2, 2007. When approximating the frequency distribution of annual maximum P data for the period 1979e2007 using the logarithmic normal distribution, the return period for the P value was calculated as 134 years. Q ranged between 0.0678 and 112 mm. The median value (M) for Q was 1.76 mm. High P values were generally recorded during summer, although low P values were recorded throughout the year (Fig. 5a). Consequently, high Q values were generally recorded during summer (Fig. 5b). Table 1 presents the results of statistical analysis for detecting trends in the precipitation pattern. Time series of the frequencies for three P classes are shown in Fig. 6 as examples. No significant trend was observed for any class (Table 1). Table 2 presents the results of statistical analysis for detecting trends in the runoff pattern. Time series of the frequencies for three Q classes are shown in Fig. 7 as examples. No significant increasing trend was observed for any class (Table 2). For one class, a decreasing trend was observed according to the tau value. Table 3 presents the results of correlation analysis for the relationship between annual precipitation and the frequency of each Q class. The correlation was significantly positive in most cases. Table 4 presents the results of statistical analysis for detecting trends in residuals of Q data frequencies. Time series of the frequencies for three Q classes are shown in Fig. 8 as examples. We did not observe significant trends in most classes (Table 4). In the two classes Q ¼ 4Me5M and Q ¼ 6Me7M, we observed significant increasing trends. The numbers of data included in these classes were 239 and 68 during the period 1979e2007 respectively. (Fig. 4b) This indicates that the data for these classes were commonly recorded at the watershed. Thus, the increase in frequencies for these classes indicates an increase in the frequency of moderate peak flow. On the other hand, we did not observe increasing trends for classes Q > 7M. This suggests no clear increase in the frequency of high peak flow in this watershed. The above results agree with those reported by Hirano et al. (2009), which were based on runoff measurements at a smallcatchment scale (w1 ha). Hirano et al. (2009) examined the runoff
regime during storms for two catchments; one was covered with C. obtusa plantation that was not thinned and the other was covered with natural broadleaved forest. They reported a distinct difference in peak flow intensity for a relatively small storm (Q was w10 mm according to Fig. 5b of their paper), and no clear difference in peak flow intensity for a large storm (Q was w50 mm according to Fig. 5c of their paper). One possible explanation for these results is as follows. As the soil layer is not generally saturated during small storms, the infiltration rate of the soil layer could affect the occurrence of overland flow and therefore catchment runoff. Indeed, Hirano et al. (2009) recorded a higher magnitude of overland flow for the coniferous catchment than for the broadleaved
Table 4 Slope of linear regression, Spearman’s rho, and Kendall’s tau for the residual (i.e., observed minus predicted) of the frequency for each daily runoff (Q) class. Numerals in bold indicate the trend was significant at the 5% level. Q class
Slope
Rho
Tau
Q < 1Ma 1M Q < 2M 2M Q < 3M 3M Q < 4M 4M Q < 5M 5M Q < 6M 6M Q < 7M 7M Q < 8M 8M Q < 9M 9M Q < 10M 10M Q < 20M 20M Q < 30M 30M Q < 40M 40M Q < 50M 50M Q
0.479 0.396 0.365 0.0351 0.158 0.0417 0.103 0.00307 0.0139 0.0261 0.0239 0.0367 0.0177 0.00695 0.0168
0.0443 0.0985 0.255 0.0591 0.371 0.0640 0.479 0.0197 0.0246 0.254 0.0394 0.229 0.0837 0.0640 0.177
0.0640 0.113 0.177 0.0616 0.266 0.106 0.330 0.0320 0.0635 0.187 0.0424 0.158 0.131 0.101 0.103
a
M is the median of all Q data (i.e., 1.76 mm).
Fig. 8. Time series of the residuals (observed minus predicted) for (a) daily runoff Q ¼ 50Me60M, (b) Q ¼ 10Me20M, and (c) Q ¼ 5Me6M, where M is the median of all Q data (i.e., 1.76 mm). Solid lines are regression lines determined using the least-squares method.
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
catchment during small storms. On the other hand, the soil layer is generally saturated during large storms. Overland flow would occur owing to return flow from the saturated soil layer, even when the infiltration rate is high (Gomi et al., 2008a). Thus, the influence of the infiltration rate of the soil layer on the magnitude of overland flow is less significant under these conditions. This could result in less considerable difference in peak flow between the two catchments during large storms. Our results also agree with those reported by Maita and Suzuki (2008), who examined changes in Q with forest clearcutting in Japan at a small catchment scale (w1 ha). They reported a significant increase in peak flow intensity during moderate storms (Q was w10 mm), but no significant increase in peak flow intensity during large storms (Q was w50 mm). 5. Conclusions As a countermeasure to increased flood risk due to decreased forestry practices, many local governments in Japan have introduced local taxes to stimulate forestry practices in coniferous plantation areas (Imawaka and Sato, 2008). However, we did not observe increases in frequencies of high peak flow with decreased forestry practices. Therefore, the increase in flood risk might be less than expected. This study is the first to quantitatively assess the increase in flood risk with decreased forestry practices for coniferous plantations in Japan on the basis of watershed runoff data. Considering that there have been many studies reporting low infiltration rates of the soil surface in C. obtusa plantations with decreased forestry practices (Miyata et al., 2007; Gomi et al., 2008a,b; Hirano et al., 2009), further studies are required to generalize our results. If the results of these future studies support ours, then countermeasures for decreased forestry practices might be unnecessary from the viewpoint of flood control. If different results are obtained among regions, current forest policies should be modified to accommodate the risk for each region. Additionally, we recommend studies examining changes in peak flow due to decreased forestry practices using watershed runoff data with high time resolution (e.g., hourly runoff data). Since we could not obtain data with high time resolution, we used daily watershed runoff data. However, previous studies reported that high-intensity overland flow occurs during short-term highintensity storms (Gomi et al., 2008a, 2010b; Miyata et al., 2010). This implies that changes in peak flow due to decreased forestry practices might be more apparent when using data with high time resolution. This paper focused on the relationship between forestry practices and flood risk. Researchers in Japan (Ohgaki, 2005; Onda, 2008) have pointed out other risks due to decreased forestry practices from the viewpoint of soil protection and water quality. Soil transport and water quality are related to the peak-flow regime (e.g., Fukuyama et al., 2005; Ide et al., 2007; Onda, 2008). Thus, the results of this study have implications for research of these risks. Acknowledgments We acknowledge Mr. K. Takeda and Mr. F. Yamamoto from the Japan Water Agency for providing precipitation and runoff data for the Terauchi watershed. We also thank Mr. O. Funakoshi and Ms. N. Ogawa from Fukuoka Prefecture for providing vegetation data for the Terauchi watershed. We are grateful to Dr. A. Kawasaki, Dr. T. Kajisa, and Dr. Y. Miyazawa from Kyushu University and Ms. A. Miyazawa from Fukuoka Prefecture for their assistance in collating and analyzing vegetation data. Thanks are also due to Mr. D. Notomi from Kyushu University for producing the topographical map. We
1535
acknowledge three anonymous reviewers for making critical comments. This research was supported by Core Research for Evolution Science and Technology of the Japan Science and Technology Agency. References Alila, Y., Kuras, P.K., Schnorbus, M., Hudson, R., 2009. Forests and floods: a new paradigm sheds light on age-old controversies. Water Resources Research 45, W08416. Alila, Y., Hudson, R., Kuras, P.K., Schnorbus, M., Rasouli, K., 2010. Reply to comment by Jack Lewis et al. on “forests and floods: a new paradigm sheds light on age old controversies”. Water Resources Research 46, W05802. Archer, D., 2003. Scale effects on the hydrological impact of upland afforestation and drainage using indices of flow variability: the River Irthing, England. Hydrology and Earth System Sciences 7, 325e338. Archer, D.R., 2007. The use of flow variability analysis to assess the impact of land use change on the paired Plynlimon catchments, mid-Wales. Journal of Hydrology 347, 487e496. Archer, D.R., Newson, M.D., 2002. The use of indices of flow variability in assessing the hydrological and instream habitat impacts of upland afforestation and drainage. Journal of Hydrology 268, 244e258. Archer, D.R., Climent-Soler, D., Holman, I.P., 2010. Changes in discharge rise and fall rates applied to impact assessment of catchment land use. Hydrology Research 41, 13e26. Ashagrie, A.G., de Laat, P.J.M., de Wit, M.J.M., Tu, M., Uhlenbrook, S., 2006. Detecting the influence of land use changes on discharges and floods in the Meuse River Basinethe predictive power of a ninety-year rainfall-runoff relation? Hydrology and Earth System Sciences 10, 691e701. Beschta, R.L., Pyles, M.R., Skaugset, A.E., Surfleet, C.G., 2000. Peakflow responses to forest practices in the western cascades of Oregon, USA. Journal of Hydrology 233, 102e120. Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experiments to determine the effect of vegetation changes on water yield and evapotranspiration. Journal of Hydrology 55, 3e23. Brown, A., Zhang, L., McMahon, T.A., Western, A.W., Vertessy, R.A., 2005. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. Journal of Hydrology 310, 28e61. Calder, I.R., 2005. Blue Revolution. Earthscan Publications, London. Duley, F.L., Ackerman, F.G., 1980. Run-off and erosion from plots of different length. Journal of Agricultural Research 48, 505e510. Farley, K.A., Jobbagy, E.G., Jackson, R.B., 2005. Effects of afforestation on water yield: a global synthesis with implications for policy. Global Change Biology 11, 1565e1576. Fukada, H., Watanabe, N., Kajihara, N., Tsukamoto, J., 2005. Altitudinal zoning of understory vegetation in planted Hinoki cypress (Chamaecyparis obtusa Endl.) forest from the view of surface soil erosion control. Japan Journal of Forest Environment 47, 77e84. Fukada, H., Watanabe, N., Kajihara, N., Tsukamoto, J., 2006. Dynamics of under growth and its application to vegetation control of planted Chamaecyparis obtusa Endl. forests with special reference to mitigation of surface soil loss. Journal of the Japanese Forest Society 88, 231e239. Fukuyama, T., Takenaka, C., Onda, Y., 2005. Cs-137 loss via soil erosion from a mountainous headwater catchment in central Japan. Science of the Total Environment 350, 238e247. Fujimori, T., 2000. Living with Forests. Maruzen, Tokyo. Geological Survey of Japan, 2007. Seamless Digital Geological Map of Japan. http:// riodb02.ibase.aist.go.jp/db084/index_e.html. Gomi, T., Sidle, R.C., Ueno, M., Miyata, S., Kosugi, K., 2008a. Characteristics of overland flow generation on steep forested hillslopes of central Japan. Journal of Hydrology 361, 275e290. Gomi, T., Sidle, R.C., Miyata, S., Kosugi, K., Onda, Y., 2008b. Dynamic runoff connectivity of overland flow on steep forested hillslopes: scale effects and runoff transfer. Water Resources Research 44, W08411. Gomi, T., Miyata, S., Onda, Y., 2010a. Overland flow and the rainfall-runoff relationship in a catchment covered with Japanese cedar plantations. Suiri-kagaku 311, 77e94. Gomi, T., Asano, Y., Uchida, T., Onda, Y., Sidle, R.C., Miyata, S., Kosugi, K., Mizugaki, S., Fukuyama, T., Fukushima, T., 2010b. Evaluation of storm runoff pathways in steep nested catchments draining a Japanese cypress forest in central Japan: a geochemical approach. Hydrological Processes 24, 550e566. Grant, G.E., Lewis, S.L., Swanson, F.J., Cissel, J.H., McDonnell, J.J., 2008. Effects of Forest Practices on Peak Flows and Consequent Channel Response: a State-ofscience Report for Western Oregon and Washington. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland. Hattori, S., Abe, T., Kobayashi, T., Tamai, K., 1992. Effect of forest floor coverage on reduction of soil erosion in Hinoki plantations. Bulletin of Forestry and Forest Products Research Institute 362, 1e34. Hirano, T., Terajima, T., Nakamura, T., Sakai, M., Aoki, F., Nanami, A., 2009. The differences in the short-term runoff characteristics between the coniferous catchment and the deciduous catchmentethe effects of storm size on stormflow generation processes of small forested catchments. Journal of the Japan Society of Hydrology & Water Resources 22, 24e39.
1536
H. Komatsu et al. / Journal of Environmental Management 92 (2011) 1528e1536
Hiraoka, M., Onda, Y., Kato, H., Mizugaki, S., Gomi, T., Nanko, K., 2010. Effects of understory vegetation on infiltration capacity in Japanese cypress plantation. Journal of the Japanese Forest Society 92, 145e150. Ichikawa, T., Takahashi, T., Asano, Y., 2003. Effects of the conversion of forest management type from natural deciduous broad-leaved forests to artificial Japanese cypress (Chamaecyparis obtusa) and Japanese cedar (Cryptomeria japonica) forests on nutrient dynamics. Japan Journal of Forest Environment 45, 35e42. Ichikawa, T., Okabe, N., Takahashi, T., Asano, Y., 2005. Comparison of soil microbial biomass and activity among site positions on a slope planted with Japanese cypress and cedar forests. Japan Journal of Forest Environment 47, 21e27. Ichikawa, T., Takahashi, T., Asano, Y., 2006. Comparison of changes in organic matter dynamics due to stand age between artificial Japanese cedar (Cryptomeria japonica D. Don) forests and Japanese cypress (Chamaecyparis obtusa Sieb. et Zucc.) forests. Journal of the Japanese Forest Society 88, 525e533. Ide, J., Nagafuchi, O., Chiwa, M., Kume, A., Otsuki, K., Ogawa, S., 2007. Effects of discharge level on the load of dissolved and particulate components of stream nitrogen and phosphorus from a small afforested watershed of Japanese cypress (Chamaecyparis obtusa). Journal of Forest Research 12, 45e56. Imawaka, S., Sato, N., 2008. Study on new forest maintenance projects by "Forest Environmental Tax". Bulletin of Kyushu University Forest 89, 75e126. Japan Forestry Agency, 2007. Forest White Paper. Japan Forestry Agency, Tokyo. Japan Meteorological Agency, 2010. Climate Statistics. http://www.jma.go.jp/jma/ index.html. Joel, A., Nessing, I., Seguel, O., Casanova, M., 2002. Measurement of surface water runoff from plots of two different sizes. Hydrological Processes 16, 1467e1478. Jones, J.A., 2000. Hydrological processes and peak discharge response to forest removal, regrowth, and roads in 10 small experimental basins, western Cascades, Oregon. Water Resources Research 36, 2621e2642. Jones, J.A., Grant, G.E., 1996. Peak flow responses to clear-cutting and roads in small and large basins, western Cascades, Oregon. Water Resources Research 32, 959e974. Kato, H., Onda, Y., Ito, S., Nanko, K., 2008. Field measurement of infiltration rate using a oscillating nozzle rainfall simulator in devastated Hinoki plantation. Journal of the Japan Society of Hydrology and Water Resources 21, 439e448. Kawasaki, A., 2005. Corporation Between Upstream and Downstream Regions for Forestry Practices. Bachelor thesis, Kyushu University. Kikuzawa, K., 1999. Forest Ecology. Kyoritsu Shuppan, Tokyo. Kiyono, Y., 1988. Dynamics and control of understories in Chamaecyparis obtusa plantations. Bulletin of Forestry and Forest Products Research Institute 359, 1e122. Komatsu, H., Tanaka, N., Kume, T., 2007a. Do coniferous forests evaporate more water than broad-leaved forests in Japan? Journal of Hydrology 336, 361e375. Komatsu, H., Kume, T., Otsuki, K., 2007b. Contemporary role of catchment water balance data for forest evapotranspiration research. Journal of the Japanese Forest Society 89, 345e358. Komatsu, H., Ide, J., Shinohara, Y., Haga, H., Fujiyama, Y., Miyano, T., Maruno, R., Chiwa, M., Kume, T., Higashi, N., Otsuki, K., 2007c. Evapotranspiration from unmanaged coniferous plantations. Suiri-kagaku 297, 107e127. Komatsu, H., Maita, E., Otsuki, K., 2008a. A model to estimate annual forest evapotranspiration in Japan from mean annual temperature. Journal of Hydrology 348, 330e340. Komatsu, H., Kume, T., Otsuki, K., 2008b. The effect of converting a native broadleaved forest to a coniferous plantation forest on annual water yield: a pairedcatchment study in northern Japan. Forest Ecology and Management 255, 880e886. Komatsu, H., Kume, T., Otsuki, K., 2009a. Effects of coniferous plantation thinning on annual wet-canopy evaporation: model verification. Journal of the Japanese Forest Society 91, 94e103. Komatsu, H., Kume, T., Otsuki, K., 2009b. Low flow changes in runoff with converting a coniferous plantation to a broad-leaved forest in the TatsunokuchiMinami catchment, Japan. Ecohydrology 2, 164e172. Komatsu, H., Kume, T., Otsuki, K., 2010a. Water resource management in Japaneforest management or dam reservoirs? Journal of Environmental Management 91, 814e823. Komatsu, H., Kume, T., Shinohara, Y., Miyazawa, Y., Otsuki, K., 2010b. Did annual runoff and low flow decrease with reduced forestry practices in Japan? Hydrological Processes 24, 2440e2451.
Kundzewicz, Z.W., Robson, A.J., 2004. Change detection in hydrological recordsda review of the methodology. Hydrological Sciences Journal 49, 7e19. Kundzewicz, Z.W., Graczyk, D., Maurer, T., Pinskwar, I., Radziejewski, M., Svensson, C., Szwed, M., 2005. Trend detection in river flow series: 1. Annual maximum flow. Hydrological Sciences Journal 50, 797e810. Kuraji, K., 2007. Expectations from society for forest hydrological sciencesetowards establishment of forest sociohydrology. In: The Editing Committee of Forest Hydrology. Forest Hydrology. Tsukiji-shokan, Tokyo, pp. 283e308. Kuraji, K., Hoyano, H., 2004. Green Dams. Tsukiji Shokan, Tokyo. Lin, Y., Wei, X., 2008. The impact of large-scale forest harvesting on hydrology in the Willow watershed of Central British Columbia. Journal of Hydrology 359, 141e149. Maita, E., Suzuki, M., 2008. The effect of forest cutting on flow duration curve of mountainous small watershed: evaluation of the influence of vegetation on inter-watershed variation in flow duration curve. Journal of the Japanese Forest Society 90, 36e45. Miura, S., 2000. Proposal for a new definition to evaluate the status of forest floor cover and floor cover percentage (FCP) from the viewpoint of the protection against raindrop splash. Journal of the Japanese Forestry Society 82, 132e140. Miyata, S., Kosugi, K., Gomi, T., Onda, Y., Mizuyama, T., 2007. Surface runoff as affected by soil water repellency in a Japanese cypress forest. Hydrological Processes 21, 2365e2376. Miyata, S., Kosugi, K., Nishi, Y., Gomi, T., Sidle, R.C., Mizuyama, T., 2010. Spatial pattern of infiltration rate and its effect on hydrological processes in a small headwater catchment. Hydrological Processes 24, 535e549. National Astronomical Observatory, 2009. Chronological Environmental Tables 2009/2010. Maruzen, Tokyo. Ohgaki, S., 2005. Nutrients in River Water. Gihodo, Tokyo. Onda, Y., 2008. Water and Sediment Discharge in Abandoned Plantations. Iwanami, Tokyo. Onda, Y., Yukawa, N., 1995. The influences of understories on the infiltration capacities of Chamaecyparis obtusa plantations (II) Laboratory experiments. Journal of the Japanese Forestry Society 77, 399e407. Onda, Y., Gomi, T., Mizugaki, S., Nonoda, T., Sidle, R.C., 2010. An overview of the field and modelling studies on the effects of forest devastation on flooding and environmental issues. Hydrological Processes 24, 527e534. Parsons, A.J., Brazier, R.E., Wainwright, J., Powell, D.M., 2006. Scale relationships in hillslope runoff and erosion. Earth Surface Processes and Landforms 31, 1381e1393. Shimizu, A., Shimizu, T., Miyabuchi, Y., Ogawa, Y., 2003. Evapotranspiration and runoff in a forest watershed, western Japan. Hydrological Processes 17, 3125e3149. Shinohara, Y., Kumagai, T., Otsuki, K., Kume, A., Wada, N., 2009. Impact of climate change on runoff from a mid-latitude mountainous catchment in central Japan. Hydrological Processes 23, 1418e1429. Sidle, R.C., Hirano, T., Gomi, T., Terajima, T., 2007. Hortonian overland flow from Japanese forest ptantationsan aberration, the real thing, or something in between? Hydrological Processes 21, 3237e3247. Svensson, C., Kundzewics, Z.W., Maurer, T., 2005. Trend detection in river flow series: 2. Flood and low-flow index series. Hydrological Sciences Journal 50, 811e824. Tsujimura, M., Onda, Y., Harada, D., 2006. The role of Horton overland flow in rainfall-runoff process in an unchanneled catchment covered by unmanaged Hinoki plantation. Journal of the Japan Society of Hydrology & Water Resources 19, 17e24. Wakiyama, Y., Onda, Y., Nanko, K., Mizugaki, S., Kim, Y., Kitahara, H., Ono, H., 2010. The estimation of temporal variation in splash detachment in two Japanese cypress plantations of contrasting age. Earth Surface Processes and Landforms 35, 993e1005. Ward, R.C., Robinson, M., 2000. Principles of Hydrology. McGraw-Hill, London. Yao, H.X., Hashino, M., Xia, J., Chen, X.H., 2009. Runoff reduction by forest growth in Hiji River basin, Japan. Hydrological Sciences Journal 54, 556e570. Yue, S., Hashino, M., 2005. Statistical interpretation of the impact of forest growth on streamflow of the Sameura basin, Japan. Environmental Monitoring and Assessment 104, 369e384. Yukawa, N., Onda, Y., 1995. The influences of understories on the infiltration capacities of Chamaecyparis obtusa plantations (I) experimental results using mist type rainfall simulator. Journal of the Japan Forestry Society 77, 224e231.