Journal of Environmental Management 91 (2010) 814823

Contents lists available at ScienceDirect

Journal of Environmental Management

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

Water resource management in Japan: Forest management or dam reservoirs?

Hikaru Komatsu 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 Sec. 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 1 April 2009
Received in revised form
8 October 2009
Accepted 23 October 2009
Available online 22 November 2009

Researchers and journalists in Japan recently proposed forest management as an alternative to dam
reservoir development for water resource management. To examine the validity of the proposal, we
compared the potential low-ow increase due to forest clearcutting with the increase due to dam
reservoir development. Here, we focused on forest clearcutting as an end member among various types
of forest management. We rst analyzed runoff data for ve catchments and found a positive correlation
between annual precipitation and the low-ow increase due to deforestation. We then examined the
increase in low-ow rates due to dam reservoir development (dQd) using inow and outow data for 45
dam reservoirs across Japan. Using the relationship between annual precipitation and the low-ow
increase due to deforestation, we estimated the potential increase in the low-ow rate for each dam
reservoir watershed if forests in the watershed were clearcut (dQf). Only 6 of the 45 samples satised
dQf > dQd, indicating that the potential increase in the low-ow rate due to forest clearcutting was less
than the increase due to dam reservoir development in most cases. Twenty-ve of the 45 samples
satised dQf < 0.2 dQd, indicating the potential increase in the low-ow rate due to forest clearcutting
was less than 20% of the increase due to dam reservoir development in more than half the cases.
Therefore, forest management is far less effective for water resource management than dam reservoir
development is in Japan.
2009 Elsevier Ltd. All rights reserved.

Keywords:
Dam reservoir
Forest clearcut
Forest management
Japan
Low ow
Water resource management

1. Introduction
Despite higher precipitation in Japan than in other temperate
regions (National Astronomical Observatory, 2001), water shortages frequently occur in Japan (Ministry of Land, Infrastructure,
Transport and Tourism, 2009a) owing to the countrys large population relative to the land surface area (127,000,000 people in an
area of 378,000 km2).
A total of 2738 dams have been developed in Japan, and
a further 331 are now under construction or have been proposed
(Tonegawa Integrated Dam and Reservoir Group Management
Ofce, 2009). One purpose of dam reservoir development is to
increase low-ow discharge to secure water resources. Most dam
reservoirs are located in forested areas upstream of metropolitan
and agricultural areas (Ministry of Land, Infrastructure, Transport
and Tourism, 2009b).
Recently, dam reservoir development has drawn criticism from
the viewpoints of biological and environmental conservation
(Harada and Yasuda, 2004). There are frequent campaigns opposing

* Corresponding author. Tel.: 81 92 948 3109; fax: 81 92 948 3119.

E-mail address: komatsu@forest.kyushu-u.ac.jp (H. Komatsu).
0301-4797/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2009.10.011

proposed dams (e.g., Hoyano, 2001), and the removal of existing

dams is also debated (e.g., Amano and Igarashi, 2004).
Some researchers and journalists (e.g., Tsukamoto, 1998;
Yorimitsu, 2001; Kuraji, 2003; Amano and Igarashi, 2004) have
proposed forest management, such as patch clearcutting and thinning of coniferous plantation forests and conversion to broadleaved
forests, as a possible alternative to the development of dam reservoirs. Japan has developed large coniferous plantation forests for
timber production, and there was active management of forests
before the 1970s (Fujimori, 2000). However, forest management has
not been economically viable in recent years because of the relatively low price of timber and has not been actively performed for
several decades (Fujimori, 2000). Researchers and journalists have
proposed that low ow could be increased by forest management
because forest management could reduce leaf area and hence
evapotranspiration. This proposal is widely believed to be valid and
many local governments have thus introduced taxes to aid forest
management (e.g., Yorimitsu, 2001; Imawaka and Sato, 2008).
Researchers in Japan have reported an increase in low ow due
to forest management such as forest clearcutting (e.g., Tamai et al.,
2004; Maita et al., 2005). On the other hand, researchers in Japan
developed a method to evaluate the increase in low ow due to
dam reservoir development (e.g., Kume and Kubota, 1998; Komatsu

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

et al., 2009b). Komatsu et al. (2009b) applied the method to evaluate an increase in low ow for seven existing dam reservoirs
located upstream of the Tokyo metropolitan area. They reported an
increase in low ow due to the dam reservoir development.
However, researchers in Japan have not compared the increase due
to forest management with that due to dam reservoir development.
Thus, the effectiveness of forest management in increasing low
ow relative to dam reservoir development has not been evaluated.
Consequently, the validity of the proposal of forest management in
Japan has not been determined (Komatsu et al., 2009a).
This study compares the increase in low ow due to forest
management with that due to dam reservoir development to
evaluate the effectiveness of forest management. Among various
types of forest management, we focused on forest clearcutting as an
end member. Changes in a ow regime due to forest clearcutting
are generally greater than changes due to other forest management
practices such as patch clearcutting and thinning of coniferous
plantation forests, and conversion to broadleaved forests (e.g.,
Bosch and Hewlett, 1982; Scott and Lesch, 1997; Komatsu et al.,
2009a,c).
Though our analysis focuses on Japan, our results are of use to
researchers in other countries. Many dam reservoirs have been
developed to secure water resources in other countries. However,
the construction of dam reservoirs is criticized because it greatly
changes natural ow regimes and therefore affects biodiversity in
river and riparian ecosystems (Poff et al., 1997, 2007; Lytle and Poff,
2004). Forest management might be an alternative to the
construction of dam reservoirs in securing water resources.
2. Materials and methods
This study comprises three parts. First, we examined the
differences in low-ow rates between forested and deforested
periods using catchment runoff data obtained where deforestation
and/or afforestation occurred in Japan. Second, we examined the
increase in low-ow rates due to dam reservoir development using
inow and outow data for dam reservoirs across Japan. Third, we
calculated the potential increase in the low-ow rate if forests in
a dam reservoirs watershed were clearcut and compared it with
the increase due to the dam reservoir development.
2.1. Catchment runoff data
We obtained daily runoff data for ve catchments where
deforestation and/or afforestation had occurred. There are several
other catchments such as Kamabuchi I and Jozankei catchments
where deforestation and/or afforestation had occurred (Maita,
2005) but for which we could not obtain daily runoff data, and thus
we did not use data for these catchments.
Fig. 1 shows the locations of the ve catchments and Table 1
briey describes the catchments. The catchments are located in
western or central Japan, where water shortages are more frequent
than in northern Japan. Thus, the results of our analysis are more
reliable for these regions than for northern Japan. In northern
Japan, catchment runoff is often inuenced by snowmelt (Komatsu
et al., 2008a; Shinohara et al., 2009), which contrasts to the case in
western and central Japan.
2.1.1. Sarukawa I and III catchments
The Sarukawa I and III catchments are adjacent catchments. The
mean annual precipitation for 19592000 in these catchments was
3032 mm and the standard deviation was 704 mm. Fig. 2a is
a histogram of annual precipitation for 19592000. The annual
precipitation ranged between 1913 and 5710 mm. Fig. 3a shows the
seasonal variation in precipitation based on data for 19671972 and

815

19941999. Precipitation amounts and their variations were

greater in summer than in winter.
Broadleaved forests in Sarukawa I and III catchments were
clearcut in 1967. Coniferous trees (Chamaecyparis obtusa and
Cryptomeria japonica for Sarukawa I and III catchments respectively) were planted just after the clearcutting and broadleaved
trees naturally regenerated in the catchments. Thus, the catchments have been covered by coniferous plantation and broadleaved
forests since the late 1970s (Shimizu et al., 2008). Daily runoff data
for 19682000 were available from the Forest Inuences Unit
Kyushu Branch Station (1982), Takeshita et al. (1996) and Shimizu
et al. (2008). Thus, 19681972 data for the deforested period and
19941999 data for the forested period were used in evaluating the
difference in low-ow rates between the forested and deforested
periods. Note that some data for 1997 were missing and therefore
data for this year were not used in the analysis. More complete
descriptions of the catchments were given by Shimizu et al. (2008).
2.1.2. Tatsunokuchi-Kita and Tatsunokuchi-Minami catchments
The Tatsunokuchi-Kita and Tatsunokuchi-Minami catchments
are adjacent catchments. The mean annual precipitation for
19372000 in these catchments was 1232 mm and the standard
deviation was 221 mm. Fig. 2b is a histogram of annual precipitation for 19372000. The annual precipitation ranged between 499
and 1680 mm. Fig. 3b shows the seasonal variation in precipitation
based on data for 19711980. Precipitation amounts and their
variations were greater in summer than in winter.
Coniferous forests, comprising Pinus densiora, in the Tatsunokuchi-Kita catchment were clearcut in 1937. Broadleaved trees
naturally regenerated and the catchment has been covered with
broadleaved forests since the 1960s (Goto et al., 2006). Daily runoff
data for 19372000 were available from the Forestry and Forest
Products Research Institute (1961), the Forest Inuences Unit
Okayama Experimental Site Kansai Branch Station, 1979) and Goto
et al. (2005). Thus, 19471951 data for the deforested period and
19931999 data for the forested period were used to evaluate the
difference in low-ow rates between forested and deforested
periods. Note that some data for 1995 and 1996 were missing and
therefore data for these years were not used in the analysis. A more
complete description of the catchment was given by Goto et al.
(2006).
The Tatsunokuchi-Minami catchment was deforested by forest
re in 1959. Coniferous trees (Pinus sylvestris) were planted
immediately after the forest re and the catchment was covered
with coniferous plantation forests in the 1970s. However, there
was an insect infestation around 1980 and the coniferous forests
of the catchment were completely destroyed. Broadleaved trees
naturally regenerated and the catchment has been covered with
broadleaved forests since the 1990s (Goto et al., 2006). Daily
runoff data for 19372000 were available from the Forestry and
Forest Products Research Institute (1961), the Forest Inuences
Unit Okayama Experimental Site Kansai Branch Station (1979),
and Goto et al. (2005). We performed three comparisons for this
catchment to evaluate the differences in low-ow rates between
forested and deforested periods: (i) 19601964 for the deforested
period versus 19731977 for the forested period; (ii) 19731977
for the forested period versus 19801984 for the deforested
period; and (iii) 19801984 for the deforested period versus
19932000 for the forested period. The comparisons evaluate
changes in low-ow rates due to coniferous afforestation, coniferous deforestation, and broadleaved afforestation respectively.
Note that some of the data for 1995, 1996, and 1997 were missing
and therefore data for these years were not used in the analysis. A
more complete description of the catchment is given by Goto et al.
(2006) and Komatsu et al. (2009c).

816

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

Fig. 1. Location of the catchments in this study. SI, SIII, TK, TM, and AS indicate Sarukawa I, Sarukawa III, Tatsunokuchi-Kita, Tatsunokuchi-Minami, and Aichi-Shirasaka catchments
respectively. Numbers in this gure correspond to the dam reservoirs in Table 2.

2.1.3. Aichi-Shirasaka catchment

The mean annual precipitation for 19301990 in the AichiShirasaka catchment was 1868 mm and the standard deviation was
286 mm. Fig. 2c is a histogram of annual precipitation for 1930
1990. The annual precipitation ranged between 1358 and 2426 mm.
Fig. 3c shows the seasonal variation in precipitation based on data
for 19301938. Precipitation amounts and their variations were
greater in summer than in winter.
The Aichi-Shirasaka catchment was deforested in the 1930s by
repetitive felling for rewood. Such felling was not frequent after
World War II and broadleaved and coniferous trees naturally
regenerated. The catchment has been covered with broadleaved
and coniferous forests since the late 1970s (Ariyakanon et al.,
2000). Daily runoff data for 19301990 are available from The
Tokyo University Forest in Aichi (1976, 1977, 1981, 1984, 1987,
1999). Thus, 19301937 data for the deforested period and 1983
1990 data for the forested period were used in evaluating the

difference in low-ow rates between forested and deforested

periods. A more complete description of the catchment is given by
Ariyakanon et al. (2000).
2.2. Inow and outow data for dam reservoirs
Daily inow Qin and outow Qout data for dam reservoirs are
available from the Database of Dams (Ministry of Land, Infrastructure, Transport and Tourism, 2009b). We selected 45 dam reservoirs across Japan from the database following three criteria. First,
one of the purposes of the dam reservoir must be to increase lowow rates. Second, the dam reservoir must be located in the most
upstream reach of the river system. This criterion enables us to
regard Qin as natural ow without modication by dam reservoirs
(Kume and Kubota, 1998; Komatsu et al., 2009b). Third, most of the
dam watershed must be covered with forest. Table 2 lists the
effective capacity, watershed area, and annual mean Qin of the dam

Table 1
Description of the catchments in this study.
Catchment name

Area (ha)

Annual precipitation (mm year1)

Annual mean temperature ( C)

Data years
Forested period

Deforested period

Sarukawa I
Sarukawa III
Tatsunokuchi-Kita
Tatsunokuchi-Minami
Aichi-Shirasaka

6.6
8.2
17.3
22.6
88.5

3032
3032
1232
1232
1868

13.2
13.2
14.3
14.3
15.4

19941999
19941999
19471951
19731977, 19932000
19301937

19671972
19671972
19932000
19601964, 19801984
19831990

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

0.4

600

817

0.3

400
0.2

200
0.1

0
0

4000

0.3

0.2

0.1

0
500
0.4

6000

Precipitation (mm month-1)

Relative frequency

0.4

2000

1000

1500

2000

300

M A

M J

A S

O N D

M A

M J

A S

O N D

M A

M J J
Month

A S

O N D

200

100

J
400

0.3

300

0.2

200

0.1

100

0
1000

2000

Precipitation (mm

3000

year -1)

Fig. 2. Histogram of annual precipitation for (a) Sarukawa I and III, (b) TatsunokuchiKita and Tatsunokuchi-Minami, and (c) Aichi-Shirasaka catchments.

Fig. 3. Seasonal variations in precipitation for (a) Sarukawa I and III, (b) TatsunokuchiKita and Tatsunokuchi-Minami, and (c) Aichi-Shirasaka catchments. Vertical bars
indicate standard errors.

reservoirs. We used data for 19941999, when there were frequent

water shortages in downstream areas across Japan (Ministry of
Land, Infrastructure, Transport and Tourism, 2009a).

nearly the same as that during the deforested period (Table 3).
Thus, differences in precipitation between forested and deforested
periods have little impact on the FDCs.
The difference in the mean daily runoff rates for the percentage
of time being 95100% between forested and deforested periods dq
ranged between 0.02 and 0.27 mm day1 for the seven cases (Table 3).
The dq values relative to mean daily runoff rates for a percentage of
time of 95100% for the forested periods ranged from 22% to 270%.
The variation in dq for the three cases of the Tatsunokuchi-Minami
catchment (dq ranging between 0.02 and 0.04 mm day1) was
much smaller than the variation among different catchments (dq
ranging between 0.02 and 0.27 mm day1, Table 3). Thus, the
difference in dq due to deforestation and afforestation was relatively minor compared with the difference in dq among different
catchments on the basis of our datasets. The differences in dq due

3. Results
3.1. Differences in low-ow rates between forested and deforested
periods
Fig. 4 shows ow duration curves (FDCs) for the forested and
deforested periods of the ve catchments. In all cases, daily runoff
rates for deforested periods were higher than those for forested
periods for a percentage of time exceeding 50%. Note that we
determined the forested and deforested periods for each case so
that the mean annual precipitation during the forested period was

818

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

Table 2
Description of the dam reservoirs in this study. The table does not include annual
outow data because annual outow was approximately equal to annual inow for
the reservoirs.
Watershed
area (km2)

Annual inow
(mm year1)

8
1
4
18
47

8
7
8
51
491

936
1676
2080
1240
1829

12
21
35
20
41

34
89
359
32
308

672
1050
1677
1617
1121

106
13
11
289
52

301
168
73
417
689

712
905
382
1673
1333

Hitokura
Shorenji
Nunome
Hasu
Mikunigawa

31
27
15
29
20

115
100
75
81
76

765
950
715
1775
4098

21
22
23
24
25

Koshibu
Akigawa
Yokoyama
Ohmachi
Naramata

37
44
33
29
85

288
82
471
193
60

591
1280
1958
2880
2063

26
27
28
29
30

Kusaki
Aimata
Shimokubo
Sonohara
Kawamata

51
20
120
14
73

254
111
323
493
179

1322
1598
528
835
1274

31
32
33
34
35

Ikari
Shichika
Tamagawa
Sagae
Shirakawa

46
100
229
109
50

271
237
287
231
205

1129
1184
3888
2728
2457

36
37
38
39
40

Kamafusa
Ishibuchi
Asaseishikawa
Iwaonai
Isarigawa

39
12
43
96
14

192
154
226
331
113

1329
2526
1731
1462
1369

41
42
43
44
45

Kanayama
Kanoko
Jozankei
Mirikawa
Taisetsu

130
36
79
14
55

470
124
104
115
292

1246
574
1565
2535
1478

No.

Dam reservoir
name

1
2
3
4
5

Kanna
Shinkawa
Hekino
Terauchi
Matsubara

6
7
8
9
10

Itsuki
Yabakei
Midorikawa
Shimachikawa
Haji

11
12
13
14
15

Yasaka
Nomura
Ishitegawa
Sameura
Ohwatari

16
17
18
19
20

Effective
capacity (106 m3)

to forest type (coniferous plantation forests for cases 4 and 5 and

broadleaved forests for case 6) were relatively minor compared
with the differences in dq among different catchments.
Fig. 5 shows the relationship between mean annual precipitation and dq. The positive correlation (p < 0.005) indicates greater
dq for higher annual precipitation. This relationship is used for
comparing low-ow rates for forest clearcutting and dam reservoir
development in Section 3.3.
3.2. Increase in low-ow rates due to dam reservoir development
Using Qin and Qout data for the dam reservoirs, we arranged
Qout  Qin data according to the order of Qin (from higher Qin to

lower Qin). Fig. 6 shows the results for the Shimokubo dam reservoir. This gure divides Qout  Qin data into 20 classes according to
the Qin value and shows the average Qout  Qin value for each class.
Qout  Qin values were negative for the percentage of time being
025% and generally positive for the percentage of time being 25
100%. Qualitatively, the same results were observed for most dam
reservoirs; that is, negative Qout  Qin for higher Qin and positive
Qout  Qin for lower Qin. Thus, dam reservoir development
generally decreased ow rates when natural ow (i.e., Qin) was
high and increased ow rates when natural ow was low.
We dened the Qout  Qin value for the percentage of time being
95100% as the increase in the low-ow rate due to dam reservoir
development (dQd). dQd ranged between 0.95 and 9.02 m3 s1 and
the mean plus or minus standard deviation was 2.63  4.12 m3 s1
(Table 4). Here, Qin for the percentage of time being 95100% ranged
between 0.00 and 8.63 m3 s1. dQd was positive for 43 of the 45 dam
reservoirs. Thus, most dam reservoirs increased ow rates when
natural ow was lowest. Fig. 7 shows the relationship between the
effective capacity of dam reservoirs and dQd. A positive correlation
(p < 0.001) indicates that larger dam reservoirs contributed more
signicantly to increasing low-ow rates.
We did not use FDCs to examine the increase in low-ow rates
due to dam reservoir development in the above analysis. Analysis
based on FDCs can lead to misinterpretation of the effect of dam
reservoir development, as detailed by Komatsu et al. (2009b).
Zero Qout is often recorded when high precipitation occurs,
which can lead to an interpretation that the dam reservoir
development decreases low-ow rates. This interpretation is
incorrect because the zero Qout is due to the operation of the
dam gate; when high precipitation occurs, the dam gate is
typically operated as for Qout being zero in Japan to prevent ood
ow. Such misinterpretation does not arise from analysis based
on Qout  Qin (Komatsu et al., 2009b).
3.3. Comparison of increases in low-ow rates for forest
clearcutting and dam reservoir development
Regressing the relationship between mean annual precipitation
and dq (Fig. 5), we obtained an empirical equation for calculating
the potential increase in low-ow rates due to deforestation from
annual precipitation data. With the input of annual precipitation P
and the area for each dam watershed A, we calculated the potential
increase in the low-ow rate due to forest clearcutting of the whole
area of each dam watershed dQf:,

dQf

h
i

h
i
m3 s1 0:000147P mm year1
i
 h
 0:0855 A km2 k

(1)

where k is a constant (86.4  106) for the conversion of units. The

above equation assumes that the results from catchment-scale
experimental studies are applicable at a watershed scale. This
assumption is supported by results from the multiscale runoff
study by Troendle et al. (2001). Annual precipitation for each dam
watershed was estimated following the methods described in
Appendix.
dQf values ranged between 0.01 and 1.51 m3 s1 (Table 4) and
the mean plus or minus standard deviation was 0.42  0.38 m3 s1.
This mean value was lower than that for dQd (2.63 m3 s1). Fig. 8
shows the relationship between watershed area and dQf. The
positive correlation (p < 0.001) indicates that the variation in dQf
among watersheds was primarily caused by the variation in
watershed area.
Fig. 9a compares dQf with dQd calculated in Section 3.2. Six of the
45 samples satised dQf > dQd and the other 39 samples satised

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

Forested period
100

100

10

10

0.1

0.1

100

Runoff (mm day -1)

Deforested period

0.01
0

20

40

60

80

100

10

10

0.1

0.1

100
10

0.01
0

100

0.01
0

20

40

60

80

819

100

40

60

80

100

20

40

60

80

100

20

40

60

80

100

0.01
0

100

20

10

1
0.1

0.01
0.001
0
100

0.1
0

20

40

60

80

100

20

40

60

80 100
Percentage of time (%)

10
1
0.1
0.01
0

Fig. 4. Flow duration curves for the forested and deforested periods for (a) Sarukawa I, (b) Sarukawa III, (c) Tatsunokuchi-Kita, (d) Tatsunokuchi-Minami (19601964 versus
19731977), (e) Tatsunokuchi-Minami (19731977 versus 19801984), (f) Tatsunokuchi-Minami (19801984 versus 19932000), and (g) Aichi-Shirasaka catchments.

dQf < dQd. This indicates that the potential increase in low-ow
rates due to forest clearcutting was less than that due to dam
reservoir development in most cases. Twenty-ve of the 45
samples satised dQf < 0.2 dQd (note that the ratio 0.2 was arbitrarily determined), indicating that the potential increase in lowow rates due to forest clearcutting was less than 20% of the
increase due to dam reservoir development in more than half the

cases. Fig. 9b shows the relationship between watershed area and

the effective capacity of dam reservoirs classied by the relationship between dQf and dQd. Data satisfying dQf > dQd and 0.2
dQd < dQf < dQd generally corresponded to greater watershed area
and smaller dam capacity. This indicates that forest clearcutting is
relatively effective for these specic conditions but is quite ineffective for other conditions.

Table 3
Annual precipitation (P), low-ow rates for forested (qf) and deforested periods (qd) and the differences in low-ow rates between forested and deforested periods (dq
qd  qf). The low-ow rate is dened as the mean daily runoff rate for ow durations of 95100%.
Case

Catchment

Forested period
1

P (mm year
1
2
3
4
5
6
7

Sarukawa I
Sarukawa III
Tatsunokuchi-Kita
Tatsunokuchi-Minami (19601964 vs. 19731977)
Tatsunokuchi-Minami (19731977 vs. 19801984)
Tatsunokuchi-Minami (19801984 vs. 19932000)
Aichi-Shirasaka

2881
2881
1189
1197
1197
1204
1853

Deforested period
1

qf (mm day
0.12
0.10
0.03
0.09
0.09
0.07
0.39

1

P (mm year
2809
2809
1179
1211
1205
1205
1806

dq (mm day1)

dq/qf (%)

0.18
0.27
0.07
0.03
0.02
0.04
0.20

150
270
233
33
22
57
51

1

qd (mm day
0.30
0.38
0.10
0.12
0.11
0.11
0.59

820

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

Table 4
Increase in low-ow rates due to dam reservoir development (dQd) and forest
clearcutting (dQf) and inow rates for the percentage of time being 95100% (Qinlow).

0.3

dq (mm day-1)

dQf (m3 s1)

0.16
0.32
0.04
0.35
0.10

0.00
0.01
0.11
0.42
8.63

0.01
0.02
0.03
0.09
1.23

0.07
1.33
1.26
0.70
1.26

1.90
0.29
5.10
0.16
1.71

0.03
0.13
0.83
0.07
0.48

2.16
0.73
0.23
23.05
2.57

1.13
0.57
0.04
1.43
5.50

0.31
0.21
0.04
0.96
1.27

Hitokura
Shorenji
Nunome
Hasu
Mikunigawa

1.56
1.31
1.06
1.13
1.67

0.28
0.36
0.07
0.53
2.79

0.13
0.13
0.08
0.20
0.42

21
22
23
24
25

Koshibu
Akigawa
Yokoyama
Ohmachi
Naramata

1.58
3.93
1.20
1.39
9.02

0.18
0.56
4.90
3.59
0.32

0.25
0.15
1.26
0.75
0.17

26
27
28
29
30

Kusaki
Aimata
Shimokubo
Sonohara
Kawamata

0.97
1.91
2.70
0.95
0.07

2.04
0.91
0.76
1.67
1.91

0.47
0.24
0.25
0.59
0.32

31
32
33
34
35

Ikari
Shichika
Tamagawa
Sagae
Shirakawa

0.23
2.91
4.84
4.90
2.55

1.66
2.33
5.92
5.42
2.67

0.43
0.39
1.51
0.86
0.69

36
37
38
39
40

Kamafusa
Ishibuchi
Asaseishikawa
Iwaonai
Isarigawa

3.00
2.67
3.49
9.96
0.16

1.86
1.49
4.31
2.25
2.60

0.35
0.53
0.54
0.58
0.18

41
42
43
44
45

Kanayama
Kanoko
Jozankei
Mirikawa
Taisetsu

12.84
1.21
5.37
1.06
0.90

4.98
0.81
0.52
1.72
4.05

0.69
0.07
0.20
0.37
0.52

Dam reservoir name

0.2

1
2
3
4
5

Kanna
Shinkawa
Hekino
Terauchi
Matsubara

0.1

6
7
8
9
10

Itsuki
Yabakei
Midorikawa
Shimachikawa
Haji

11
12
13
14
15

Yasaka
Nomura
Ishitegawa
Sameura
Ohwatari

16
17
18
19
20

0
1000

1500

2000

2500

3000

Precipitation (mm year-1)

Fig. 5. Relationship between mean annual precipitation and the difference in the
mean daily runoff rates for ow durations of 95100% between forested and deforested periods dq. The regression line, determined by the least-squares method, is
y 0.000147x  0.0855. The correlation is signicantly (p < 0.005) positive according
to Pearsons correlation coefcient test.

4. Discussion
4.1. Differences in low-ow rates between forested and deforested
periods
We observed increases in low-ow rates with deforestation and
decreases in low-ow rates with afforestation. This agrees with
results of many earlier studies in Japan (e.g., Tamai et al., 2004;
Maita et al., 2005) and in other countries (e.g., Brown et al., 2005;
Farley et al., 2005).
We found a positive correlation between mean annual precipitation and dq. This is supported by two other studies (Farley et al.,
2005; Maita, 2005). Farley et al. (2005) summarized measurements
for 26 catchments around the world and reported that changes in

Qout - Qin (m3s-1)

Qinlow (m3 s1)

No.

R2 = 0.80

dQd (m3 s1)

-2
low-ow rates due to afforestation correlated with changes in
annual runoff. Maita (2005) summarized measurements for seven
catchments in Japan and reported that changes in annual runoff
due to afforestation and deforestation correlated with annual
precipitation. Therefore, a positive correlation between mean
annual precipitation and the difference in low-ow rates between
forested and deforested periods is expected.

-4

-6

4.2. Increase in low-ow rates due to dam reservoir development

-8

20

40
60
80
Percentage of time (%)

100

Fig. 6. The difference in outow Qout and inow Qin arranged according to the order
of Qin for the Shimokubo dam reservoir. Vertical bars indicate standard errors.

Our results show that dQd was positive for most dam reservoirs
and that dQd was positively correlated with the effective capacity of
dam reservoirs. These results agree with ndings in a previous
study (Komatsu et al., 2009b), which evaluated increases in lowow rates for seven existing dam reservoirs located upstream of the

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

821

dQf = dQd

25

dQf = 0.2 dQd

2

20
15
dQf (m3s-1)

dQd (m3 s-1)

R2 = 0.64

10
5

0
0

100

200

300

-5

0
-5

Dam reservoir capacity (106 m3)

Fig. 7. Relationship between the effective capacity of dam reservoirs and the increase
in low-ow rates due to dam reservoir development dQd. The solid and dotted lines
indicate relationships for dQf dQd and dQf 0.2 dQd respectively. The regression line,
determined by the least-squares method, is y 0.0585x  0.434. The correlation is
signicantly (p < 0.001) positive according to Pearsons correlation coefcient test.

4.3. Comparison of increases in low-ow rates for forest

clearcutting and dam reservoir development
We compared increases in low-ow rates for forest clearcutting
and dam reservoir development. In reality, forest clearcutting of the

b
Dam reservoir capacity (106 m3)

Tokyo metropolitan area. Thus, the ndings of the previous study

hold not only in the specic area but all over Japan.
The positive dQd suggests that the dam reservoirs generally
homogenize natural ow dynamics. This agrees with what was
noted by Poff et al. (1997, 2007) and Lytle and Poff (2004), who
examined river ow data in the United States and reported that
river ow dynamics are homogenized by dam reservoir
development.

R2 = 0.63

dQf (m3 s-1)

5
dQd (m3s-1)

10

15

dQf > dQd

0.2 dQd < dQf < dQd
dQf < 0.2 dQd
300

200

100

0
1.5

200
400
Watershed area (km2)

600

Fig. 9. (a) Comparison of the potential increase in low-ow rate due to forest clearcutting dQf with the increase due to dam reservoir operation dQd. Note that this gure
does not include the sample for the Sameura dam reservoir (dQd 23.05 and
dQf 0.96) because of its extreme dQd value. (b) Relationship between watershed area
and effective capacity of dam reservoirs classied by the relationship between dQf and
dQd, where dQf is the potential increase in the low-ow rate due to forest clearcutting
and dQd is the increase in the low-ow rate due to dam reservoir development.

0.5

0
0

200
400
Watershed area (km2)

600

Fig. 8. Relationship between watershed area and the potential increase in low-ow
rates due to forest clearcutting of the entire watershed of each dam reservoir. The
regression line, determined by the least-squares method, is y 0.00192x  0.0265. The
correlation is signicantly (p < 0.001) positive according to Pearsons correlation
coefcient test.

whole dam watershed area is unrealistic because it can cause

environmental problems relating to oods, soil erosion, water
quality, and biodiversity (e.g., Tsukamoto, 1998; Ohte et al., 2001;
Brooks et al., 2003). Realistic forest management methods that have
been proposed by researchers and journalists and implemented by
local governments in Japan are patch clearcutting and thinning of
coniferous plantation forests, and conversion to broadleaved forests
(e.g., Tsukamoto, 1998; Yorimitsu, 2001; Kuraji, 2003; Amano and
Igarashi, 2004). Changes in the ow regime following these
methods are generally less signicant than those due to forest
clearcutting of the whole watershed (e.g., Bosch and Hewlett, 1982;

822

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

Scott and Lesch, 1997; Maita, 2005; Komatsu et al., 2009b).

Furthermore, changes in the ow regime due to thinning of coniferous plantation forests becomes less signicant within several
years of the treatment possibly because forest gaps developed by
the treatment are lled with branch extensions (Komatsu et al.,
2009a). Thus, the effectiveness of forest management in realistic
cases should be less than that evaluated in this study for the end
member case (i.e., forest clearcutting of the whole watershed).
To our knowledge, this study is the rst evaluating the potential
increases in low-ow rates due to forest management relative to
the increases due to the dam development. The results of this study
would not be directly applicable outside Japan because increases in
low-ow rates due to forest management and dam reservoir
development would vary with climatological conditions (e.g.,
Fig. 5). However, our method of comparing the low-ow increase
due to forest management and that due to dam reservoir development is applicable to many other countries. Our method only
requires daily runoff data on deforested or afforested catchments
and daily inow and outow data of dam reservoirs, which are
readily available in many countries. Thus, this study can enhance
examinations on the effectiveness of forest management relative to
dam reservoir development for securing water resources.

Sawano, 2006) owing to the large variation in precipitation

amounts and sparse distribution of precipitation measurement
points in forested areas.
Note that determination of annual evapotranspiration is not
critical for our results because variations in annual evapotranspiration between regions are much smaller than variations in annual
runoff (Komatsu et al., 2008b), and annual Qin is generally greater
than annual evapotranspiration in Japan (Table 2).
When using precipitation data taken by the Japan Meteorological Agency (2009) at a station in/near each dam watershed, our
results varied somewhat but not substantially. We compared the
potential increase in the low-ow rate when forests in each dam
reservoirs watershed were clearcut (dQf) with the increase due to
dam reservoir development (dQd) in Section 3.3. When the
measured precipitation data were used, four of the 45 samples
satised dQf > dQd and the other 41 samples satised dQf < dQd.
Twenty-nine of 45 samples satised dQf < 0.2 dQd. Data satisfying
dQf > dQd and 0.2 dQd < dQf < dQd usually corresponded to greater
watershed area and smaller capacity of dam reservoirs, similar to
what is seen in Fig. 9b.

5. Conclusions

Amano, R., Igarashi, T., 2004. A Way to Dam Removal. Tokyo Shoseki, Tokyo.
Ariyakanon, N., Numamoto, S., Suzuki, M., 2000. Sixty-year decreasing trend of bare
land in Shirasaka Watershed, University Forest in Aichi, revealed by aerial
photography. Bulletin of Tokyo University Forest 103, 339348. Available at: http://
repository.dl.itc.u-tokyo.ac.jp/dspace/bitstream/2261/22508/1/esrh103008.pdf.
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, 323.
Brooks, K.N., Ffolliott, P.F., Gregersen, H.M., DeBano, L.F., 2003. Hydrology and the
Management of Watersheds. Iowa University Press, Ames.
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, 2861.
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 11,15651576.
Forest Inuences Unit Kyushu Branch Station, 1982. Statistical report of hydrological
observation at Sarukawa experimental watersheds (January, 1967December,
1976). Bulletin of Forestry and Forest Products Research Institute 317, 147190.
Available at: http://www.ffpri.affrc.go.jp/labs/kanko/317-5.pdf.
Forest Inuences Unit Okayama Experimental Site Kansai Branch Station, 1979.
Statistical report of hydrological observation at Tatsunokuchiyama experimental watershed. January, 1959December, 1977. Bulletin of Forestry and
Forest Products Research Institute 308, 133195. Available at: http://www.ffpri.
affrc.go.jp/labs/kanko/308-6.pdf.
Forestry and Forest Products Research Institute, 1961. Observation report of
hydrological experimental watersheds. Forestry and Forest Products Research
Institute, Tokyo.
Fujimori, T., 2000. Living with Forests. Maruzen, Tokyo.
Goto, Y., Tamai, K., Kominami, Y., Miyama, T., 2005. Hydrological observation reports
in Tatsunokuchi-yama experimental forest. January, 1981December, 2000.
Bulletin of Forestry and Forest Products Research Institute 394, 87133. Available at: http://www.ffpri.affrc.go.jp/labs/kanko/394-6.pdf.
Goto, Y., Tamai, K., Miyama, T., Kominami, Y., Hosoda, I., 2006. Effects of disturbance on
vertical stratication of broad-leaved secondary forests in Tatsunokuchi-yama
Experimental Forest. Bulletin of Forestry and Forest Products Research Institute
400, 215225. Available at: http://www.ffpri.affrc.go.jp/labs/kanko/400-2.pdf.
Harada, J., Yasuda, N., 2004. Conservation and improvement of the environment in
dam reservoir. Water Resources Development 20, 7796.
Hoyano, H., 2001. Why Dam-Free in Nagano? Tsukiji Shokan, Tokyo.
Imawaka, S., Sato, N., 2008. Study on new forest maintenance projects by "Forest
Environmental Tax. Bulletin of Kyushu University Forest 89, 75126. Available at:
http://www.forest.kyushu-u.ac.jp/pdf/bkuf_089/bkuf_089_075-126_imawaka.pdf.
Japan Meteorological Agency, 2009. Meteorological Statistics. http://www.jma.go.
jp/jma/menu/report.html.
Komatsu, H., Kume, T., Otsuki, K., 2008a. The effect of converting a native broad-leaved forest to a coniferous plantation forest on annual water yield: a paired-catchment study in northern Japan. Forest Ecology and Management 255,
880886.
Komatsu, H., Maita, E., Otsuki, K., 2008b. A model to estimate annual forest
evapotranspiration in Japan from mean annual temperature. Journal of
Hydrology 348, 330340.
Komatsu, H., Kume, T., Otsuki, K., 2009a. Effects of coniferous plantation thinning
on annual interception evaporation: model verication. Journal of the Japanese
Forest Society 91, 94103. Available at: http://www.jstage.jst.go.jp/article/jjfs/
91/2/94/_pdf/-char/ja/.

The potential increase in the low-ow rate due to forest clearcutting was estimated as being lower than the increase due to dam
reservoir development in most cases. Thus, our analysis claried
that forest management is generally far less effective for water
resource management than the construction of dam reservoirs is in
Japan. Only when the watershed area is great and the capacity of
dam reservoirs small can forest management be an effective
method of water resource management.
The Japanese public widely believes that forest management
can be an alternative to dams and many local governments have
introduced taxes to aid forest management. However, our results
do not support this common belief. Thus, the present public policy
on water resource management in Japan needs to be reconsidered.
Acknowledgments
We express sincere thanks to Ms. Chiyoko Kumagai and Mr.
Kenji Tsuruta (Kyushu University, Japan) for providing catchment
runoff data. We also thank Mr. Yoshinori Shinohara (Kyushu
University, Japan) for fruitful discussion on the selection of dams
and accuracy of precipitation data. Thanks are also due to two
anonymous reviewers for their critical and constructive comments.
This research has been supported by a Grant-in-Aid for Scientic
Research from the Japanese Ministry of Education, Culture, Sports,
Science and Technology (#20780119) and a CREST project (Development of innovative technologies for increasing in watershed
runoff and improving river environment by the management
practice of devastated forest plantation).
Appendix. Methods for estimating annual precipitation
Annual precipitation was estimated by summing the mean
annual inow Qin and mean annual evapotranspiration for each
dam watershed. Mean annual evapotranspiration was assumed as
600 mm year1 for the Hokkaido region, 1000 mm year1 for the
Okinawa region, and 800 mm year1 for the other regions (Fig. 1),
which were typical values (Komatsu et al., 2008b). Watershedaveraged precipitation is often estimated more accurately if based
on Qin data than if based on precipitation data taken at a station in
the watershed or near the watershed in Japan (Tsuchiya, 2005;

References

H. Komatsu et al. / Journal of Environmental Management 91 (2010) 814823

Komatsu, H., Kume, T., Otsuki, K., 2009b. Evaluating the contribution of dam
reservoirs located upstream of the Tokyo metropolitan area during lowow
periods. Bulletin of Kyushu University Forest 90, 111.
Komatsu, H., Kume, T., Otsuki, K., 2009c. Low ow changes in runoff with converting a coniferous plantation to a broad-leaved forest in the TatsunokuchiMinami catchment. Japan. Ecohydrology 2, 164172.
Kume, T., Kubota, J., 1998. Evaluation of effects of dam reservoir on dischargeduration curve in forested catchment. Journal of the Japan Society of Hydrology
& Water Resources 11, 317323.
Kuraji, K., 2003. Effects of Forests on Stabilizing Streamow. Nihon Chisan-chisui
Kyokai, Tokyo.
Lytle, D.A., Poff, N.L., 2004. Adaptation to natural ow regimes. Trends in Ecology
and Evolution 19, 94100.
Maita, E., 2005. Effects of forest clearcut on runoff at Fukuroyamasawa catchment in
Chiba research forest. Ph.D. Thesis, The University of Tokyo, Tokyo, 194 pp.
Maita, E., Suzuki, M., Ohta, T., 2005. Change in the annual discharge after clear-cutting
a 70-year-old mixed plantation of Sugi and Hinoki in a Tertiary formation area.
Journal of the Japanese Forestry Society 87,124132. Available at: http://nels.nii.ac.
jp/els/110002975571.pdf?idART0003858245&typepdf&langjp&hostcinii
&order_no&ppv_type0&lang_sw&no1252996943&cp.
Ministry of Land, Infrastructure, Transport and Tourism, 2009a. Water Resources in
Japan. Available at: http://www.mlit.go.jp/tochimizushigen/mizsei/hakusyo/
index5.html.
Ministry of Land, Infrastructure, Transport and Tourism, 2009b. Database of Dams.
Available at: http://www2.river.go.jp/dam/.
National Astronomical Observatory, 2001. Chronological Scientic Tables 2002.
Maruzen, Tokyo.
Ohte, N., Mitchell, M.J., Shibata, H., Tokuchi, N., Toda, H., Iwatsubo, G., 2001.
Comparative evaluation on nitrogen saturation of forest catchments in
Japan and northeastern United States. Water Air and Soil Pollution 130,
649654.
Poff, N.L., Allan, J.D., Bain, M.B., Karr, J.R., Prestegaard, K.L., Richter, B.D., Sparks, R.E.,
Stromberg, J.C., 1997. The natural ow regime. Bioscience 47, 769784.
Poff, N.L., Olden, J.D., Merritt, D.M., Pepin, D.M., 2007. Homogenization of regional
river dynamics by dams and global biodiversity implications. Proceedings of the
National Academy of Sciences 104, 57325737.
Sawano, S., 2006. Evaluation of Water Resources Available from Forested areas in
Japan using GIS data. Ph.D. Thesis, The University of Tokyo, 86 pp.
Scott, D.F., Lesch, W., 1997. Streamow responses to afforestation with Eucalyptus
grandis and Pinus patula and to felling in the Mokobulaan experimental
catchments, South Africa. Journal of Hydrology 199, 360377.

823

Shimizu, A., Miyabuchi, Y., Shimizu, T., Ogawa, Y., Daimaru, H., Sato, T., Takeshita, M.,
2008. Statistical report of hydrological observations at the Sarukawa experimental watershed. January 1987 to December 2000. Bulletin of Forestry and
Forest Products Institute 406, 1365. Available at: http://www.ffpri.affrc.go.jp/
labs/kanko/406-3.pdf.
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, 14181429.
Takeshita, M., Shimizu, A., Miyabuchi, Y., 1996. Statistical report of hydrological
observation at Sarukawa experimental watershed. Bulletin of Forestry and
Forest Products Institute 370, 3175. Available at: http://www.ffpri.affrc.go.jp/
labs/kanko/370-3.pdf.
Tamai, K., Goto, Y., Miyama, T., Kominami, Y., 2004. Inuence of forest decline by
forest re and pine wilt disease on discharge and discharge duration curve: In
the case of Tatsunokuchi-yama Experimental Forest. Journal of the Japanese
Forest Society 86, 375379.
The Tokyo University Forest in Aichi, 1976. Report of the daily runoff in the Tokyo
University Forest in Aichi (I). The Tokyo University Forest 20, 3964.
The Tokyo University Forest in Aichi, 1977. Report of the daily runoff in the Tokyo
University Forest in Aichi (II). The Tokyo University Forest 21, 4889.
The Tokyo University Forest in Aichi, 1981. Report of the daily runoff in the Tokyo
University Forest in Aichi (III). The Tokyo University Forest 22, 84191.
The Tokyo University Forest in Aichi, 1984. Report of the daily runoff in the Tokyo
University Forest in Aichi (IV). The Tokyo University Forest 23, 5788.
The Tokyo University Forest in Aichi, 1987. Report of the daily runoff in the Tokyo
University Forest in Aichi (V). The Tokyo University Forest 25, 135151.
The Tokyo University Forest in Aichi, 1999. Report of the daily runoff in the Tokyo
University Forest in Aichi (VI). The Tokyo University Forest 38, 127146.
Tonegawa Integrated Dam and Reservoir Group Management Ofce, 2009. Lecture
of dams. http://www.ktr.mlit.go.jp/tonedamu/lecture/d04/d04E.html.
Troendle, C.A., Wilcox, M.S., Bevengerc, G.S., Porth, L.S., 2001. The Coon Creek Water
Yield Augmentation Project: implementation of timber harvesting technology
to increase streamow. Forest Ecology and Management 143, 179187.
Tsuchiya, I., 2005. Interannual variation of annual precipitation more than 12000
mm on Mt. Chokai between 1979 and 2003: estimation based on the dam ow
data. Tenki 52, 681684. Available at: http://nels.nii.ac.jp/els/110001868242.
pdf?idART0002038386&typepdf&
langjp&hostcinii&order_no&ppv_type0&lang_sw&no1252997611&cp.
Tsukamoto, Y., 1998. Conservation of Forest, Water, and Soil. Asakura, Tokyo.
Yorimitsu, R., 2001. Environmental Conservation at Catchment Scale. Nihon-KezaiHoso Shuppan, Tokyo.