Research Paper
Advances in River Engineering, JSCE, Vol.18, 2012, June
LONG TERM VARIABLE PROPERTIES
OF RUNOFF PROCESS
IN A MOUNTAINOUS FORESTED CATCHMENT
Edwina Zainal1, Keisuke OHASHI2, Masaki SAKURAI3,
Toshiharu KOJIMA4 and Seirou SHINODA5
1 Civil Engineering Division, Graduate School of Engineering, Gifu University
(1-1 Yanagido, Gifu City 501-1193, Japan)
2 Member of JSCE, Dr. of Eng., Asistant Professor, Faculty of Engineering, Gifu University
(1-1 Yanagido, Gifu City 501-1193, Japan)
3 Student Member of JSCE, Dept. of Civil Engineering, Faculty of Engineering, Gifu University
(1-1 Yanagido, Gifu City 501-1193, Japan)
4 Member of JSCE, Dr. of Eng., Associate Professor, River Basin Research Center, Gifu University
(1-1 Yanagido, Gifu City 501-1193, Japan）
5 Member of JSCE, Dr. of Eng., Professor, Information and Multimedia Center, Gifu University
(1-1 Yanagido, Gifu City 501-1193, Japan)
This paper investigates the changes of fundamental hydrological properties with paying attention to climate change
and forest growth change after thinning by using long term (24 years) data records at Futatsumori research area in
Nakatsugawa City, Japan. The discharge ratio estimated as the ratio between streamflow and rainfall volume over a
certain area and time, as a function of the catchment state conditions. Some seasonal cyclic variations or long term
trends are derived from the hydrological data. The long term discharge ratio in all analysis cases exhibits a decreasing
trend and it is assumed to be caused from increases of temperature, evapotranspiration and rainfall interception with
growth of forest. Comparison of discharge ratio between the period of 1984 to 1989 and 2002 to 2007 have resulted for
100 and 200 mm rainfall event are 11% and 14%, respectively. The highest difference of discharge ratio was observed in
summer season (Jun-Aug) pointing at 15% and 19%, respectively. Therefore the results of decreasing recent discharge
ratio imply that the dealing with discharge ratio coefficient as constant value is not appropriate. In other word the result
of response against both growth of forest and temperature rising also show water resource may decrease in future, if
rainfall situation was same.
Key Words:
Long Term Hydrological Monitoring, Climate Change, Growth of Forest, Mountainous Forest Catchment, Thinning Effect
500
0
branch cutting
thinning
under tree cutting
plantation
400
300
100
200
1988
200
300
100
400
0
plantation area (ha)
Forests are the most plentiful source of the
cleanest water are often located in the mountains
and thus provide the first opportunity to store,
filter, release water for downstream uses; provide
the earliest opportunity to measure precipitation
and streamflow; and its change frequency in seasonal, flood peak, and others1) , thereby allowing
water managers to forecast supplies and adjust
downstream water storage systems. Recently,
the hydrological impacts of climate change and
other forms of forest management and vegetation
growth are recognized globally. The Forest Division of Gifu Prefecture is implementing catchment and forest management projects by estab-
thinning and cutting area (ha)
1. Introduction
500
1985
1990
1995
2000
fiscal year
Fig-1 Distribution time of thinning, cutting and planting area at
Futatsumori experimental forest
- 471 -
136˚
137˚
130˚
138˚
37˚
37˚
36˚
36˚
35˚
35˚
34˚
137˚
40˚
30˚
130˚
138˚
137˚24'00"
30˚
140˚
137˚24'36"
137˚25'12"
800
137˚23'24"
140˚
40˚
34˚
136˚
700
11
00
35˚34'12"
1200
0
100
00
10
900
35˚33'36"
800
GS
700
900
80
0
90
0
35˚33'00"
N
80
0
W
E
90
0
lishing experimental forest2) at Futatsumori research area in Nakatsugawa City, Japan. It were
continually thinning, branch cutting and removal
of the smaller trees, grasses (under tree cutting)
in cover area around 208.5 ha for the period of
1983 to 1988 and small thinning area in 1995
and 1998. Forest is recovering with young trees
gradually at same area around 298.5 ha (Fig-1).
Responses in streamflow to forest recovery may
be almost immediate or considerably delayed, depending on climate, topography, soils and other
factor3) . Several studies have documented the reduction of water yield in region undergoing expansion of growing forest plantation4) . Hibbert5)
reported increases in streamflow proportional to
the amount of cover removed. Thinning, logging and planting operation can alter the conditions and processes involved in the generation of
streamflow6) . Therefore the objective of this paper is to quantify the change in streamflow ratio
(discharge ratio) as a function of the catchment
state condition. To provide this objective we analyzed long term of precipitation and streamflow
data, furthermore to investigates the potential influencing factor of the fundamental hydrological
properties changes.
km
0
0.5
S
1
elevation(m)
300
2. Case Study
(1) Study Area
Fig-2 shows the topographical, geographical
and contours of study area in Futatsumori,
Nakatsugawa City of Gifu Prefecture, Japan.
It is one of part of the Kiso River system.
The study area was established in mountainous
forested catchment of Futatsumori with longi0
0
tude arorund 137◦ 24 E, latitude 35◦ 33 N and elevation was 500 m to 1300 m. The average annual
precipitation was 2284.7 mm. GS (the Gaman
Stream lower catchment, 300 ha) was observed to
analyze the effect of forest growth on discharge
ratio, hydrological cycle and water yield where
dominant forest types are Japanese cedar and
cypress. The long-term hydrological data for the
period of 1984 to 2007 were measured by the Forest Division of Gifu Prefecture.
(2) Climate Trend
In particular, temperature has proved useful in
investigations of streamflow generation7, 8, 9) . We
measured the monthly temperature data from
500
700
900 1100 1300
Fig-2 Study area in Futatsumori, Nakatsugawa City of Gifu Prefecture which is part of the Kiso River system, Japan
three nearby site because of no long term temperature data in Futatsumori site. The sites
were Higashimori site (one of sub-catchment
area in Futatsumori area) with longitude around
0
0
137◦ 24 E, latitude 35◦ 34 N and elevation was
1019 m, Kurokawa site and Nakatsugawa site.
Kurokawa and Nakatsugawa data were collected
during 24 years data from AMeDAS (Automated Meteorological Data Acquisition System)
of Japan Meteorological Agency. Fig-3 presents
comparison mean and quartiles monthly temperature data during 24 years. Nakatsugawa
city temperature is higher than Kurokawa and
Higashimori temperature. It shows the urban
area was around 2 ◦ C higher than the rural land
and mountain area. In order to determine the
temperature as climate trend surrounding catchment area, Kurokawa long term temperature
- 472 -
monthly mean temperature
Quartiles (Nakatsugawa 1984-2007)
Quartiles (Kurokawa 1984-2007)
monthly mean temperature (HM 1998-2004)
Summary statistics
3rd Quartile
median
temperature (oC )
20
1st Quartile
min
15
10
5
0
-5
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
month
Kurokawa temperature 1984-2007
spring
autumn
annual
summer
winter
trend line
dT/dt=+0.01042
11.0 ◦ C
11.4 ◦ C
11.9 ◦ C
12.0 ◦ C
12.6 ◦ C
slope
intercept
determination R2
0
1000
200
2000
400
3000
600
daily
monthly
annual
800
dT/dt=+0.00595
10
dT/dt=+0.04193
4000
1985 1988 1991 1994 1997 2000 2003 2006
10
annual
monthly
daily
8
1.0
0.8
6
0.6
4
0.4
2
0.2
0
0.0
1985 1988 1991 1994 1997 2000 2003 2006
dT/dt=+0.00540
0.00540
0.00427
0.00410
0
1.5
20
15
minimum
1st quartile
median
3rd quartile
maximum
daily mean
discharge
25
11.8 C
0.5 ◦ C
mean discharge ratio
seasonal & annual mean temperature T ( oC )
30
mean
standard deviation
Qd (m3/day)
Fig-3 Monthly mean temperature at three nearby sites,
Kurokawa, Nakatsugawa and HM (Higashimori)
Linear regression
◦
Ry (mm/year)
max
Rd (mm/day), Rm (mm/month)
25
monthly & annual
mean discharge Q m , Q y (m3/day)
30
Table-1 Summary statistics and linear regression of variation
Kurokawa temperature
monthly
monthly (May-Nov)
annual
1.0
0.5
0.0
5
1985 1988 1991 1994 1997 2000 2003 2006
year
0
dT/dt=+0.00044
Fig-5 Long time variation of daily, monthly and annual rainfall
(a), discharge (b) and discharge ratio (c) during 24 years
-5
1985
1988
1991
1994 1997
time t (year)
2000
2003
2006
Fig-4 Seasonal Variation of Kurokawa temperature
data is appropriate. According to Fig-3, the season defined as follows; Dec, Jan, Feb are winter with the lowest average temperature; Mar,
Apr and May are spring with snow begins to
melt, stream well with runoff and temperature
increases more rapidly; Jun, Jul, Aug are summer with the highest temperature; Sep, Oct and
Nov are autumn with decrease temperature more
rapidly. Mountainous forest systems are particularly sensitive to changes in climate, with small
changes having the potential to produce significant effects. Higher temperatures and changing rates of temperature may cause changes to
growth trees10) . Fig-4 shows seasonal and annual
mean temperature at Kurokawa as assume as
catchment temperature. Annual mean temperature trend is +0.0054 ◦ C over the last 24 years.
Winter, spring, summer and autumn temperature have increased by 0.00044 ◦ C, 0.00595 ◦ C,
0.01042 ◦ C and 0.04193 ◦ C, respectively. The
summary statistics (Table-1) shows the coefficient
correlation is small and have weak trend.
3. Methodologies
(1) Rainfall, Discharge and Discharge Ratio
Bosch and Hewlett3) suggest that streamflow
response also depends on the mean annual precipitation of the area. The daily data from 1984 to
- 473 -
1.0
- 0.01 < dQ/dt < 0
0
0.8
discharge ratio
Re =0
rainfall R (mm)
discharge Q (m3/day)
duration of 1 event
0.6
0.4
spring
summer
autumn
winter
without winter
0.2
Qe
Qs
time t (day)
0
0.0
10
direct runoff
base flow
dQ/dt > + 0.01
Qe/Qs < 5
Fig-6 Conceptual representation of extracting unit hydrograph
100
1000
total rainfall in each event (mm)
Fig-7 Seasonal trend between total rainfall in each event (mm)
and discharge ratio
1.0
∫
f =
Qdt
10−3 RA
(1)
The results in Fig-5 (a), (b), (c) shows two contrasting in peak flow before and after 1994. In
these case, an aliasing hydrograph after 1994 at
lower rates. It may an effect of ten minute data
that has many sinusoidal amended to daily data
that may plotted at low sinusoidal. Fig-5(c) also
shows monthly discharge ratio with a higher variability of the error measures in snow dominated
term as a dashed line compared to rainfall dominated term as a solid line. Thus we must pay
attention to discuss about winter precipitation
and discharge ratio later. Periods of rainfall and
discharge data available for analysis have been
taken from 1984 to 2007, but fewer data have not
been recorded on 1995, 1996 and 1997 caused by
instrument trouble.
(2) Unit Hydrograph
Under the natural conditions, the assumptions
of the unit hydrograph cannot be satisfied perfectly. The unit hydrograph at the catchment
0.8
discharge ratio
1993 and ten minute data from 1994 to 2007 have
been recorded by the Forest Division of Gifu Prefecture. The ten minute data were amended to
daily data in order to produce similar responses.
Fig-5 (a), (b), (c) show daily, monthly and annually
data for rainfall, discharge and discharge ratio f
during period. f in Fig-5(c) is taken as combination of the proportion of rain that becomes
streamflow (discharge), discharge event period,
and daily extract event flow. f estimated as the
ratio between streamflow Q (m3 /day) and rainfall
R (mm) over a certain area A (m2 ) and time t (day),
as a function of the catchments state conditions.
0.6
0.4
0.2
events without winter
thinning(1984.4-1989.3)
no thinning(1989.4-2008.3)
0.0
10
100
1000
total rainfall in each event (mm)
Fig-8 Differences of discharge ratio f in period of thinning and
no thinning without winter events data
outlet is then estimated the value of unit discharge with time. Suitable duration of unit hydrograph of each event was one day. These
were drawn in Fig-6, event of the parameters is
−0.01 < dQ/dt < 0. The relative value of terminal
discharge of ending against starting Qe /Q s < 5.
This conception is mainly based on slope that
observed during the discharge increases as the
stage rises and conversely the discharge decreases
as the stage falls.
4. Result
Season is important variable that affect the
streamflow response to logging6) . We analyzed
the trend in different seasons with rainfall as the
predictor variable. Seasonal analyses in Fig-7 indicate that trends exist in spring, autumn and
summer allow us to infer if forest cover change
were affecting rainfall-discharge ratio relationship and the winter trend had some variability
- 474 -
num events
0
8
10
23
64
total rainfall of unit event Rev (mm)
corr.
1.0
0.6
0.4
0.2
1985
1988
1991
1994
2000
2003
df/dt
no event
no event
-0.00572
-0.00624
-0.00207
corr.
1.0
discharge ratio f
0.4
0.2
1988
1991
1985
1988
1991
1994
1997
1994
1997
2000
2003
400<Rev
200<Rev <400
100<Rev <200
50<Rev <100
0<Rev < 50
2003
df/dt
2006
corr.
1 event
1 event
no event no event
-0.00711 -0.32468
-0.00034 -0.01417
+0.00121 0.03801
0.8
0.6
0.4
0.2
0.0
2006
2000
num events
1
0
14
34
111
total rainfall of unit event Rev (mm)
no event
no event
-0.56475
-0.32984
-0.17416
0.6
1985
0.2
(b) summer
0.8
0.0
0.4
(a) spring
num events
0
0
6
20
77
corr.
-0.57496
-0.49286
-0.54243
-0.24599
-0.30418
0.6
time t (year)
400<Rev
200<Rev <400
100<Rev <200
50<Rev <100
0<Rev < 50
df/dt
-0.00971
-0.00842
-0.00998
-0.00403
-0.00517
0.8
0.0
2006
400<Rev
200<Rev <400
100<Rev <200
50<Rev <100
0<Rev < 50
time t (year)
total rainfall of unit event Rev (mm)
1.0
1997
num events
13
16
30
38
81
total rainfall of unit event Rev (mm)
no event
-0.41781
-0.21392
-0.38951
-0.25586
0.8
0.0
discharge ratio f
df/dt
no event
-0.00641
-0.00662
-0.00703
-0.00599
discharge ratio f
discharge ratio f
1.0
400<Rev
200<Rev <400
100<Rev <200
50<Rev <100
0<Rev < 50
1985
1988
1991
1994
1997
time t (year)
time t (year)
(c) autumn
(d) winter
2000
2003
2006
Fig-9 Relationship between partition of total rainfall event and discharge ratio during period
of errors and interpreted as an evidence of Fig5(c). Fig-8 shows comparison of discharge ratio trend between the period of thinning and no
thinning were analyzed without winter data. We
estimated an average reduced of discharge ratio
about 11% and 14% for 100 mm and 200 mm of
rainfall, respectively. Therefore the results imply
that the dealing with discharge ratio as constant
value is not appropriate. It can be attributed to
growth of forest effect when forest cover change
after 14 years after planting. Land use change
can direct the runoff response of a catchment,
depending on the rainfall partitioning11) . The total rainfall of unit event has a large variation
through the catchment area. The shorts rain
occur mainly bigger than 400 mm and the long
rain occur smaller than 50 mm. The relationship
rainfall variability with discharge ratio in season
can be described by Fig-9, -10. Fig-9(a), (b), (c) related to Fig-7 which decreasing of discharge ratio
in each event rainfall during the growing season.
This is likely due to relatively a new vegetation
slightly growth at the early stage and become
more mature, their take up more water through
transpiration. In addition, more mature forest
typically have greater leave area index and consequently higher canopy interception, leading to
more water loss12) . In winter(Fig-9(d)), the trend
of discharge ratio is increasing in total rainfall
per event smaller than 50 mm with 111 events.
It may some error data have been recorded and
may snow dominated at area as same as seasonal analysis in Fig-5(c), -7.Fig-9(b) showed that
response of discharge ratio were very rapid and
significant with proportionately larger rainfall in
summer. The analysis of Fig-10 during periods
showed that a break in the trend or change point
that indicates the effect of thinning and no thinning activity. Discharge ratio increases shown
in proportion of high rainfall due to thinning,
but they are short-lived due to rapid growth
of vegetation. During period without thinning
showed reduction of discharge ratio where plantation took place cumulatively. In other word it
can reduce high flow by storing them in the soil,
which means less chance for floods and enhance
low flow which means less chance for droughts.
Factors such as the total forested area, forest age
and tree density can interactively affect duration
of these responses12) .
- 475 -
1.0
thinning
discharge ratio f
(1984.4-1989.3)
no thinning
(1989.4-2008.3)
0.8
df/dt
corr.
df/dt
corr.
0.6
-0.00508
-0.00629
-0.00738
-0.00692
-0.00424
-0.29991
-0.33985
-0.25992
-0.29173
-0.23729
+0.03649
+0.00342
-0.01998
-0.00319
+0.00070
1.00000
0.06991
-0.35068
-0.03129
0.00903
0.4
0.2
total rainfall of unit event Rev (mm)
0.0
1985
1988
1991
1994
1997
2000
2003
time t (year)
thinning
2006
400<Rev
200<Rev <400
100<Rev <200
50<Rev <100
0<Rev < 50
no thinning
Fig-10 Discharge ratio change in period thinning and no thinning without winter events data
5. Conclusion
Consider long-term annual changes, adjustment time scales, the seasonal pattern of flows,
and changes in both annual and seasonal figure
was indicating that removal of trees was designed
to improve the growth potential of the younger
tree. This research also focusing on the understanding the changes in the hydrological regime is
important because it will affect downstream water availability, vegetation cover stage influence
at this partitioning point compare the forested
catchment including the old situation and the
new situation. A good understanding of the hydrological processes is important for the assessment of the water resources, their management
and conservation on global and regional scales13) .
Factors such as forest age and thinning should be
incorporated to more accurately predict the hydrological effects of forest growth in the future.
Acknowledgment
This work was supported by JSPS KAKENHI Grantin-Aid for Scientific Research (B) (18310021) and Grantin-Aid for challenging Exploratory Research (22651012)
(Prof. Seirou SHINODA). Gifu Univ Grant for revitalization of Regional Collaboration 2008 and Research Projects
2011 (Prof. Seirou SHINODA). We acknowledge the important data in experimental forest and the helpful support
provided by Gifu Prefecture and Forestry Agency, MAFF.
References
1) Shi Q., Yungi W., Yujie W., Effect of reforestation on the hydrological function of a small watershed in the Three Gorges
Reservoir Area, Springer-Verlag, pp. 148-156, 2007.
2) Report of integrated model work of reinforcing water and soil
conserving function (Kiso River), Gifu Prefecture, 2004.
3) Bosch J. M., Hewlett J. D, A review of catchment experiments
to determine the effect of vegetation changes on water yield
and evapotanspiration, J. Hydrology, Vol. 55, pp. 3-23, 1982.
4) Little C., Lara A., McPhee J., Urrutia R., Revealing the impact of forest exotic plantations on water yield in large scale
watersheds in South-Central Chile, J. Hydrology, pp. 162170, 2009.
5) Hibbert A.R., Forest treatment effects on water yield, Proceedings of the International Symposium on Forest Hydrology, edited by W.E. Sopper and H. W. Lull, pp. 527-290,
Pergamon, New York, 1967.
6) Keppeler E.T., Ziemer R.R., Logging effects on streamflow:
water yield and summer low flows at Caspar Creek in Northwestern California, Water Resources Research, Vol. 26, pp.
1669-1679, 1990.
7) D. Kobayashi, Separation of the snowmelt hydrograph by
stream temperatures (Saporo, Japan), J. Hydrology. Vol. 76,
pp. 155-162, 1985.
8) Pangburn T, Forecasting of snowmelt runoff using temperature data, Annual Eastern Snow Conference, pp. 108-113,
1987.
9) Shanley J. B., Peters N. E., Preliminary observations of
streamflow generation during storms in a forested Piedmont
watershed using temperature as a tracer, J. Contaminant Hydrology 6(3), pp. 349-365, 1988.
10) M. Tani, I. Hosoda, Dependence of annual evapotranspiration on a long natural growth of forest and vegetation change,
J. Japan Society of Hydrology and Water Resources, Vol. 25,
No. 2, 2012.
11) Mul L. M., Understanding hydrological processes in
Ungauged Catchment in sub-Saharan Africa, CRCpress/Balkema, Netherlands, 2009.
12) Yao Y., Cai T., Wei X., Zhang M., Ju C., Effect of forest
recovery on summer streamflow in small forested watersheds,
Northeastern China, Hydrological Processes, 2011.
13) Virtual campus in hydrology and water resources, Ecole Polytechnique Federale de Lausanne,
http://echo2.epfl.ch/VICAIRE/, March 2012.
- 476 -
(Received April 5, 2012)