Hydrology of Humid Tropical Regions with Particular Reference to the Hydrological
Effects of Agriculture and Forestry Practice (Proceedings of the Hamburg
Symposium, August 1983). IAHS Publ. no. 140.
Effects of deforestation on flood characteristics
with particular reference to Hainan Island,
China
QIAN WANGCHENG
Hydrological
General Station
Province,
Canton, China
ABSTRACT
of Guangdong
Hainan Island, the second largest island in
2
China, has an area of 33 900 km and lies in a humid
tropical region. In the last 30 years the forest cover
has been drastically reduced from about 50 to 21%. To
study the effects of deforestation on flood characteristics,
trends in annual storm rainfalls and floods, rainfall/
runoff relationships, and convergence of runoff in the
water courses, are analysed. Since deforestation should
not change the atmospheric circulation, there should be
no noticeable change in annual depth of storm rainfall and
flood. Hainan Island is in a humid tropical region with
heavy storm rainfall where scrub grows rapidly after
deforestation; therefore flood characteristics are not
noticeably affected by deforestation.
Influence
de la deforestation
sur les
exemple concernant l'île
d'Hainan
caractères
des
crues:
RESUME
L'île Hainan, avec sa superficie de 33 900 km z
située dans une région humide et chaude, est la deuxième
grande île en Chine. Le pourcentage de couverture
forestière dans l'île a été réduite brutalement depuis les
30 dernières années de 50 à 21%. Pour l'étude de
l'influence de la deforestation sur les crues, une analyse
particulière a été présentée dans le présent rapport,
relative aux trois points suivants: tendance des variations
pluriannuelles des averses et des crues, relation entre les
précipitations et l'écoulement, et la convergence de
l'écoulement des cours d'eau. Comme la deforestation ne
peut pas modifier les conditions de circulation atmosphérique dans l'île d'Hainan, les tendances de variations
pluriannuelles des averses et des crues demeurent
inchangées. Par suite de sa situation en région humide et
chaude, des concentrations d'averses intenses se produisent
souvent; après la deforestation, le fourré de végétation
secondaire croît abondamment en peu de temps, par suite
il n'y a pas de changement notable dans le caractère
des crues.
INTRODUCTION
2
Hainan, with an area of 33 900 km , is the next largest island in
China after Taiwan. It is located between 108°-110°N, and 18°-20°E
(Fig.l), and is a tropical island with evident tropical monsoon
249
250 Qian
Wangcheng
climate. The annual mean air temperature of the whole island is
about 24°C, the accumulated temperature âlO°C is 8200-9200°C. The
mean annual precipitation of the whole island is about 1800 mm. As
a result of the seasonal alternations of the East Asia monsoon
circulation, the space-time distribution of precipitation is uneven,
with greater precipitation in the southeast than in the northwest,
and a distinct wet and dry season. The island lies in the region
much affected by typhoons and typhoon rain is the main source of
precipitation, with high concentration and high intensity; the amount
of annual precipitation is also closely related to typhoons. The
Altitude increases towards the centre of the island: the mean
elevation is about 220 m a.m.s.l., and about a quarter of the total
area has an altitude >500 m a.m.s.l. The summit of Mt Wuzhi is
1879 m a.m.s.l. Nandu River, Wanguan River and Changhua River are
the three main rivers on the island, together they drain 47% of the
total area of the island. There are 10 rivers with drainage areas
of 500-2000 kmz; these are all rather short, steep and fast flowing
rivers.
C'^.J
-^
"•3?
River
Basin Boundary
Division Boundary
Mount
Streamgauge
Ramgauge
1
FIG. I
liai nan Is land .
In the last 30 years, due to the lack of forest conservation and
afforestation, the forested area was reduced from about 50 to 21%
by 1980 (including artificial forest). Only 11% of the island is
covered by natural forest.
In order to study the effects of deforestation on flood characteristics, trends in the annual variation of storm rainfall and runoff,
the rainfall-runoff relationship, and the convergence of runoff, are
analysed in this paper.
Large-scale deforestation occurred in 1958, 1968 and 1978.
However, since hydrological stations were only established in the
late 1950's, analyses can only be made of variations in flood
characteristics in the 1960's and 1970's.
Effects
of deforestation
on flood
characteristics
251
VARIATIONS IN ANNUAL STORM RAINFALL AND RUNOFF
According to storm characteristics, topography and other meteorological factors (such as maximum 24 h precipitation, number of days with
rainfall of 80 mm, annual precipitation and the precipitation in
early and late flood periods, mean flow patterns in early and late
flood periods, typhoon tracks, etc.), the island may be divided into
three regions:
I
southwestern heavy rainfall region;
II southeastern heavy rainfall region; and
III northern average rainfall region.
Four stations in the central mountainous area seriously affected
by deforestation were selected to study the effects of deforestation
on the annual variation of storm rainfall and runoff:
2
Region
I:
?
Basia (75.3 km ) and Fucai (508 km " ) , located upstream
H 24
(mm)
1950
60
70
1950
60
70
80
MAOZHI QM
WENGCHANG
CHENGPO
H 24
(mm)
1950
60
70
QUNHAI
H 24
(mm)
80
1950
400
H 24
(mm)
200
1950
60
60
70
LINGS -I III
ill m
1J
rf
1950
60
70
YAXIAN
400
H 24
(mm)
200
400
H 24
(mm)
200
1950
60
70
1950
1960
60
70
FIG.2
Histograms and 5-year
Qm at each
station.
70
80
moving
averages
for
H24 and
252 Qian
Wangcheng
on Nandu River; Maozhi (610 km ) , located upstream on
Changhua River;
Region II:
Chengpo (727 km located upstream on Wanquan River.
The maximum 24 h rainfall H 2 4 and peak discharge Q m of each
station were calculated and are shown in Fig.2; their 5-year moving
average is also calculated and shown in Fig.2.
For comparison, eight precipitation stations in the coastal region
only slightly affected by deforestation were selected:
Region I:
Yinggehai, Dongfong (Basuo) and Daxigiao;
Region II:
Wencheng, Qunhai (Jiaji), Wanning reservoir, Lingshui
and Yaxian.
The maximum 24 h rainfall H24 and 5-year moving average rainfall
of these stations were also calculated (Fig.2). From Fig.2 it can be
seen that in both regions the annual variations of H04 are in phase
and have a similar range and cycle. This indicates that the annual
variations of storm rainfall are not obviously affected by deforestat ion.
In region II, most of the storm rainfall in the early flood period
is produced by frontal troughs coming from the east, and that in the
late flood period produced mainly by typhoons passing over Hainan
Island or to the south of the island (north of 17 N ) . This is also
one of the major convergence zones of typhoons and cold air. In
region I, storm rainfall is mainly produced by typhoons. After
crossing longitude 110 E, the typhoon air currents passing over
Leizhou Peninsula or the north of Hainan Island on their way to Beibu
Bay, are forced to rise steeply over the Jianfengling range of
mountains, and this brings about extraordinarily heavy storm rainfall.
Most maximum 24-h rainfalls were caused by typhoons. For Fucai
stations flood peak discharges of 2990 and 2710 m s - 1 corresponding
to maximum 24 h storm rainfalls of 417 and 300 mm resulted
respectively from typhoon no. 7703 and typhoon no. 6311. These
were the annual maximum values in 1977 and 1963 respectively.
It may be seen that, the annual variation of maximum 24-h storm
rainfall of the above-mentioned station is closely related to the
variation of atmospheric circulation, especially the moving track of
typhoons. Deforestation could not change the atmospheric circulation
of Hainan Island.
On Hainan Island flooding is produced by storm rainfall. From
Fig.2 it can be seen that the annual variation of peak discharge Q m
of four of the stations corresponds well with that of maximum 24-h
storm rainfall. The coefficient of correlation r is calculated to be
about 0.8. (Baisa r = 0.89; Fucai r = 0.82; Maozhi r = 0.79; Chengpo
r = 0.84). Therefore, combined with the annual variation of maximum
24 h storm rainfall, it may be concluded that there is no obvious
relationship between deforestation and variations in annual runoff.
Rainfall-runoff
relationship
During storm r a i n f a l l , t h e e f f e c t s of f o r e s t cover on the r a i n f a l l runoff r e l a t i o n s h i p are c h i e f l y t h e i n t e r c e p t i o n of r a i n f a l l by
branches and l e a v e s , the a b s o r p t i o n of r a i n f a l l by t h e l a y e r of
l i t t e r and t h e changes in s o i l p e r m e a b i l i t y and s o i l s t o r a g e c a p a c i t y
due to the a c t i v i t y of t h e root system. In a s i n g l e storm, the
i n t e r c e p t i o n may be very s m a l l , while t h e a c t i v i t y of t h e root system
Effects
of deforestation
on flood
characteristics
253
may be going on through the soil regime. The absorption by the
litter depends upon its thickness and local climatic factors. The
effect depends not only on the tree species and its growing state,
but also to a limited extent on any antecedent precipitation. If
continuous rainfall occurs before a storm, the litter and soil will
be saturated, and runoff will be greater. On the other hand if there
is no rainfall for a long period of time, the soil moisture and the
water content of the litter will be low due to évapotranspiration.
There will be larger pore spaces for absorption and runoff will be
less. On Hainan Island, the litter layer is not very thick and the
storm rainfall is heavy.
The observed maximum 24 h rainfall of all stations is >300 mm.
For regions I and II, maximum 24 h rainfalls >500 mm occur quite
often. At Jianfengling, H 2 4 = 777 mm (8 September 1963) and at
Qilingchang H24 = 783 mm (13 June 1974); therefore, even if there is
no rainfall for a long period of time, the amount of rainfall intercepted and retained by the forest cover is small.
In order to ascertain the full effect of deforestation on the
rainfall-runoff relationship, the total loss of precipitation in a
river basin after a long dry period, I was analysed. In addition
to the above-mentioned four stations, four other stations, Dawang
(337 k m 2 ) , Xinzheng (72.6 k m 2 ) , Maowan (14.9 km 2 ) and Shirang
(6.96 km ) were also considered (Fig.l).
The storms selected for analysis all occurred after long dry
periods and the resulting runoff was small because the interception
and retention capacity of the forest cover was very high. The
criteria are as follows:
Antecedent precipitation index (index representing soil moisture
before the storm) Pa 530 mm, calculated by
30
Pa = Zl
.
P t K*
where K = 0.9, P-j- = average precipitation of the basin on the t-th
day before the storm; t - calculated up to the thirtieth day before
the storm.
The average storm rainfall of the P â 80 mm, and the rainfall at
each station in the basin should not be much less than 80 mm.
The coefficient of runoff a = R/P 230%, where runoff depth R is
calculated from the flood hydrograph after deducting the baseflow.
The baseflow is the minimum daily flow during the month before the
rise of flood.
Since most of the water entering the stream in the form of shallow
groundwater by infiltration has already been included in the runoff
depth R, the value of I m is calculated by
Im = ? + p a - R
where I m includes total losses of precipitation - such as interception,
depression storage, absorption of soil and litter and évapotranspiration in flood periods.
Based on a preliminary classification for the Guangdong province
by incorporating the values for neighbouring stations in the same
storm, and making a comprehensive study of the relationship between
P + P a and Im it may be concluded that the eight stations considered
254 Qian
Wanucheng
belong to the same type.
To compare the variation of I m before and after deforestation,
P + P a and I m data are used in power correlation analysis in three
separate cases. The results are as follows:
curve I : all
data
I m = 2.228 (P + P a ) !
curve
r
= 0.899.
r = 0.948
II : data for 1968 and prior
to 1968
0.897
r = 0.947
i .090 ( p + p a r
curve III : data for 1969 and after
1969
I m = 3.319 (P + P a ) 0 - 7 1 G .
r 2 = 0.872,
r = 0.934
All correlation coefficients are close to 0.95.
P+Pa
[mm]
200
p*p, I m (mm)
mm) I
Error7°
n m
I
II
70 65.2 63,0 69,4 -34 • 6 4
100 86,6 85.4 89,0 -14 +35
120 100199.7 102,1 -0A +20
150 [119.6 120,6 119,7 'OB +0,1
200 1503 154,0 1W1 +25 -21
1
2
3
4
5
SHIRANG
MAOWEN
XINCHENG
DAWANG
MAOZHi
Data of 1968 and
before 1968
6 CHENGP0
? FUCAI
8
Data of 1969 and
after 1969
BASAI
-4— U m n ]
50
50
FIG.3
Relationships
100
150
between
P +P
and I
Figure 3 shows that within the range of observation, the curves
are all very close to each other. The error of curves II and III is
less than ±5% with respect to curve I. It may be concluded that
after deforestation I m shows no notable difference from that before
deforestation. It clearly shows that even for storms occurring
after a long dry period, the effect of deforestation on the rainfallrunoff relationship is not significant.
CONVERGENCE OF RUNOFF
The convergence of flood flow may be roughly divided into two stages:
convergence of sheetflow and convergence of the stream network. The
effect of retarding runoff by forest cover occurs chiefly at the
sheetflow stage. Therefore, a small basin where sheetflow was the
dominant effect was selected for analysis, to find the effect of
deforestation on the convergence of flood flow. In order to obtain
fairly corresponding records of rainfall and runoff, Shirang station
was chosen for the analysis.
The drainage area of Shirang station is 6.96 km , with main river
Effects
of
deforestation
on flood
characteristics
255
length of 4.07 km and main river gradient of 9.2%. The mean elevation
of the basin is 371 m, and the terrain inclines from west to east.
Thirty-eight per cent of the total areas has an altitude above 400 m.
The soil of the basin may be roughly divided into three groups:
yellow soil, red soil and paddy soil. The soil forming rocks all are
granite. Yellow soil is mainly distributed over the areas above
400 m, and comprises a layer of sandy clay over a layer of clay. Red
soil is mainly distributed over hilly land with elevation below 400 m,
and comprises a layer of gravelly sandy loam over a layer of light
loam. Paddy soil is distributed over the paddy fields cultivated to
a depth of 20-30 cm, and comprises fine sandy loam and silty loam.
The total basin has thick soil layer, no bare waste slopes, and
slight soil erosion.
The investigation results for 1960-1965 are shown in Table 1. No
data are available after 1965.
TABLE 1
(in km2')
Land use
Paddy
fie id
Dry land
Meadow
Scrub
Forest
roa ds
Villages,
distribution
from
I960
to
1965
i960
1961
1962
1963-64
0, .820
0. .704
0. .854
i ..51
3. .03
0 .042
0. .820
1 ..40
0. .858
± ..51
2 ..33
0 .042
0. .820
1. .50
0. .858
2 ..13
1 ..61
O..042
O..820
1 ..00
O..858
2 .62
1 .61
0 .042
1965
0. .820
0. .518
1 ..11
2 ..89
i
± ..58
O .042
Observations at Shirang station began in July 1959 and ended at
the end of 1971. All precipitation stations in the basin are selfrecording. The distribution of the stations was improved after 1965.
The rainfall-runoff relationship is often nonlinear. In general
with increasing precipitation intensity, the peak of the unit hydrograph increases, and the duration of the unit hydrograph decreases.
By using similar storms for the analysis, the interference of the
nonlinear effect can be eliminated and the analysis made easier.
The method of analysis is to select the storms for the 1960's and
1970's with similar precipitation and peak flow, solve the unit
hydrographs separately by use of the method of least squares and
compare them with each other. The storms selected for analysis are
listed in Table 2, and the corresponding unit hydrographs are shown
in Fig.4.
From Table 2 and Fig.4 it may be seen tbat the peak flows differ
only slightly. The peak of storm 65724 is slightly less than that
of storm 71930. The peak flows of all the other storms in the 1960's
are slightly greater than those of the 1970's. With regard to the
time to peak flow, two storms of the 1970's, storms 70929 and 71930,
occur slightly earlier than the storms of the 1960's, but apart from
these all other storms occur at the same time.
The above analysis indicates that the effect of deforestation on
the convergence of runoff, at least after a definite period of time,
256
Qian
Wangcheng
5
FIG.4
5
10(h)
Comparison
of
unit
10(h)
hydrographs.
TABLE 2
Storm
no .
p
a
(mm)
P
(mm)
63924
67909
70929
HO.O
110.O
77 .9
61.9
73.5
93.6
621007
70819
HO.O
82.4
44.8
49.2
17 .8
65724
71930
IIO.O
96.0
651118
71929
§F
28 .2
23.9
- maximum
mean loss
o
(mm)
R
(mm)
Ji
8.5
28.3
41.6
49.4
23.2
24.2
27 .6
8.3
11.1
14.2
11.1
7 .2
35.9
25.5
0.2
0.0
15.7
15.8
13.5
12 .4
9 .83
1 .82
67.1
78 .8
7.0
15.7
18.1
6.3
10.6
10.7
7.89
8 .88
J
14.1
6.5
rainfall
rate of
excess
later
-m+
(mm h~
intensity
stage.
h~l)
(mm
3.68
2 .32
3.01
15.6
11 .1
during
Qm
(m s'1)
Um ^
29 7
32 6
33 7
1 .195
1 .098
1 .02 6
9
7
1 .285
1 .200
14
14
interval
11 7
12 . 6
8 .96
9 19
3
(m s
- 1l
)
0.726
0.822
0.800
0.650
At
is insignificant. The interception of branches and leaves is mostly
eliminated by évapotranspiration. A part of the absorption water is
a component of the convergence of groundwater flow and runoff.
Nevertheless, the difference between absorption capacity before and
after deforestation is small, and it cannot substantially affect the
regulation of flood flow, especially in the case of continuous
antecedent rainfall or storms with heavy rainfall.
CONCLUSION
Comparing (a) annual variations of maximum 24-h precipitation and
Effects
of deforestation
on flood
characteristics
257
peak flow at four stations in the central hilly region where
considerable deforestation has occurred, with those of eight stations
in the coastal region only slightly affected by deforestation; (b)
total precipitation losses at eight stations in the hilly region in
the I960's with those in the 1970's; and (c) unit hydrographs for
Shirang station in the I960's with those for the 1970's for similar
storm rainfalls, it may be concluded that because Hainan Island is
situated in the humid tropical region with heavy storm rainfall,
deforestation does not affect the atmospheric circulation. Scrub
grows rapidly after deforestation, therefore deforestation has no
notable effect on flood characteristics.
Forest plays a role in preventing soil erosion, conserving water
resources, and regulating streamflow, as well as retarding storm
runoff to some extent, therefore emphasis must be placed on forest
conservation and afforestation.
ACKNOWLEDGEMENTS
The author gratefully acknowledges the assistance
of the staff of Compilation Office of Atlas of PMP of Guangdong
Province for providing storm rainfall data and analyses.