Analysis of flood inundation in Cambodia's Mekong Delta

農業農村工学論文集 Research Paper
Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
9 | P a g e
A Hydrologic Analysis on Inundation in the Mekong Delta, Cambodia
KHEM Sothea *, GOTO Akira ** and MIZUTANI Masakazu **
* United Graduate School of Agriculture Sciences, Tokyo University of Agriculture and Technology,
Utsunomiya University, Mine Machi 350, Utsunomiya city, Tochigi Pref. 321-8505, Japan
** Faculty of Agriculture, Utsunomiya University, Mine Machi 350,
Utsunomiya city, Tochigi Prefecture, 321-85005, Japan
Abstract
Since the watershed of the Mekong River is situated under the typical monsoon climate, the Mekong River presents
a distinct variation in its water levels between the dry and rainy seasons, and causes heavy inundation in the Delta area
along the river every year. In order to analyze the inundation process and for agricultural practice, this study tried to
establish three models: the Mekong Runoff Model, Tonle Sap Lake Water Balance Model and Delta Water Balance
Model. The Tank Model consisting of 4 columns and 3 layers with a ground water layer was employed to calculate the
runoff of the Mekong River and its sub catchments from Pakse to Kompong Cham. It was also used to estimate the
inflow to the Tonle Sap Lake from its catchments. The daily water balance of the lake was formulated based on the
relationship between the storage volumes and water levels of the lake surface. The deltaic area was divided into four
zones, and the water balance of each zone was formulated by considering the zone’s inflow/outflow between rivers and
flooding areas. A combination of these three models could provide a basic framework for modeling the Delta
inundation.
Key words: 4 columns Tank Model; Water Balance; Tonle Sap Lake; Flood inundation; Mekong Delta
1. INTRODUCTION
1.1 Background
The territory of Cambodia belongs mostly to the
watershed of the Mekong River, which is the largest river
in Southeast Asia. Since the watershed of the Mekong
River is situated under the typical monsoon climate, the
Mekong River shows high seasonal variation in water
level (or discharge flow rate) between the dry season and
the rainy season, and it causes heavy inundation in the
Delta area along the river every year. For example, at
Phnom Penh, the difference between the minimum and
maximum water level reaches as much as 8-10 meters. The
most unique hydrologic feature occurs during the flood
season from May to October, when massive floodwater
from the Mekong River flows into Tonle Sap Lake and
swells the lake surface up to five times its normal size.
This long and deep inundation is forcing the agricultural
activities in the Delta to be limited. Modeling of runoff
from catchments, river flow in channels, and analysis of
inundation processes on the floodplain can contribute to
both security and economy in lowland communities.
Identifying the inundation processes has been attempted
for many years, and still has not been completely attained.
To accomplish this goal, this study tried to establish three
basic models that could represent the inundation process
and flow phenomena of the Mekong Delta in Cambodia.
These three models are: the Mekong runoff model, the
Tonle Sap Lake water balance model, and the Delta water
balance model. Over the last four decades, several
advanced models have been developed for the Mekong
River to simulate hydrological processes, such as flood
control, water balance and related water resources
development, by various organizations and consultants.
Carbonnel, et al. (1962-63) conducted a research project
on the sedimentology and hydrology of Tonle Sap Lake.
Though this project provided useful data and information,
up-to-date topographic, hydraulic and hydrologic data are
still necessary for accurate calculation of the current water
balance. The first mathematical model for the Mekong
Delta was developed by SOGREAH/UNESCO (1964) to
describe the interaction of the flows into the Mekong Delta.
Although this model is no longer used, its topographic and
hydraulic data remain valuable. On the other hand,
Kazama et al. (2002) presents the combinations of a one-
dimensional (1D) model for river flow routing and a
two-dimensional (2D) model for inundation depth of the
Mekong Delta, Cambodia. Although many models have
been developed for flood prevention, the capabilities of
those models are still restricted. At the mean time, the
series tank model is widely used in the Mekong watershed
to estimate rainfall-runoff with satisfactory results
(Sugawara M. 1974). For instance, Tatano (1999)
performed runoff analysis of the Mekong River from
Chiang Sean to Pakse and obtained good results using the
“4 columns-type” Tank Model. The present study employs
the Tatano’s approach for the Mekong runoff model from
Pakse to Kompong Cham (Kg.Cham).
1.2 Description of the Study Area
The Mekong River flows through six countries: China,
Myanmar, Thailand, Laos, Cambodia and Vietnam. Two
other rivers (the Basac and the Tonle Sap) connect to the
Mekong River at the Chaktomuk junction in Phnom Penh,
Cambodia. From Phnom Penh, this river system forms a
deltaic structure throughout the southern part of Cambodia
and Vietnam, before flowing into the South China Sea.
The whole study area is shown in Fig. 1, which covers the
Mekong River from Pakse to Kompong Cham, the entire
Tonle Sap Lake, and the Mekong Delta in Cambodia for
water balance analysis.

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
10 | P a g e
80 0 80 Kilometers
N
EW
S
T-2
T-3
T-4
T-5
T-9
I
II
IV
III
V
VI
VII
VIII
IX
T-1
MekongRiver
BasacRiver
MekongRiver
St.Treng
Pakse
Kg. Cham
T
Discharge St.
Rainfall St.
Catchment Tank Model
Watershed boundary
Sub-catchment
80 0 80 Kilometers
N
EW
S
T-2
T-3
T-4
T-5
T-9
I
II
IV
III
V
VI
VII
VIII
IX
T-1
MekongRiver
BasacRiver
MekongRiver
St.Treng
Pakse
Kg. Cham
T
Discharge St.
Rainfall St.
Watershed boundary
Sub-catchment
T-2
T-3
T-4
T-5
T-9
I
II
IV
III
V
VI
VII
VIII
IX
T-1
MekongRiver
BasacRiver
MekongRiver
St.Treng
Pakse
Kg. Cham
T
Discharge St.
Rainfall St.
Watershed boundary
Sub-catchment
The area of Stung Treng sub-catchment: 92292 km2
1995 1996 1997
Rainfall (mm) 2011 2990 2517
Discharge (mm) 1315 1877 1645
Loss between rainfall and discharge (mm) 696 1113 872
Mean ETp (mm) 1880 1712 1959
ETa ratio 0.37 0.65 0.45
Average ETa ratio {(1)+ (2)+(3)}/3 0.50
Year for available data
Description
(2) (3)(1)
Tonle Sap Lake is the largest freshwater lake in
Southeast Asia, and is a crucially important source for
agricultural production and fishery in Cambodia. The
Lake connects to the Mekong River via the Tonle Sap
River, and reverse flows occur during flood and dry
seasons between the Lake and the Mekong River. When
the water level of the Mekong River is high in the flood
season, water flows from Mekong River through the Tonle
Sap River and fills the Lake from about 2,500 km2
to
15,000 km2
and increases the volume of the lake also from
about 2.53 x 1012
m3
to 173.95 x 1012
m3
, depending on the
flood intensity. During the year, the lake’s water level
varies from 1 m MSL (mean sea level) in the dry season to
9 m MSL in the rainy season.
The selected area for the Mekong Runoff Model lies
between latitudes 100
33’ N and 160
00’ N along the
Mekong mainstream. The climate of the Mekong River is
dominated by the two monsoon seasons: the rainy season
(southwest monsoon) from May to October; and the dry
season (northeast monsoon) from November to April.
Eighty-five percent of annual precipitation falls between
the months of May and October, with the mean annual
total ranging from 1,200 mm in the southern part, to more
than 2,500 mm in the mountainous areas (northeastern
part). The rise of runoff discharges of the Mekong River
actually begins in May, and its peak flow appears normally
in September or October. The area of the Mekong Delta
focused in this study is located in the lower part of the
Mekong River, beginning from Kompong Cham in
Cambodia and continuing to wider areas at TanChau and
ChauDoc in Vietnam. The area of the Mekong Delta forms
the best land for agricultural production in Cambodia and
Vietnam, because of the abundance of arable land and
water resources. The main branch rivers in the Delta
support domestic water and irrigation water supplies and
also contribute to transport in the region. The Mekong
Delta in Cambodia comprises about 1.5 million hectares.
The Delta has a rather flat topographical feature and lies at
a low elevation varying from 0.6 to 10 meters on average.
During the rainy season, the mainstream water level
(Mekong, Basac) rises and causes flood to enter canals or
overflow over banks of rivers into the Deltaic area. In
1996, Cambodia experienced serious floods due to the
rapid flow of the Mekong River caused by storm rainfall in
the upstream countries and in the mountainous regions.
Consequently, the years of 1995, 1996 and 1997 were
chosen for this analysis.
2. MATERIALS AND METHODS
2.1 Mekong Runoff Model
2.1.1 Data collection and water balance analysis
Such hydrologic data as daily rainfall, discharge and
pan evaporation were collected from the Lower Mekong
Hydrologic Year Books from 1995 to 1997. An average
rainfall in each sub-catchment was calculated based on the
Thiessen polygons method. The water balance was
analyzed before the runoff modeling. As shown in Fig. 1,
discharge at Stung Treng, which consists of eight
sub-catchments (I-VIII), the corresponding rainfall and
pan evaporation were compared on both monthly and
yearly basis. The pan evaporation (ETo) data in Pakse was
adopted to estimate mean potential evapotranspiration
(ETp). The result of this analysis is shown in Table 1. It
indicates that annual actual evapotranspiration (ETa) was
estimated about half of the annual potential
evapotranspiration. The relation among rainfall data,
discharge data and ETa was considered to be reasonable.
Therefore, these data can be used for runoff analysis.
Fig. 1 The outline of the Mekong Watershed
Table 1 Water balance analysis at Stung Treng (T-5)
2.1.2 Description of the 4 Columns Tank Model
Sugawara (1974) proposed a series “4*4”-type Tank

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
11 | P a g e
p
QKW *=
inq
y
Q
t
W
=
∂
∂
+
∂
∂
S4
S3
S2
S1
S4
S3
S2
S1
S4 : S3 : S2 : S1
B1
A1
A2
A3
A4
z1
z2
z3
Q upper zones
Q root zones
Rainfall
ETa
h1
h2
Model for watersheds having an intense dry season.
However, according to Tatano’s application (1999), the
“4*4”-type Model had a tendency to show too much
increase in stored water in the lowest tank of the nearest
column to the river. Hence, Tatano modified the model
structure from the “4*4” to “3*4+1”-type model, and
obtained a good result through this modification. The
model consists of 13 storage tanks with four columns, in
which each column has three storage tanks. Each of the
four columns is connected with a ground water tank as
illustrated in Fig. 2. As shown in Fig. 2, rainwater is added
to the uppermost storage layers, and stored water moves to
the next storage layers or to the same level storage layers
of the neighboring lower columns. Based on water balance
analysis, the evapotranspiration (ETa) from the total basin
was approximated 50% of pan evaporation (ETo).
Evapotranspiration takes place in the uppermost storage
layers first where the input ETo is regarded as ETp, then if
the stored water there is insufficient, the deficit in
evapotranspiration is subtracted from the next layers.
Considering the resistance in evapostanspiration from
lower layers, the deficit in k-th tank (Dk) is multiplied by a
parameter of the reduction factor CER, then CER* Dk is
subtracted from k+1-the tank. The value of CER was
determined 0.65 through trial-error, so that the total ETa
should become about 50% of the total ETo. The
coefficients for the hole sizes and hole heights are the
model parameters. Though the model contains many
parameters to be determined, for simplicity and
convenience of model calibration, the same set of
parameter values can be adopted in every column. In a
non-humid basin, some parts are wet and the remaining
parts are dry. When the rainy season begins, the wet area
grows larger from a small area along the river. On the
contrary, when the dry season comes, free water in the
highest zone decreases faster than in the lower zones. To
approximate the gradually change of wet area, the basin is
divided into four zones of tank columns (S1, S2, S3, S4)
from the lowest to the highest part as shown in Fig. 2. The
sizes of the cross-sections of the columns are taken in a
geometric ratio; for instance 1:2:4:8 when r=2, or 1:3:9:27
when r=3. The ratio r is a parameter for the catchment
characteristic.
2.1.3 Model application
To separate the channel flow from the slope water
movement, an open book scheme was employed, in which
the catchment is presented as a quadrangle with a straight
channel. In this scheme, the “3*4+1” Tank Model is used
to calculate water movement on the slope, and its output is
given as uniform lateral inflow to the channel flow. The
governing equations for channel flow having a uniform
lateral inflow were written in the form of continuity and
motion equations assuming a kinematics wave flow as
follows:
(1)
(2)
Where, W is the cross-section of flow (m2
), Q is the
discharge (m3
/s), y is the distance along the channel (m),
qin is the lateral inflow per unit length as slope runoff
(m2
/s), and K and p are the kinematic wave parameters for
particular channels.
The “3*4+1”-type model, together with the kinematic
wave for channel routing sub-model, was employed to
calculate the runoff from the 116,199 km2
drainage area of
the Mekong River from Pakse to Kompong Cham. The
whole drainage area was divided into nine sub-catchments
based on the distribution of tributaries, stream gage
stations and assuming the similarity of land use or
topography, in which one Tank Model was applied to each
sub-catchment, as shown in Fig. 1. Among the nine
sub-catchments, six sub-catchments have stream gauge
stations at their outlets, which are represented by T-1, T-2,
T-3, T-4, T-5 and T-9. The other three sub-catchments (VI,
VII and VIII) have no stream gauge stations. Considering
those three sub-catchments should have similar land use
and topography with sub-catchment V (T-5), their model
parameters were assumed to be the same values as those of
sub-catchment V.
Changing wet areas
Fig. 2 The structure of the 4 columns Tank Model
For evaluating model performance of the
goodness-to-fit between the calculated and observed
discharge, the Mean Relative Error (MRE) in percentage
was used as the criterion, which calculated by the
following equation:

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
12 | P a g e
( ) (%)100
1
×
−
= ∑
obs
calobs
Q
QQ
N
MRE
Storage change
Calculated water level of
the Lake
Compared calculated water level of the Lake
with observed level at Prekdam
DEM derived from 1m
contour map
Established curve of
storage-water level relationship
Equation of storage and
water level
Estimated runoff from Pursat+Sen’s catchments,
using “3*4+1-type” Tank Model
Catchment inflow
Estimation of total inflow by
regression analysis between discharge
data and results of Tank Model
Input water level and
cross-section data
Outflow to downstream
Storage change
Calculated water level of
the Lake
Compared calculated water level of the Lake
with observed level at Prekdam
DEM derived from 1m
contour map
Established curve of
storage-water level relationship
Equation of storage and
water level
Estimated runoff from Pursat+Sen’s catchments,
using “3*4+1-type” Tank Model
Catchment inflow
Estimation of total inflow by
regression analysis between discharge
data and results of Tank Model
Input water level and
cross-section data
Outflow to downstream
0
20000
40000
60000
80000
Jan-95 Jan-96 Jan-97
Date
Discharge,(m3
/s)
0
50
100
150
200
250
300
350
400
450
Rainfall,(mm/day)
Rainfall Qobs Qcal
(3)
Where, N is the number of data, Qobs is the observed
discharge (m3
/s), and Qcal is the calculated discharge
(m3
/s). The best performance can be obtained when MRE
is equal to zero.
River cross sections were assumed to be trapezoidal for
each section of the rivers. The flow regime is assumed to
follow Manning’s resistant law. Manning’s roughness
coefficient (n) was estimated according to the Handbook
Method (referred to Chow V.T., 1959).
2.1.4 Results and Discussion
Model calibration was done for the year 1995 through
the trial-and-error approach, and the data for 1996~97
were used for model validation.
The computations were performed using a one-day
time-step, and the program was written in the Pascal
computer language. Fig. 3 shows the obtained hydrograph
of the catchment outlet at Kompong Cham (T-9). The
correspondence between the simulated and observed
hydrographs was considered satisfactory.
The parameter values and model performances are
summarized in Table 2. It shows that the MRE values are
less then 0.5 for model calibration. Consequently, the
“3*4+1 Tank Model” is considered to have the ability to
represent the watershed properly, and to be used as an
effective tool for estimating inflow to the flooding area of
the Mekong Delta.
2.2 The Tonle Sap Lake Water Balance Model
The digital elevation data was developed from the
elevation contour map with a 1-meter interval provided by
the Mekong River Commission (MRC). It was used to
identify the Tonle Sap Lake Basin and its stream network.
From the digital elevation data, the relationship curve
between storage volumes and water levels was established.
The storage change of the Lake was calculated from the
water balance of the lake, and then converted to daily
change in water level. The calculation procedure of the
catchment’s modeling and the Tonle Sap Lake water
balance model is described in Fig. 4.
Fig. 3 Kompong Cham (T-9) simulated Runoff Hydrographs
2.2.1 Estimation of daily change in storage volume
Fig. 4 Computation procedure of Tonle Sap Lake water balance model
Table 2 Summary of model performance and parameter values
Sub-catchment T-1 T-2 T-3 T-4 T-5 T-9
Area, km2
14796 3267 3613 9570 47797 23594
Upper overland flow coefficient, (B1) 0.230 0.310 0.240 0.280 0.570 0.560
Lower overland flow coefficient, (A1) 0.220 0.280 0.210 0.270 0.560 0.550
Upper runoff threshold, (h1) 40.00 35.00 28.00 40.00 30.00 30.00
Lower runoff threshold, (h2) 13.00 15.00 12.00 15.00 7.00 5.00
Infiltration coefficient in surface zone, (Z1) 0.250 0.300 0.300 0.330 0.050 0.050
Upper root zone runoff coefficient, (A2) 0.023 0.035 0.022 0.028 0.070 0.050
Infiltration coefficient in upper root zone, (Z2) 0.020 0.010 0.018 0.015 0.010 0.025
Lower root zone runoff coefficient, (A3) 0.004 0.020 0.012 0.015 0.002 0.001
Infiltration coefficient in lower root zone, (Z3) 0.020 0.010 0.018 0.015 0.001 0.005
Ground water runoff coefficient, (A4) 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
Manning's roughness coefficient, (n) 0.003 0.003 0.003 0.003 0.006 0.007
Channel width, (Bc,m) 100 100 100 100 1300 1500
MRE-Calibration period (1995) 0.480 0.350 0.280 0.240 0.220 0.230
MRE-Validation period (1996-97) 0.590 0.450 0.620 0.547 0.290 0.250

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
13 | P a g e
L
HH
n
AR
Q 01
3/2
−
=
( ) 988.60*0266.2 −= +SenPursatinflow QQ
2543.12384.00026.000001.0 23
++−= VVVH
opa ETKET *=
tETPQQVV aoutinnn ∆−+−+=+ )(1
Lake
Flooded area
Modeling catchment
Catchment area
Discharge St.
Rainfall St.
Flow Direction
Water level St.
0 8000000 km8000000
Sen-Catchment
Pursat-Catchment
Prekdam
Tonle Sap Lake
Phnom Penh
Tonle Sap RiverN
EW
S
Lake
Flooded area
Modeling catchment
Catchment area
Discharge St.
Rainfall St.
Flow Direction
Water level St.
0 8000000 km8000000
Sen-Catchment
Pursat-Catchment
Prekdam
Tonle Sap Lake
Phnom Penh
Tonle Sap Riveronle Sap RiverN
EW
S
The mass balance equation that represents the inflow
and outflow of the Lake was used to estimate the daily
change in storage volume, which express as follows:
(4)
Where, n and n+1 are the time steps, Vn is the storage
volume at time step n (mm), Qin is the inflow into the lake
(mm), Qout is the outflow/reverse flow of the lake (mm), P
is the rainfall (mm), and ETa is evapotranspiration of the
lake (mm).
Pan evaporation data at Surin and Pakse obtained from
the hydrologic year book (1995-96) was used in this
calculation. The evapotranspiration (ETa) of the Tonle Sap
Lake was determined by multiplying the reference
pan-evaporation (ETo) by a coefficient (Kp), which can be
expressed as:
(5)
Where, ETa is evapotranspiration (mm/day), ETo is
pan-evaporation (mm/day), Kp is a pan coefficient. A value
of 0.75 for pan coefficient (Kp) was obtained, based on the
Food and Agriculture Organization (FAO)’s equation for
pan-evaporation of Class-A pan evaporation.
2.2.2 Estimation of total inflows from the catchment
area
The 3*4+1 Tank Model was also employed to calculate
the inflow from the two sub-catchments (the Pursat and
Sen). These two sub-catchments are presented in Fig. 5,
where daily data records of discharge, water levels in 1995
to 1996, and rainfall for the same period are available.
Table 3 summarizes the model performance and
parameter values of these two sub-catchments. Model
calibration was done for the year 1995, whereas data for
1996 was used for the model validation. Due to
insufficient runoff data from other sub-catchments of the
Tonle Sap area, the result of regression analysis between
discharge data derived from Carbonnel’s studies in
1962-63 and the calculated discharge of the two rivers by
the Tank Model was used to estimate the total inflow from
the whole catchment to the lake. Fig. 6 shows the
performance of this analysis, in which the equation of total
inflow was derived and written as shown below:
(6)
Where, Qinflow is the total inflow of the Lake’s catchment
(m3
/s), and QPursat+Sen is the summation of the calculated
discharges at the Pursat and Sen Catchments in (m3
/s).
2.2.3 Calculation of outflow/reverse flows of the Lake
The flow of water between the Mekong River and the
Tonle Sap Lake is seasonal, and its flow direction changes
depending on the water level of the Mekong River. When
the water level of the Mekong River becomes high in the
flood season, water is pushed into the lake (reverse flow),
and when the water level of the Mekong River recedes in
the dry season, water flows from the lake to the Mekong
River (outflow). As shown in Fig. 5, Prekdam station was
regarded as an outlet point of the lake, and the
outflow/reverse flow of this station was calculated based
on the channel’s cross-section, and the water levels
between the two stations (Prekdam and Phnom Penh Port)
along the Tonle Sap River. The Manning equation was
used in this calculation, as expressed as follows:
(7)
Where, Q is the outflow/reverse flow (m3
/s) at Prekdam,
A is the cross-sectional area (m2
), R is the hydraulic radius
(m), n is Manning’s roughness coefficient, H1 is the water
level inside the reach (m), H0 is the water level outside the
reach (m), and L is the distance between the two key
stations (m).
The cross-section of the river was assumed to be
trapezoidal for open channel flow.
2.2.4 Results and discussion
Based on the analysis of 1-meter contour data of the
Digital Elevation Model (DEM) of Tonle Sap Lake, the
relationship between the storage volume and water level
(V-H) was established. By comparing with the study of
Geoff K. (2000), the equation of (V-H) relation was quoted
and written as follows:
(8)
Where, H is the water level of the lake (m), and V is the
storage volumes of the lake (m3
).
Fig. 5 The Tonle Sap Lake’s catchment model

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
14 | P a g e
0
2
4
6
8
10
12
1-May 1-Sep 1-Jan 1-May 1-Sep Date
Waterlevel(m,MSL)
Obs. Water level in Prekd Cal. Water level in the lake
R
2
= 0.7125
0
500
1000
1500
2000
2500
10 210 410 610 810 1010 1210 1410
Tank Model (Pursat+Sen), (m
3
/s)
Totaldischargedata(62-63),
(m
3
/s)
Total discharge
8000000
6000000
4000000
2000000
0
2000000
4000000
6000000
8000000
May-95 Oct-95 Mar-96 Aug-96
Month
Inflow,Outflow,Storage
change(mm,monthly)
0
20
40
60
80
100
120
Rainfall,(mm,monthly)
Inflow Outflow/reverse flow
Rainfall Storage change
Sub-catchment Pursat Sen
Area, km
2
6074 14378
Upper overland flow coefficient, (B1) 0.13 0.30
Upper overland flow coefficient, (A1) 0.11 0.26
Upper runoff threshold, (h1) 65.00 35.00
Lower runoff threshold, (h2) 0.00 0.00
Infiltration coefficient in surface zone, (Z1) 0.22 0.60
Root zone runoff coefficient, (A2) 0.00035 0.043
Infiltration coefficient in upper root zone, (Z2) 0.023 0.0003
Infiltration coefficient in lower root zone, (Z3) 0.0003 0.02
Ground water runoff coefficient, (A4) 0.00002 0.00002
Manning's roughness coefficient, (n) 0.02 0.02
Channel width, (Bc, m) 50.00 100.00
MRE-Calibration period (1995) 0.62 0.69
MRE-Validation period (1996) 0.68 0.73
Fig. 6 Regression analysis for total inflow to the Tonle Sap Lake
Water balance of the lake was computed based on the
equation of storage-water level relation combined with
evapotranspiration, rainfall, inflow and outflow of the lake.
The summarized results of the monthly changes in the
water balance components of the Tonle Sap Lake are
depicted in Fig. 7. To examine model performance,
calculated water level was compared with observed water
level of the lake. Fig. 8 presents the calculated daily
change in water level (above mean sea level, MSL) of the
lake, compared with observed water level at Prekdam. The
results indicate that the calculated water level is fairly
matched with observed one, although some discrepancies
are found, which maybe caused by inadequacy of rainfall
and discharge data. Consequently, the Tonle Sap Lake
model was proved to be a better way for seasonal water
balance, local drainage analysis and for floodplain
management.
Table 3 Runoff parameters values of the Tonle Sap Lake Catchment
Fig. 7 Calculated monthly change in Tonle Sap Lake water balance
Fig.8 Verification of the Tonle Sap Lake Water Balance Model
2.3 Delta Water Balance Model
2.3.1 The determination of the flooding area
Water level data at Kompong Cham on the upper
Mekong River, Prekdam on the Tonle Sap River, Tan Chau
on the lower Mekong River, and ChauDoc on the Basac
River were used as the boundary conditions for assessing
the water balance of the Mekong Delta in Cambodia.
These water level data were derived from the MRC’s
hydrological yearbooks from 1995 to 1997.
Cross-sectional data were obtained from the results of the
MRC’s “Updating of the Hydrographic Atlas” project. The
extent of the flooded area was determined by choosing an
area below 10-meter contour lines in the 1996 topographic
map (scale, 1:100,000).The Mekong Delta covers the main
Mekong River, Tonle Sap River and Basac River, as well
as inundated areas. These rivers network divides the whole
deltaic area in Cambodia into four zones. Fujii H. et al.
(2003) divided the flooded area into five zones, in which
the Tonle Sap area was included. The characteristic of the
four zones in this study are as follows:
Zone 1 (SD1): This floodplain is the area north of the
Mekong River constrained by the road from Prekdam to
Kompong Cham.
Zone 2 (SD2): This floodplain is intersected by a
number of roads, which in effect creates local storage
areas along the Tonle Touch River. This floodplain
continues on to Tan-Chau on the Vietnamese border.
Zone 3 (SD3): This floodplain is to the southwest and
along the right side of the Basac River, and is constrained
by national road No. 2 from Phnom Penh to Chau-Doc on
the Vietnamese border.
Zone 4 (SD4): This floodplain is in the middle of the
area constrained by the Mekong and Basac Rivers.
In this paper, only Zone 3 (SD3) was chosen for use in
describing its water balance. Fig. 9 presented the zoning
areas of the Mekong Delta and a schematic modeling of
the delta water balance model. To calculate the water level
in the inundated area, basically, two modules having
different functions were established. Fig. 10 shows the
procedure of this calculation. The first module is used to
calculate the overflow and return flow between the rivers

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
15 | P a g e
tETPQQQQSS aoutinreturnover
n
SD
n
SD ∆−+−+−+=+
)( 333423
1
( )β
γ 0WLdWLdS −∗=
Initial water level (WLin)
Dayloop
No
Yes
S-WLdcurve
Actual flood
situation
CalculateQover, Qreturn
WLd=-F(S)
Assumedfunction
F(Qover, Qreturn)
Assumedfunction
F(Qover, Qreturn)
Assumedfunction
F(S)
Assumedfunction
F(S)
First module
Secondmodule
Calculate dailychangeStorage volumeSSD
Qout3
N
Z o n in g .s h p
Z o n e 1
Z o n e 2
Z o n e 3
Z o n e 4
SD 1
SD 2
SD 4SD 3
Prekdam
TanChau
ChauDoc
Kg.Cham
WL1
WL2
WL3
WL5
WL4
Qin3
Qover23
Qover21
Qreturn21
Qreturn4
Qreturn34
Qout2Qout4
Qreturn12
Qover12
Qover32
Qreturn24
Mekong R
Mekong R
TonleSapR
BasacR
P.Penh
SD
CaFlow
Sub-zone
Water Level (WL) St.
Qover4
PrekThnot
Tonle Touch
1 0 0 1 0 2 0 K m
Qout3
N
Z o n in g .s h p
Z o n e 1
Z o n e 2
Z o n e 3
Z o n e 4
SD 1
SD 2
SD 4SD 3
Prekdam
TanChau
ChauDoc
Kg.Cham
WL1
WL2
WL3
WL5
WL4
Qin3
Qover23
Qover21
Qreturn21
Qreturn4
Qreturn34
Qout2Qout4
Qreturn12
Qover12
Qover32
Qreturn24
Mekong R
Mekong R
TonleSapR
BasacR
P.Penh
SD
CaFlow
Sub-zone
Qover4
PrekThnot
Tonle Touch
1 0 0 1 0 2 0 K m
N
Z o n in g .s h p
Z o n e 1
Z o n e 2
Z o n e 3
Z o n e 4
SD 1
SD 2
SD 4SD 3
Prekdam
TanChau
ChauDoc
Kg.Cham
WL1
WL2
WL3
WL5
WL4
Qin3
Qover23
Qover21
Qreturn21
Qreturn4
Qreturn34
Qout2Qout4
Qreturn12
Qover12
Qover32
Qreturn24
Mekong R
Mekong R
TonleSapR
BasacR
P.Penh
SD
CaFlow
Sub-zone
Qover4
PrekThnot
Tonle Touch
1 0 0 1 0 2 0 K m1 0 0 1 0 2 0 K m
and the inundated area, and determined storage change of
the inundated area. The second module is considered for
the relation between storage volume and water level (m,
MSL) in the inundated area by establishing S-WLd curve
equation. Water level of the inundated area (WLd) was
calculated based on the combination of these two modules.
2.3.2 The calculation of the water balance of the
inundated area
Considering each divided zone as a storage reservoir,
the zone inflow and outflow water balance was formulated.
By using the same mass balance equation as the Tonle Sap
Lake model, water balance in SD3 was calculated, based
on the following:
(9)
Where, SSD is the storage volume in Zone SD3 (mm), Q
over23 is the overtop bank flow from the river at Prekdam
(mm), Q return34 is the return flow from the floodplain into
the river at ChauDoc (mm), Qin3 is the inflow into SD3
from Prek Thnot’s catchment (mm), Qout3 is the outflow
from SD3 to the connected low area (mm), P is the daily
rainfall (mm), and ETa is evapotranspiration (mm). The
same calculation of the Tonle Sap model was used for ETa,
in which Kp is about 0.7. The infiltration was assumed to
be negligible. Daily rainfall (P) was formulated based on
the data from MRC’s hydrological yearbook.
Because the detail elevation data was not available in
this stage, the Storage-Water level relationship (S-WLd)
curve was estimated from a rough topography. A curve of
the S-WLd relationship can be presented showing the
relationship between the storage volume and water level of
the zone. The equation of storage volume (S) as a function
of water level (WLd) can be obtained as follows:
(10)
Where, S is the storage volume of the zone (m3
), WLd is
the water level in the zone (m, MSL), WLdo is the
minimum water level of the inundated area (m, MSL), and
γis the coefficient and β the scaling component of the
S-WLd curve equation, both of which are used as model
parameters.
2.3.3 Calculation of the overflow and return flow
The route of the overflow and return flow were assumed to
go through the colmatage canals, natural rivers, and
natural low levees. The colmatage is French-based
technology, which was dug crossing the low natural levee
of the river in order to lead flood water from the main
rivers into the back marsh area behind those
levees, where the sedimentation rises the land elevation
and eventually increases the fertile land of the natural
levee zone (Kakudo et al. 1995). The inflow into the
inundated area is considered as overflow from the main
rivers (or inflow from upper catchment) during the rainy
season, and return flow from the inundated area to the
main river occurs when the dry season starts.
By considering the water level in the flooding areas to
be equal to the water level in the canal systems, flooding
water from the main river (overflow) was calculated from
water levels of the zones and main rivers by using
formulas for hydraulic structures. Referred to Fig. 11, the
equations used in this calculation are express by:
Fig.9 Zoning and schematic modeling of the Mekong Delta
Fig.10 Delta Water Balance Model procedure

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
16 | P a g e
Inundated area
River
Q1
hstg2
hstg1
Zcr
Q2
0 asl
H SA
HR
Z
Inundated area
River
Q1
hstg2
hstg1
Zcr
Q2
0 asl
H SA
HR
Z
1111
1
2
2
3
2
stgstg
stg
stg
ghBhQ
h
h
µ=⇒








<
crhstg ZHh −=1
( )21222
1
2
2*
3
2
stgstgstg
stg
stg
hhghBQ
h
h
−=⇒








≥ µ
crlstg ZHh −=2
0
2
4
6
8
10
12
14
16
18
20
Jan-95 Jul-95 Jan-96 Jul-96 Jan-97 Jul-97 Date
Waterlevel,(m,MSL)
0
50
100
150
200
250
300
350
Rainfall,(mm/day)
Rainfall
WLcal in SD3
WLobs at Prekd (upstream)
WLobs at ChauDoc (downstream)
0
1000
2000
3000
4000
0 1 2 3 4 5 6
Water level, (m, MSL)
Storage*10
12
,(m
3
)
S-WLd curve of SD-3
Description Parameter Overflow Returnflow
Initial water level (m) WLin 0.35 0.15
Coefficient of flow in Canal µ1,2 0.20 0.91
Crest elevation (m) Zcr 8.00 7.00
Side slope of canal (m) z 1.30 1.50
Length of weir over flow/return flow (m) B 21 20
Coefficient of outflow α
Coefficient of S-WLd equation γ
Scalling exponent of S-WLd equation β
Minimum water level of inundated area (m) WLo
0.02
2.65
2.50
0.50
Fig.11 The overflow structure
For the complete overflow:
if (11)
For the submerged flow:
if (12)
in which, and (13)
Where, Q1, Q2 are the flow rates of overflow and
submerged flow, respectively (m3
/s), µ1 and µ2 are the
discharge coefficients of complete overflow and
submerged flow between the river and inundated area, g is
the acceleration due to gravity (g =9.81 m/s2
), Zcr is crest
elevation (m), hstg1, hstg2 are water stages above crest
elevation in river and inundated area, respectively(m), Hh
is higher value obtained by comparison of water level in
the river HR and inundated area (m), Hl is lower value
obtained by comparison of water level in the river and
inundated area HSA (m), B refers to the length of weir for
overflow and returned flow (m).
Fig.12 SD3 calculated water level in comparison with observed
water level between upstream and downstream
Fig.13 The verified S-WLd curve in Zone 3 (SD3)
Table 4 Determined parameter values for the Delta model
After a flood, water level in the river recedes, which
means that the water level of the flooding area is higher
than the level in the river. Consequently, return flow from
the flooding area to the river occurs. Considering this flow
in the same way as an overflow, equation (12) was used for
return flow calculation. The water balance of each zone
reflects these flows.
Due to the lack of field data to calculate the inflow and
return flow, the parameter’s values need to be assumed.
The process was examined and iterated to make the final
result a reasonable one. For instance, outflow (Qout3) from
Zone SD3 at the most downstream point was assumed to
have a strong correlation with upstream overflow (Qover23)
at Prekdam. If α is a coefficient factor of this correlation,
Qout3 can be written as: Qout 3=α* Qover 23.
Prek Thnot catchment is a tributary inflow to zone SD3.
Since its water level data record was available in this
catchment, it was converted into the inflow (Q in3) based
on the “flow rating curve” (stage-discharge equation)
established by the Ministry of Water Resources in Phnom
Penh, Cambodia.
2.3.4 Results and discussion
For the calculations of flooding inflow and return flow,
assumed values for the parameters were examined by
simultaneous equations. This process of calculation was
iterated until final solution was obtained. The model
parameters values used for the Delta model of Zone SD3
were tabulated in Table 4.
The storage change was calculated from the water
balance of the Delta zone and was converted to a water
level based on the relationship between storage volumes

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
17 | P a g e
and water level (S-WLd). The water level of Zone SD3
(assumed to be uniformly flat) was obtained based on the
S-WLd curve equation. Fig. 12 shows a result of the
calculated water level of SD3 in comparison with
observed water levels upstream at Prekdam, and
downstream at ChauDoc of the Mekong Delta. The
simulated results show that the inundation depth of SD3
ranged in between the upstream and downstream water
levels, and reached as high as about 6 meters in 1996 and
1997. The considerable variation of the range of calculated
water level in SD3 can be noticed in Fig. 12 that during the
rainy season the calculated water level (WLd) became in
between Prekdam and Chau-Doc, whereas in the dry
season calculated WLd was higher than the river’s water
level. This can be caused by the water inflow from the
outside catchments. The validity of the S-WLd
relationship curve was also examined in the same iteration
process. Fig. 13 presented the validation of the S-WLd
curve obtained from the relation between inundated area
(A) and its water level (WLd) in SD3.
Due to the lack of available data, the inundation depth
could not be compared with actual values. However,
according to the results of interview that was conducted to
the farmers in the study area, the calculated water level of
the SD3 were similar, compared to the actual flood
situation in the year 1995 to 1997. Therefore, this analysis
on water balance was considered to provide the basic
framework for modeling in the Mekong Delta.
3. CONCLUSION
The establishment of the “3*4+1 Tank Model” model
for estimating runoff of the Mekong River having the
distinct dry season was considered to have a capacity of
representing the watershed properly. The results of the
Tonle Sap Lake model were also proved to be a better way
for water balance analysis. This model is also suitable for
providing useful information on multi-functional roles of
hydrology in floodplain area of the Tonle Sap Lake. These
two models are considered to be capable of estimating the
inflow to the Mekong Delta. The Delta Water Balance
Model was also established and considered to provide the
basic framework for modeling the Delta inundation. For
more accurate modeling of the Delta water balance, it will
be crucially important to collect detailed data on
topography, as well as water level records. For the next
inundation study of the Mekong Delta, it is necessary to
consider the distribution of water on the floodplain, using
a hydraulic model and Geographic Information System
(GIS) technique for examining the possibility and
effectiveness of the semi-control of flooding in order to
improve the agricultural land conditions in the region.
ACKNOWLEDGMENTS: Our thanks to the Mekong River
Commission (MRC) in Phnom Penh and the Ministry of Water
Resources of Cambodia, for their extensive assistance in
providing necessary data for this study. The authors also express
gratitude to the Ministry of Education, Sport, and Technology,
Japan for providing scholarship in support of this study. “This
research was also partially supported by a Grant-in-Aid of
CREST of the Japan Science and Technology Agency.
REFERENCES
Carbonnel J.P., Guiscafre J. (1963): Grand Lac du Cambodge.
Sedimentologie et Hydrologie. Rapport de mission. Mekong
Secretariat, p.64-176, (in French).
Chow V. T., (1959): Open-Channel Hydraulics, McGraw-Hill, New York,
NY. p. 106.
Fujii H., Gardal H., Ward P., Ishii M., Morishita K., Boivin T., (2003):
Hydrological roles of the Cambodian floodplain of the Mekong
River. Intl.J. River Basin Management Vol. 1, No. 3 (2003), p.1-14.
Geoff K. (2000): Developing a Hydrological Model for the Mekong
Basin: Impact for Basin Development on Fisheries Productivity.
Working Paper 2. ISBN: 92-9090-424-0.
Kakudo H., Kawai T., Goto A., Mase T. (1995): Colmatage system as an
appropriate technology in Cambodia. Journal of JSIDRE. Vol.
63(4), p. 357-362 (in Japanese).
Kazama S., Yasunori M., Keiji N., Kazuya I. (2002): Study on the 2000
Flood in the Lower Mekong by field Survey and Numerical
Simulation. Proceedings of the 13th
congress the APD/IAHR, Vol.1,
p. 534-539.
MRC. (1995-97): Lower Mekong Hydrologic Yearbook. Mekong River
Commission.
SOGREAH/UNESCO: (1964): Mathematical Model of the Mekong
Delta. Flood prediction in the Cambodian Delta, Vol. 1, p. 44.
Sugawara M., Ozaki E., Watanabe I., Katsuyama Y (1974): Tank Model
and its application to Bird Creek, Wollombi Brook , Bikin River,
Kitsu River, Sanaga River, and Nam Mune, Research Notes of the
National Research Center for Disaster Prevention, No. 12, p. 1-64.
Tatano M., Goto A. (1999): Real-Time Flow Forecasting in the
Midstream Basin of the Mekong River by Combination of a
Deterministic Model and a Stochastic Model. Proceeding Annual
Meeting JSIDRE, 1998, p. 126-127, (in Japanese).
[Received 2005.4.26, Accepted 2006.2.14]
[Questions and/ or discussions on this paper for public debate will be
accepted before 2006.10.24]

Trans. J S I D R E
No. 242, pp. 9-17 (2006-02)
18 | P a g e
カンボジア領メコンデルタにおける洪水氾濫解析
ケムソティア *
後藤章**
水谷正一**
* 東京農工大学大学院連合農学研究科－宇都宮大学農学部．〒321-8505 栃木県宇都宮市峰町 350
**宇都宮大学農学部、〒321-8505 栃木県宇都宮市峰町 350
要旨
メコン川は典型的なモンスーン気候下にある河川の特徴として雨季・乾季できわめて大きな水位変動を呈する。本研究では洪
水氾濫との農業形態と関係を分析するため，氾濫過程を再現できる基礎モデルを構築した。洪水氾濫解析モデルはメコン川流
域モデル，トンレサップ湖モデル，デルタ水収支モデルの 3 つのモデルから構成される。メコン川流出モデルとして，3 段 4 列の下
に下層地下水 1 段を考慮した 3×4+1 型のタンクモデルを Pakse 地点から Kompong Cham 地点までの流域に適用した。このモデ
ルによってメコン川本流からデルタ地帯への流入量，トンレサップ湖水位の季節変動とモデル地帯への流入量，そしてデルタ地
帯での水収支などが良く再現された。最後に上のモデルから得られる流入量からデルタ水収支モデルの構築を試みた。これらに
よりデルタ地域の洪水氾濫モデルの基本構造の大枠を提示することができたものと考える。
キーワード：4 段のタンクモデル，水収支モデル，トンレサップ湖；洪水氾濫モデル，メコンデルタ

Analysis of flood inundation in Cambodia's Mekong Delta

Recommended

Recommended

More Related Content

What's hot

What's hot (8)

Similar to Analysis of flood inundation in Cambodia's Mekong Delta

Similar to Analysis of flood inundation in Cambodia's Mekong Delta (20)

Analysis of flood inundation in Cambodia's Mekong Delta