See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227316197

Improved version of BTOPMC model and its application in event-

based hydrologic simulations

Article in Journal of Geographical Sciences · February 2007

DOI: 10.1007/s11442-007-0073-2

CITATIONS READS
17 274

4 authors:

Guoqiang Wang Maichun Zhou

Beijing Normal University South China Agricultural University
70 PUBLICATIONS 692 CITATIONS 58 PUBLICATIONS 544 CITATIONS

SEE PROFILE SEE PROFILE

Kuniyoshi Takeuchi Hiroshi Ishidaira

University of Yamanashi University of Yamanashi
200 PUBLICATIONS 3,912 CITATIONS 170 PUBLICATIONS 2,297 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Response to Reviewer's comments on HESSD paper titled "Indicators of Necessary Storages for Flood and Drought Management: Towards
Global Maps" View project

Global Center Of Excellence (GCOE) View project

All content following this page was uploaded by Guoqiang Wang on 26 September 2014.

The user has requested enhancement of the downloaded file.

Journal of Geographical Sciences

© 2007 Science in China Press Springer-Verlag

DOI: 10.1007/s11442-007-0073-2

Improved version of BTOPMC model and its

application in event-based hydrologic simulations
WANG Guoqiang, ZHOU Maichun, TAKEUCHI Kuniyoshi,
ISHIDAIRA Hiroshi
Department of Civil & Environmental Engineering University of Yamanashi, 4-3-11, Takeda, Kofu, Yamanashi
400-8511, Japan

Abstract: In this paper, a grid-based distributed hydrological model BTOPMC (Block-wise

use of TOPMODEL) is introduced, which was developed from the original TOPMODEL. In
order to broaden the model’s application to arid regions, improvement methodology is also
implemented. The canopy interception and soil infiltration processes were incorporated into
the original BTOPMC to model event-based runoff simulation in large arid regions. One de-
signed infiltration model with application of time compression approximation method is
emphasized and validated for improving model’s performance for event hydrological
simulations with a case study of Lushi River basin.
Keywords: BTOPMC; infiltration; time compression approximation; interception

1 Introduction
TOPMODEL developed by Beven and Kirkby (1979) is widely used for hill slope hydro-
logical simulations. Based on the basic concepts of TOPMODEL, a Block-wise use of
TOPMODEL (BTOPMC) was developed by the University of Yamanashi for the hydrologi-
cal modeling in large river basins (Ao, 2001), in which MC represents the Muskin-
gum-cunge routing method. Like in TOPMODEL, the runoff generation in BTOPMC model
is based on saturation excess mechanism. Though this approach for runoff generation works
well in humid regions yet it performs poor for arid catchments as the runoff generation
mechanism are different in arid regions.
In this study, in order to improve the application of the BTOPMC model for event-based
hydrological simulations in arid catchments, Leaf Area Index (LAI) based canopy intercep-
tion model and time compression approximation (TCA) based infiltration model were in-
corporated into the model structure. A practical importance of this study on the infiltration is
that by using TCA method, the atmosphere controlled infiltration conditions can be switched
to soil controlled conditions (ponded infiltration). This work demonstrated that infiltration

Received: 2006-08-01 Accepted: 2006-11-12

Foundation: 21st Century Center Of Excellence (COE) Program supported by the Ministry of Education, Culture, Sports,
Science and Technology of Japan
Author: Wang Guoqiang (1978–), Ph.D., specialized in hydrological modeling. E-mail: [email protected]

www.scichina.com www.springerlink.com
74 Journal of Geographical Sciences

could be calculated easily with some accuracy while keeping high computation efficiency.
Liu (1998) and Salvucci (1994) discussed the relation of TCA to one rigorously defined flux
concentration relation and used the latter to demonstrate the validity of TCA. They found
that the largest error of TCA in calculating cumulative infiltration is only 2.5%. In this study,
some modifications have been done to the traditional TCA method to make it capable for the
infiltration simulations during erratic rainfall storms without additional parameters included.

2 Description of BTOPMC model

The Block-wise use of TOPMODEL (BTOPMC) is a grid-based distributed hydrological
model developed for large river basins, in which the core modules are BTOP model for run-
off generation and Muskingum-cunge routing method for routing the flow. According to the
concept of BTOPMC, the basin is divided into several blocks or sub-basins considering of
the basin’s scale and the heterogeneity of basin’s topography. In BTOPMC model, runoff
generation is based on the concept of TOPMODEL, which assumes that the ground water
flow is driven by the surface topographic gradient.
Although the basic mathematical equations are kept unchanged, several improvements
have been done to the BTOPMC model, which enable its application for large basins. One
important improvement is that groundwater flow from each grid-cell is proportional to ex-
ponential of saturation deficit of the grid cell. In BTOPMC model, the topographical index
was redefined and a groundwater dischargeability parameter D is introduced into the model,
replacing the transmissivity T0 of the original TOPMODEL. In the original TOPMODEL,
the base flow qbi of hill slope i can be represented as riai, in which ri is homogeneous re-
charge rate and ai is the hill slope area. While in BTOPMC model, the base flow should not
only reflect the groundwater discharge from the most downstream cells, but also upper grid
cells, therefore, the groundwater flow relative to its catchments can be interpreted as:
qbi = rk ⋅ f ( ai ) ⋅ ak (1)
where qbi is the specific base flow from grid-cell i to the stream segment in the grid cell, rk is
the mean areal groundwater recharge over the block k, ak is the runoff coefficient of that
block and f(ai) is an effective contributing area fraction to the catchment above, which is a
function of real upstream catchment area ai.
Like TOPMODEL, the BTOPMC model is treated in three zones vertically, that is, root
zone, unsaturated zone and saturation zone (Figure 1).

2.1 Root zone

In the original TOPMODEL, all rainfall is assumed to infiltrate into the root zone until its
storage capacity is reached. The actual evapotranspiration, ET, is calculated as either PET or
the water storage in the root and unsaturated zones plus rainfall:
ET (t ) = min{PET (t ), S rz (t − 1) + P (t ) + Suz (t − 1)} (2)
S rz (t ) = S rz (t − 1) + P (t ) − ET (t ) if 0 ≤ S rz (t ) ≤ S r max (3)
where P(t) is the gross rainfall, Srz(t) and Suz(t) are the soil water in root and unsaturated
zones respectively (mm), Suz expresses the part above field capacity and Srmax is the root
zone water storage capacity (mm), estimated by defining appropriate root depth, field capac-
WANG Guoqiang et al.: Improved version of BTOPMC model and its application 75

ity and wilting point from the grid land cover and soil texture types.

Figure 1 Schematic diagram of the BTOPMC model, where P denotes gross rainfall, ET0 interception evapora-
tion, Imax interception storage capacity, I interception state, Pa net rainfall on land surface, ET actual evapotran-
spiration, Srmax available water storage capacity of root zone, Srz soil moisture state in root zone, SD and Suz soil
moisture storage deficit and state in unsaturated zone, qo overland runoff, qv groundwater recharge and qb
groundwater release

2.2 Unsaturated zone

The unsaturated zone also refers to gravity drainage zone where soil moisture is higher than
field capacity but not saturated. As shown in Figure 1, the inactive part reflects the field ca-
pacity and is static, while the active part (or drainage store) receives excess water from root
zone which changes with time:
EX (t ) = [ Srz (t − 1) + P(t ) − ET (t )] − Sr max (4)
and Suz(t) can be derived from the equations as:
⎧ S (t − 1) EX (t ) ≤ 0
Suz (t ) = ⎨ uz (5)
⎩ Suz (t − 1) + EX (t ) EX (t ) > 0
where EX(t) is the storage excess in root zone.

2.3 Saturation zone

Soil moisture content in this zone does not change. However, it receives recharge from un-
saturated zone at qv(t) and releases a base flow at qb(t).
qv (t ) = K 0 exp [ − SD(t ) m ] if qv ( t ) ≤ Suz (t ) (6)
and
76 Journal of Geographical Sciences

( dl ) T0 exp [ − SD(t ) m] tan β

qb (t ) = (7)
1000A
where K0 is the saturated conductivity at land surface, T0 is the lateral transmissivity of soil
profile when saturated to surface, dl is the effective length (m) orthogonal to groundwater
flow at the grid cell, assumed the same direction as surface flow, A is the area of grid cell
(m2), m is a parameter (mm) describing the exponential decay of lateral transmissivity of the
soil column with its storage deficit, SD(t) (mm), tanβ is the hydraulic gradient of saturated
zone (m/m), approximated by the surface slope (Beven et al., 1995). K0 and T0 are related as
K0 = T0/ds at the grid cell, where ds is the effective distance of surface flow over the cell (m)
(assuming equal to A/dl). The values of T0 (or K0) and m are calibrated for each block. The
calculation of dl and ds is based on the flow direction algorithm. In this research, the single
flow direction algorithm is used (Ao, 2001). The storage deficit at the grid cell is calculated
as
{
SD ( t ) = SD ( t ) + m γ − ln ⎡⎣ a (T0 tan β ) ⎤⎦ } (8)
where γ is the block average of soil-topographic index, ln[a/(T0 tanβ)], and the block aver-
age of storage deficit is obtained from
{
SD(t ) = SD(t − 1) − qv (t − 1) − qb (t − 1) } (9)
where qv (t − 1) and qv (t − 1) are averaged values of qv(t − 1) and qb(t − 1) in the block.
The initial value of SD(0) is first given subjectively and then determined by a warming-up
running of the model in calibration, and the results of previous running the model in the op-
eration simulation.
The water storage in unsaturated zone is updated after the percolation as:
Suz ( t ) = Suz ( t ) − qv ( t ) (10)
The overland runoff occurs when the storage from Eq. (10) exceeds its deficit
qo ( t ) = Suz ( t ) − SD ( t ) if qo ( t ) ≥ 0 (11)
The storage state is updated again:
Suz ( t ) = Suz ( t ) − qo ( t ) (12)
The concept of ‘MC’ in the BTOPMC structure means the Muskingum-cunge routing
method, which was developed by Cunge (1969).
∂A ∂Q ∂ 2Q
+c = μ 2 + cqL (13)
∂t ∂x ∂x
where A is the channel cross section area, Q is the discharge volume, μ is diffusion coeffi-
cient and qL is lateral water inflow, and c is the discharge celerity, which can be solved by
considering c = dQ / dA together with Manning’s equation as:

c = 1.7Q 0.4 S 0.3 n −0.6 B −0.4 (14)

where S is grid slope, n is manning roughness and B is the channel width.
Takeuchi (1999) introduced the BTOPMC model for the application in large ungauged
river basins, and summarized that the BTOPMC model has the advantages of both lumped
and distributed models. It has relatively few parameters requiring calibration. The parame-
WANG Guoqiang et al.: Improved version of BTOPMC model and its application 77

ters for the model have physical interpretations, representing the effects of topography,
vegetation (Srmax), soil properties (T0, M, n0). BTOPMC is also a GIS based model, which
abstracts the spatial data from DEM and relates parameters to basin features. Like
TOPMODEL, BTOPMC is a saturation excess based hydrological model. Therefore in wet
regions, it performs well, but in arid regions, the performance is poor. To solve this problem,
it is necessary to incorporate a simple infiltration method preferably with few parameters,
though improving its performance. At the same time the rainfall interception by the vegeta-
tion cover should also be considered, especially during hydrologic events.

3 Canopy interception modeling

Canopy is the first-layer interface between soil and atmosphere. It works for both intercep-
tion and evapotraspiration in water cycles. For event-based simulation, due to the short du-
ration of the storms, the evapotraspiration can be neglected, while the canopy interception is
really important for calculating the net rainfall volume in arid regions. One interception
method based on LAI is used in this study as:
I v (t ) = m ⋅ vc ⋅ LAI (t ) (15)
where Iv(t) is the interception capacity (mm), m is a coefficient, vc is the vegetation-covered
area and LAI is the leaf area index.
The deficit of canopy water storage is calculated
I vd (t ) = I v (t ) − Sc (t ) (16)
where Ivd(t) is the deficit of canopy water storage and Sc(t) is the canopy water storage. Con-
sidering the stepwise rainfall intensity P(t), which is constant within each time increment,
the interception can be interpreted as:
⎧ P (t ) if P(t) ⋅ Δt ≤ I vd (t )
I A (t ) = ⎨ (17)
⎩ I vd (t ) if P(t) ⋅ Δt > I vd (t )
where IA(t) is the actual water interception.

4 TCA based infiltration model

Time compression approximation, first used by Sherman (1947), is derived from the as-
sumption that maximum capacity at which a soil column can infiltrate at any given time
during storms depends only on the initial moisture state of the column and the cumulative
infiltration. Under this assumption, for given the initial soil conditions, the storm infiltration
capacity f(t) can be expressed as a function of cumulative storm infiltration F(t) by TCA
after the ponding time.
f (t ) = GTCA (F (t )) (18)
The actual infiltration rate during the event f ∗ (t ) is the lesser of potential infiltration
capacity and the rainfall intensity.
f ∗ (t ) = Min( f (t ), P(t )) (19)
This relation between potential infiltration capacity and cumulative infiltration can be
mapped either by experiments or analytic models. In this study, the Philip’s equation was
78 Journal of Geographical Sciences

selected to generate this relation. Philip (1957) expressed the point infiltration as:
f (t ) = 0.5St −1/ 2 + cK S (20)
where S is the soil sorptivity, which depends on the soil water potential (S denotes the initial
sorptivity), Ks is the saturated hydraulic conductivity and c is a coefficient. S can be derived
from the equation given by Smith (1993).
θs
⎛ dψ ⎞
[ S (θ i )] = 2(θ s − θ i ) ∫ K (θ ) ⎜
2
⎟dθ (21)
θi ⎝ dθ ⎠
where θ i and θ S represent the initial water content and the saturated water content re-
spectively. The K (θ ) in Eq. (21) is assumed to fit Brooks and Corey model (1964) and
ψ (θ ) is obtained from Smith’s equation (1993) as:
b+(a / λ )
⎡ θ − θr ⎤
K (θ ) = K s ⎢ ⎥ (22)
⎣θ s − θ r ⎦
1/ c
⎡⎛ θ − θ ⎞ − ( c / λ ) ⎤
ψ (θ ) = ψ b ⎢⎜ r
⎟ − 1⎥ (23)
⎢⎝ θ s − θ r ⎠ ⎥
⎣ ⎦
where ψ b is the air entry potential, θ r is the residual soil water content, a, b, c and λ
are dimensionless coefficients.
Under continuous rainfall, there are two important time points, one is the compressed
time tc, which is the time needed for the potential infiltration capacity decreasing to the
rainfall intensity under continuous ponding soil surface; another is the ponding time tp,
which happens when the actual infiltration capacity equals to the rainfall rate p (Figure 2).
By using Philip’s equation, the compressed time can be easily got by assuming the potential
infiltration capacity equals to the rainfall rate p under the condition of ponded soil surface.
tc = 4S −2 ⋅ ( p − cK S )2 (24)
The potential cumulative infiltration under this assumption can be calculated to this time.

Figure 2 Illustration of TCA for infiltration under constant rainfall

F (tc ) = cK S ⋅ tc + S ⋅ tc 0.5 (25)
Ponding time is considered to be the boundary condition for the infiltration to be domi-
nated by soil processes. Under TCA assumptions, ponding will occur at a time when the
WANG Guoqiang et al.: Improved version of BTOPMC model and its application 79

cumulative infiltration volume (due to rainfall infiltration) is equal to total effective rainfall
volume, which doesn’t consider that part of rainfall intercepted by vegetation. With the
Philip’s equation, the ponding time can be solved as:
⎡ S2
p2 ⎤
tp =
⎢ − 1⎥ (26)
4 p ⋅ cK S
⎣ ( p − cK S )
2
⎦
Once ponding occurred, infiltration occurs at a rate less than precipitation rate. We define
a relative time value (t−tp)+tc, such that the actual infiltration rate, which equals the infiltra-
tion capacity at any time greater than tp is given by
f (t ) = 0.5 ⋅ S ⋅ (t − t p + tc )−0.5 + cK S (27)
Till now, we are discussing the infiltration under constant rainfall; let’s consider the rain-
fall is defined with variability for each time step i. Under TCA, we should be able to use a
similar approach with different rainfall intensities, while during each new time increment the
ponding time should be re-calculated with variable rainfall rates.
t p ,i = [ S ⋅ (tc ,i ) 0.5 + cK S ⋅ tc ,i − F (t( i−1) )] / P(t ) (28)
The sorptivity S in Philip’s equation is a time-dependent variable, although its variability
was usually neglected. During continuous events, for each new event, it should be
re-calculated according to the initial soil moisture content, which is calculated based upon
cumulative infiltration and evaporative loss since preceding storm, but within a given storm,
we prefer to calculate the its behavior on the infiltration capacity using a time offset. The
infiltration capacity at the start of the next time increment always equals the value of infil-
tration at the end of the last time step, tf,i can be solved by relating the corresponding time
for both time increments under the ponded conditions.
f (ti +1 + t f ,i ) = f (ti − (t p ,i − tc,i )) (29)
Combining Eq. (29) with Eq. (20), Eq. (24) and Eq. (28), the offset time can be solved as:
2cK S F (t(i −1) ) + S 2 − S 4cK S F (t(i −1) ) + S 2
t f ,i = (30)
2(cK S ) 2
Finally, under stepwise variable rainfall, there are three possible situations: if Δt ≤ t p ,i , no
ponding occurs; if Δt ≤ t p ,i , ponding occurs and if tp,i < 0, it shows ponding happened at the
exact beginning. All three possible expressions can be summarized as function of the infil-
tration volume during one given time step.
ΔF (ti ) = P(t ) ⋅ Δt if P(t ) ≤ cK S
ΔF (ti ) = cK S ⋅ Δt + S ⋅ ( Δt + t f ,i − t f ,i ) if P(t ) > cK S , t p ,i < 0 (31)
ΔF (ti ) = P(t ) ⋅ t p ,i + cK S ⋅ (Δt − t p ,i ) + S ⋅ ( Δt − (t p ,i − tc ,i ) − tc ,i ) if P(t ) > cK S , t p ,i ≥ 0

5 Case study
5.1 Introduction of the basin

A case study at Lushi River basin was selected to observe the effect of the improvement in
80 Journal of Geographical Sciences

the model structure. Lushi River basin is located southeast of the Yellow River basin in
China (109º40′-110º10′E, 33º45′-34º30′ N). The basin has an area of 4623 km2 and length of
196.3 km. Lushi River basin belongs to the semi-arid region with mean annual precipitation
of 720 mm and potential evaporation of 966 mm. In this study, the topographical character-
istics of the basin were derived from a 100 m digital elevation model, hourly rainfall from
33 gauging stations and runoff data of Lushi discharge station are available from the year
1960 to 1997 (Figure 3).

Figure 3 Map of Lushi River basin

The red point on the map is the discharge station– Lushi station; the black points are the 33 rainfall stations, that
is, Lingkou, Shimenyu, Lushi, Luoyuan, Zhangping, Changmadi, Maping, Huanglongpu, Shuangmiao, Luonan,
Xinxing, Boyusi, Xunjian, Manli, Sangping, Gucheng, Miaotai, Liheng, Sanyao, Mutonggou, Dongxia, Guanpo,
Henancun, Panhe, Longqujie, Nianzigoukou, Hengjian, Liangchakou, Songpinghe, Shahejie, Jianbei, Zhulinyu
and Liuguan stations

5.2 Parameter initialization

Using TCA concept, the difficulties caused by considering temporal variation of soil mois-
ture can be avoided; however, the quantification of the initial infiltration parameters be-
comes important for the model’s performance. In this study, the infiltration parameters for
each soil type are initialized referring to the engineering properties of soil (1937) and finely
tuned. Table 1 listed the parameters used for both infiltration and interception modules.

Table 1 Values for some parameters used in the modified BTOPMC model

Parameters m vc LAI Ks θS θI θr λ ψb
Values 0–0.2 0–0.8 0–6.9 0.5–0.7 0.4–0.5 0.3–0.35 0.1–0.15 0.2 −25

5.3 Simulation results

In this case study, in order to justify the improvement of the model’s performance by incor-
porating the interception and infiltration models for arid regions, simulation results from
WANG Guoqiang et al.: Improved version of BTOPMC model and its application 81

both original BTOPMC model and improved version of BTOPMC are compared. One ob-
jective of this modification is to amplify the application of BTOPMC model for event-based
simulation, especially during continuous rainfall storms; therefore, in this study,
multi-storms in several years were selected, which have short but obvious time gap between
each two successive storms. Here example simulations for 7 storms in 3 years are shown,
that is, 3 continuous storms in the year of 1988, 2 storms in 1992 and 2 storms in 1994. Al-
though the improved version of BTOPMC model includes more parameters than the original
version, the common components of BTOPMC shared the same parameter values, which
makes this comparison more reasonable. The multi-storm hydrographs at Lushi station in
those years for both versions are shown in Figures 4-9. The parameters for infiltration mod-
ule were finely tuned using the data in the year of 1988 and validated during the simulations
in the other years.

Figure 4 Hydrograph from the original BTOPMC model at Lushi station in 1988

Figure 5 Hydrograph from the improved BTOPMC model at Lushi station in 1988

On the whole, the simulation results from the improved version of BTOPMC model with
infiltration excess and canopy interception are in better accordance with the observed data of
Lushi station compared to the original model in most of the simulated years. In this study,
we used the Nash efficiencies to validate the models’ performances in all those years. For
the cases of the simulations in those three years, the Nash efficiencies are verified to be im-
proved obviously (Table 2).
82 Journal of Geographical Sciences

Figure 6 Hydrograph from the original BTOPMC model at Lushi station in 1992

Figure 7 Hydrograph from the improved BTOPMC model at Lushi station in 1992

Figure 8 Hydrograph from the original BTOPMC model at Lushi station in 1994

Figure 9 Hydrograph from the improved BTOPMC model at Lushi station in 1994
WANG Guoqiang et al.: Improved version of BTOPMC model and its application 83

Table 2 Nash efficiencies for Lushi discharge simulation results from both versions of BTOPMC models
Versions/Year Year 1988 Year 1992 Year 1994 Average
Original BTOPMC 86.4% 78.7% 88.1% 84.4%
Improved BTOPMC 93.8% 92.3% 93.1% 93.07%

Therefore, the modification done in this study can be validated to be efficient for the im-
provement of the original BTOPMC model’s performances. While it should be mentioned
that the improvement had good feedbacks mainly in the high discharge periods, for the base
flow calculation, the modification doesn’t show obvious advantages.

6 Conclusions
The BTOPMC model has successfully amplified the applications of the TOPMODEL con-
cept in the large river basins, while it is not a hydrological modeling package. It is rather a
set of conceptual tools that can generate the hydrological behavior of catchments in a dis-
tributed way. This study improves the BTOPMC model package by introducing the canopy
interception and infiltration components.
(1) The canopy interception is considered into the model, which accounts for the rainfall
loss intercepted by plants during events and leads to effective rainfall intensities for runoff
generation.
(2) Developed from TOPMODEL, the BTOPMC model keeps the saturation-excess based
flow generation mechanism, which limits its application in event-based simulations, spe-
cially in arid regions. A TCA based infiltration model is introduced and incorporated into the
BTOPMC model, which strengthens the predictive capability of the BTOPMC model in
dealing with hydrologic events.
The prediction performance of the modified BTOPMC model is validated with a case
study in Lushi basin. 44 events, ranging from 1960 to 1997, were simulated for this purpose.
From all the selected events, the improvements in the model’s performances are observed for
36 cases. The mean value of Nash efficiencies derived for the simulated flows from the
modified BTOPMC reaches 88.4%, while the mean Nash efficiency value calculated from
the BTOPMC is only 79.3%. Through comparison, the modification approaches are verified
to be able to efficiently improve the predictive performance of BTOPMC model, specially
the designed infiltration model, which can be used for improving the prediction of infiltrated
water in arid regions. This improvement turns to be more important for hydrologic events,
since the soil surface has a high capacity to infiltrate the rainfall at the beginning of the
event.

References
Ao T, Takeuchi K, Ishidaira H, 2000. On problems and solutions of the Muskingum-Cunge routing method ap-
plied to a distributed rainfall runoff model. Annual J. Hydraulic Eng., JSCE, 44: 139–144. (in Japanese)
Beven K J, Kirkby M J, 1979. A physically based, variable contributing area model of hydrology. Hydrol. Sci.
Bul., 24(1): 43–69.
Beven K J, Lamb R, Quinn P et al., 1995. TOPMODEL. In: Signh V P (ed.), Computer Models of Watershed
Hydrology. Water Resources Publications, Highlands Ranch, CO, 627–668.
84 Journal of Geographical Sciences

Brooks K, Corey A, 1964. Hydraulic properties of porous media. Hydrology Paper No.3. Colorado State Univer-
sity Press.
Cunge J A, 1969. On the subject of a flood propagation computation method (Muskingum Method). Journal of
Hydraulic Research, 7(2): 205–230.
Hogentogler C A, 1937. Engineering Properties of Soil. McGraw-Hill Book Company.
Liu M C, Parlange J Y, Sivapalan M et al., 1998. A note on the time compression approximation. Water Resour.
Res., 34(12): 3683–3686.
Philip J R, 1957. The theory of infiltration: 1. The infiltration equation and its solution. Soil Science, 83: 345–357.
Salvucci G D, Entekhabi D, 1994. Equivalent steady soil moisture profile and the time compression approxima-
tion in water balance modeling. Water Resour. Res., 30(10): 2737―2749.
Sherman L K, 1943. Comparison of F-curves derived by the methods of Sharp and Holtan and of Sherman and
Mayer. Eos. Trans. AGU., 24: 465–467.
Smith R E, Corradini C, Melone F, 1993. Modeling infiltration for multistorm runoff events. Water Resour. Res.,
29(1): 133–144.
Takeuchi K, Ao T Q, Ishidaira H, 1999. For hydroenvironmental simulation of a large ungauged basin: introduc-
tion of block-wise use of TOPMODEL and Muskingum-Cunge method. Hydrological Science Journal, 44(4):
633–646.

View publication stats