Natural Hazards (2023) 118:875–902

https://doi.org/10.1007/s11069-023-06047-1

REVIEW ARTICLE

A comprehensive review of watershed flood simulation

model

Yanxia Shen1 · Chunbo Jiang1

Received: 3 January 2023 / Accepted: 31 May 2023 / Published online: 8 June 2023
© The Author(s), under exclusive licence to Springer Nature B.V. 2023

Abstract
Flooding is a major threat that presents a significant risk to human survival and develop-
ment worldwide. Regarding flood risk management, flood modeling enables understanding,
assessing, and forecasting flood conditions and their impact. This paper gives an overview
of prevailing flood simulation models given their potentials and limitations for reflecting
pluvial floods in watershed settings. The existing models are categorized into hydrologic,
hydrodynamic, and coupled hydrologic-hydrodynamic models. The coupled hydrologic-
hydrodynamic model can be further classified into full, external, and internal coupling
models. The definitions, advantages, and limitations of each coupling model are discussed.
It is found that the existing coupling types cannot accurately reflect the flood evolution pro-
cesses. A dynamic bidirectional coupled hydrologic-hydrodynamic model is then detailed,
where the watershed is spatially divided into inundation and non-inundation regions. These
two regions are connected by a coupling moving interface. Only 2D hydrodynamic mod-
els are applied to the local inundation regions to ensure numerical accuracy, whereas the
fully distributed hydrologic model is applied to non-inundation regions to improve compu-
tational efficiency. Future investigation should focus on the development of a dynamic bidi-
rectional coupling procedure that can accurately represent the complex physical interac-
tions between upstream rainfall-runoff and the local inundation process. This paper would
help flood managers and potential users undertake effective flood modeling tasks, balanc-
ing their needs, model complexity, and requirements of input data and time.

Keywords Watershed flood simulation · Hydrologic model · Hydrodynamic model ·

Coupled model · Computational efficiency and accuracy

Abbreviations
DBCM Dynamic bidirectional coupled hydrologic-hydrodynamic model
USLE Universal soil loss equation
RUSL Revised universal soil loss equation
MUSLE Modified universal soil loss equation
Ann AGNPS Annualized agricultural non-point pollutant loading model

* Chunbo Jiang
jcb@tsinghua.edu.cn
1
State Key Laboratory of Hydroscience and Engineering, Department of Hydraulic Engineering,
Tsinghua University, Beijing 100084, China

13
Vol.:(0123456789)
876 Natural Hazards (2023) 118:875–902

SCS-CN Soil conservation service curve number

SWMM Storm water management model
SWAT​ Soil and water assessment tool
HEC Hydrologic Engineering Center
HSPF Hydrological simulation program-Fortran
TOPMODEL Topography-based hydrologic model
INCA Integrated nitrogen model for catchments
HYPE Hydrological predictions for the environment
GSSHA Gridded surface/subsurface hydrologic analysis
WAM Watershed assessment model
ANSWERS Areal nonpoint source watershed environment response simulation
HM2D Full 2D hydrodynamic model
BMI Basic model interface
LARSIM Large area runoff simulation model
HEC-RAS Hydrologic Engineering Center-River Analysis System
HEC-HMS Hydrologic Engineering Center-Hydrologic Modeling System
GEMSS Generalized environmental modeling system for surface waters
SCS-LR Soil conservation service and a lag-and-route (LR) model
EFDC Environmental fluid dynamics code
WASP Water quality analysis simulation program
WATLAC A water flow model for lake catchments
HIMS Hydroinformatic modeling system

1 Introduction

Flood has become the supreme catastrophic natural hazard with significant economic dam-
age and loss of lives. According to the statistics, in the 20 years between 2000 and 2019,
Emergency Events Database (EM-DAT) recorded 7348 disaster events worldwide. Among
these disasters, floods (44%) and storms (28%) are the most frequent, and floods (41%) and
droughts (35%) are the most affected (UN Office for Disaster Risk Reduction 2020), which
indicates that the flood has been increasingly understood as a major threat that has pre-
sented a significant risk for human life and infrastructure worldwide. With global climate
change, the possibility and uncertainty of extreme rainstorms and flood increase gradu-
ally, posing a huge threat to people’s life and property (Evans et al. 2004). Therefore, in
response to this adversity, it is of great significance to simulate the flooding process timely
and accurately to improve the flood risk prevention ability and reduce the loss of flood
disasters.
Regarding flood risk management, flood modeling provides the distribution and extent
of inundation and its dynamics, which enables one to understand, assess and forecast
flood conditions and their impact (Bates 2022). These are helpful input information for
planning to mitigate the flood and reduce its effects. Numerous hydrologic and hydrody-
namic models have been developed for simulating flood processes. Their applications have
also been realized through commercial and open-source simulation tools (Henonin et al.
2013). Hydrodynamic modeling is a state-of-the-art approach for simulating inundation
processes, but it has computational limitations in large-scale applications (Hunter et al.
2007; Yamazaki et al. 2014). In contrast, hydrologic modeling requires less computation
time while representing the physical processes with lower resolution; therefore, it has been

13
Natural Hazards (2023) 118:875–902 877

widely used to simulate overland flows in flood modeling (Liu et al. 2019). Substantial
research has been conducted to improve the computational efficiency and accuracy of flood
models by coupling hydrologic and hydrodynamic modeling (Wang et al. 2021).
Aiming these numerous models, from an application point of view, different applica-
tions require diverse kinds of information and level of accuracy (Bulti and Abebe 2020).
For example, flood disaster assessment in small watersheds relies on the precision of flow
velocity. The flow velocity should be sensibly modeled for flood disaster assessment, while
the maximum water depth and extent of inundation could be sufficient for hazard mapping.
As for the large-scale regions, the maximum water depth and extent of inundation may
be detailed. Moreover, a shorter computational time is needed for real-time applications.
However, it is well-known that a shorter computational time means low simulation accu-
racy, while satisfactory simulation accuracy requires more computation time. This puts
additional demands on various aspects of the underlying modeling approach to available
flood models.
This study aims to provide an overview of the flood simulation model in watersheds.
The characteristics and classification of hydrologic and hydrodynamic models were dis-
cussed. Based on this, the coupling of hydrologic and hydrodynamic models was pre-
sented, and their advantages and limitations in modeling watershed flood disasters were
introduced. Following this, a dynamic bidirectional coupled hydrologic-hydrodynamic
model (DBCM) was detailed and the advantages and current challenges of DBCM were
also discussed.

2 Numerical models

2.1 Hydrologic model

The hydrologic model is used to simulate the migration and transformation process of
water in the atmosphere, surface, and soil (Sitterson et al. 2017). It is usually applied at
the basin scale and has many parameters that need to be calibrated or estimated region-
ally. Hydrologic processes consist of runoff generation and routing. The runoff generation
module transforms rainfall into surface runoff via the hydrologic processes in the vertical
direction. Laterally, these runoffs are routed to an outlet point by the runoff routing. The
runoff routing schemes can be divided into 1D way (e.g., unit hydrograph methods, linear
reservoir, and 1D kinematic wave model) (Nguyen et al. 2016; Felder et al. 2017) and 2D
way (e.g., 2D kinematic wave model) (Kim et al. 2012; Li et al. 2021).
Based on the spatial interpretation of the model’s catchment area, the hydrologic model
can be divided into lumped, semi-distributed, and distributed hydrologic models (Arnold
et al. 2010). It is easy to confuse the difference between the semi-distributed and distrib-
uted models based on these classification methods. For example, some researchers defined
the SWAT as a distributed model (Zhao et al. 2013; Zhang et al. 2016), but some research-
ers considered it a semi-distributed model (Sitterson et al. 2017). Li et al. (2021) classified
the hydrologic model into process-based or fully physical models and conceptual or empir-
ical models based on the routing calculation scheme. The process-based model solves the
mathematical governing equations, while the conceptual model does not, but instead of
parsimonious expressions or simplified degrees of representations. Although there are
many ways to classify hydrologic models, not all models fit into a single category because
they are developed for a variety of purposes.

13
878 Natural Hazards (2023) 118:875–902

Table 1  Classification of hydrologic models

Models Definition

Lumped hydrologic model The computational domain is considered a unit and not discretized.
The empirical methods are used to calculate the runoff routing
Semi-distributed hydrologic model The computational domain is divided into multiple sub-catchments
and a 1D routing scheme is used to calculate the runoff routing
Distributed hydrologic model The computational domain is divided into multiple hydrologic
response units (HRUs) and a 1D routing scheme is used to calcu-
late the runoff routing
Fully distributed hydrologic model The computational domain is divided into HRUs or grids, and a 2D
routing scheme is used to calculate the runoff routing

Fig. 1  Visualization of spatial structure for each type of hydrologic model: a lump model; b semi-distrib-
uted model; c distributed model; d fully distributed model

In this work, the hydrologic model is categorized into four types based on spatial discre-
tization and its routing scheme, as shown in Table 1. Visualization of spatial structure for
each type of hydrologic model is shown in Fig. 1.
Lumped hydrologic models do not consider spatial variability within the catchment, and
the entire catchment is modeled as one unit (Singh 1995). Averaged values over the catch-
ment are used such as mean soil storage and uniform precipitation amounts (Beven 2012).
Empirical methods are used to simulate total runoff and streamflow at the outlet point in
lumped models, and no detailed information on water depth and flow is available in the
catchment. The lumped hydrologic model includes USLE, RUSL (Spaeth et al. 2003),
MUSLE (Gwapedza et al. 2021), CREAMS (Kniselw 1980), AnnAGNPS (Parajuli et al.
2009), and so on. Lumped models simulate average runoff conditions with high computa-
tional efficiency, but there are a lot of assumptions which is not ideal for flood simulation.
What is more, with this type, the individual changes exerted in one part of the catchment
cannot be accounted for spatially, and thus, assessment management within the watershed
is problematic.
In the semi-distributed hydrologic model, the computational domain is divided into
multiple sub-catchments that have similar land use, soil type, and topography-like slope
(Rinsema 2014). In each sub-catchment, the water mass is calculated by the water balance
equation. Only the average value of variables, such as the flow rate, can be obtained on
each sub-catchment. And then, the empirical equations, such as the SCS-CN curve (SCS
1972) and Muskingum equation, usually reflect the surface runoff process. HEC (Feld-
man 2000) uses the SCS curve to calculate the surface flow volume. TOPMODEL is a
semi-distributed model which uses land surface slope and soil characteristics to sub-divide
the catchment (Beven and Freer 2001). Additionally, HSPF (Becknell et al. 1993), INCA

13
Natural Hazards (2023) 118:875–902 879

(Rankinen and Lepisto 2002), and HYPE (Lindstrom et al. 2010) are all such an example.
This approach is simple and feasible easily and indicates the effect of management prac-
tices within the sub-catchments. A drawback is the manipulation of input data. For exam-
ple, spatially distributed rainfall data must be averaged within the sub-catchment, or rain
gauge data at specific locations must be distributed to the area using certain interpolation
methods.
In the distributed hydrologic model, based on overlaying maps of land cover, soil group,
and elevation, the computational domain is divided into many HRUs in which the region
within the HRU responds to rainfall the same way (Rinsema 2014). The water mass is cal-
culated by the water balance equation in each HRU. Only the average value of variables,
such as the flow rate, can be obtained from each HRU. And then, similar to the semi-dis-
tributed model, the 1D routing scheme is used to reflect the surface runoff process. For
example, the 1D nonlinear reservoir model including water balance and Manning equa-
tions is used to simulate the rainfall-runoff in SWMM (Fig. 2a) (Ada et al. 2010; Rossman
2015). SWAT uses HRUs to divide the computational domain (Arnold et al. 1998), and the
SCS curve is adopted to calculate the surface flow volume. The benefits of the distributed
model are high computational efficiency and the ability to use fewer data and parameters.
But it fails to show the interaction between the HRUs as they are not internally linked
within the watershed but are all routed individually to the basin outlet. Besides, it is not
appropriate to use a 1D routing scheme to reflect surface routing as rainfall-runoff is 2D
flow (Fig. 2c). The runoff routing schemes are often oversimplified for most of the distrib-
uted hydrologic models, which is perceived as a weakness for flood forecasts.

Fig. 2  Representation of surface runoff of a SWMM (Rossman 2015); b distributed model and c actual
processes

13
880 Natural Hazards (2023) 118:875–902

In a fully distributed hydrologic model, the computational watershed is divided into

hydraulically connected elements such as grids. Each grid has a distinct hydrologic
response and is calculated separately, but incorporates interactions with bordering cells.
By calculating runoff routing for every grid cell using a 2D scheme, the model provides
detailed runoff information at various points within the catchment (Rinsema 2014). The 2D
scheme is more consistent with runoff routing, which can obtain high simulation accuracy.
A fully distributed model demands substantial amounts of data and typically has much
larger computational requirements than lumped or semi-distributed approaches. CASC2D
(Marsik and Waylen 2006), Mike SHE (Danish Hydraulic Institute 2009) and GSSHA
(Downer and Ogden 2004; Brendel et al. 2021) are all fully distributed hydrologic models.
What is more, the grid approach is implemented in WAM (Bottcher et al. 2012), AnnAG-
NPS (Bingner and Theurer 2002), and ANSWERS (Beasley et al. 1980).

2.2 Hydrodynamic model

The hydrodynamic models are used to reflect the flood processes in rivers or 2D inunda-
tion regions, such as lakes and reservoirs. The hydrodynamic models are classified into
1D, 2D, and 3D hydrodynamic models. 3D hydrodynamic models are not widely used in
flood simulation due to their complexity. 1D Saint-Venant equations are used to represent
the 1D hydrodynamic model (Audusse and Bristeau 2007), which can be used to simulate
the flow in 1D regions such as rivers or pipes. 2D shallow water equations (SWEs) are used
to represent the 2D hydrodynamic model (Toro 2001), which is used to reflect the flooding
process in 2D inundation regions.

2.3 Coupled hydrologic‑hydrodynamic model

The hydrologic models are widely used for streamflow time series generation, but they are
rarely expanded to simulate the flood propagation processes, which require less computa-
tional time at the price of representing less detailed physical processes. The hydrodynamic
model played a major role in inundation processes, which are computationally prohibi-
tive for large-scale applications. Consequently, substantial research has been conducted to
develop flood simulation models by coupling hydrologic and hydrodynamic models (Wang
et al. 2021). Generally, there are two categories of methods for coupling hydrologic and
hydrodynamic models, as shown in Fig. 3. The first category is the full 2D hydrodynamic

Fig. 3  Classification of coupled hydrologic-hydrodynamic model

13
Natural Hazards (2023) 118:875–902 881

model, where the 2D hydrodynamic model is used to simulate the overland flow (runoff
routing and flood inundation), while only the runoff generation is calculated by the hydro-
logic model and added into the source term of the 2D hydrodynamic model. The second
category uses hydrologic modeling for upland areas and hydrodynamic modeling for main
channels and flood-prone areas, and the hydrologic model provides the discharge boundary
for the hydrodynamic model. And therefore, this category can be more efficient and thus,
applied to large-scale studies. This category can be further divided into external and inter-
nal coupling models, according to whether the hydrodynamic model provides feedback to
the hydrologic model. Flowcharts for each coupling technique are shown in Fig. 4.

2.3.1 Full 2D hydrodynamic model

In this coupling type, only the runoff generation is calculated by the hydrologic model and
considered as the source term of the continuity equation of 2D hydrodynamic model, and
then, both the overland flow migration and inundation processes are all calculated by 2D
hydrodynamic model. This coupling type is also called a full 2D hydrodynamic model
(HM2D). The HM2D can be used to simulate complex flow patterns and achieve satisfac-
tory results.
To explore the trends and development of HM2D, a topic search was conducted to cap-
ture the maximum possible amount of relevant literature using the Web of Science Core
Collection database. The terms “2D shallow water equation” and “flood” are used as the
search queries. The search results were deduplicated, sorted, and irrelevant entries were
removed, and finally, 319 relevant documents were obtained as analysis data. In these pub-
lications, England and China are the wise dominance of studies using such models, and the
proportion of their publications is 18% and 17%, respectively.
Figure 5 shows the yearly distributions of papers published. It is observed that the stud-
ies of HM2D began in the 1980s. It is an initial stage for the exploration of HM2D. After
2000, publications about HM2D are fluctuating upward. The development of computer
technology facilitated the development of the model, and the HM2D was widely used to
simulate floods in different scenarios. Typical applications of HM2D in recent years are
listed in Table 2.

Fig. 4  Flowcharts representing the coupling technique of different coupling models

13
882 Natural Hazards (2023) 118:875–902

Fig. 5  Counts of publications of HM2D research by year based on the Web of Science database

Table 2  Summary of the publications obtained that used HM2D

Reference Study areas

Liang (2010) Thamesmead near London (36 k­ m2)

Wang and Liang (2011) in the floodplain at Thamesmead (36 ­km2)
Singh et al. (2011) dam break laboratory experimental
Kim et al. (2012) Peacheater Creek watershed (64 ­km2)
Yu and Duan (2014, 2017) Goodwin Creek Watershed (21.3 ­km2)
Liang and Smith (2015) Glasgow and Thamesmead in the UK (0.39 ­km2)
Vacondio et al. (2016) Secchia River, Italy (180 ­km2)
Vacondio et al. (2017) Flood event induced by a levee breach in a long river reach (83 ­km2)
Hou et al. (2020) Fengxi New City in Xi’an (China) (56 ­km2)
Garcia-Navarro et al. (2019) Three small areas
Pilotti et al. (2020) Cancano II dam
Li et al. (2021) Hurricane Harvey in Spring Basin
Cea and Lopez-Nunez (2021) Some cases about dam-break
Costabile and Costanzo (2021) Three actual catchments (3.9 ­km2, 2.3km2, ­43km2) and one ideal-
ized basin (1.62km2)
Zhang et al. (2021) a small catchment located in Shenzhen City, China (0.35 ­km2)

1D hydrodynamic models have long been popular for reasons of their speed of cal-
culation, ease of parameterization, and easy representation of hydraulic structures in
the flow domain. Some researchers have also coupled the HM2D with 1D hydrody-
namic model, where the HM2D is used to simulate the runoff in 2D regions, while the
1D hydrodynamic model is used to simulate the flow in 1D rivers or pipes, as shown
in Fig. 6. Underground sewer flow, overflow to the surface, and return flow from the

13
Natural Hazards (2023) 118:875–902 883

Fig. 6  Diagram of a actual water flow and b coupling of HM2D and 1D hydrodynamic model

surface through the manholes are also considered bidirectionally. Mike11 and Mike21
are the typical couplings of 1D and 2D hydrodynamic models (Timbadiya et al. 2014;
Tansar et al. 2020). Some other similar models for coupling of HM2D and 1D hydro-
dynamic models can be found in Liu et al. (2015), Zhang et al. (2020), Kuiry et al.
(2010), and so on.
The rainfall-runoff was simulated by solving the 2D SWEs in HM2D, which can
better understand the full process of flood process physically. Therefore, the HM2D
has satisfactory simulation accuracy. However, as the 2D hydrodynamic equations
need to be solved in the entire watershed, the HM2D is still computationally prohibi-
tive for large-scale applications, especially in regions where the high-resolution rep-
resentation of complicated topographic features is necessary. Therefore, the HM2D is
typically applied to small and medium-sized watersheds (Table 2).
It is noted that the flood inundation region is only a part of the watershed. The 2D
hydrodynamic model can be solved in the local inundation region to ensure numerical
accuracy, and the hydrologic model can be solved in other areas (i.e., non-inundation
regions) to save computational time and resources, that is, different models are used in
different spatial region.

13
884 Natural Hazards (2023) 118:875–902

2.3.2 External coupling model

External coupling, also called one-way coupling, is the simplest and most common type.
This coupling type can be easily developed based on the available hydrologic and hydrody-
namic models, which have been widely used. A topic search was performed to capture the
maximum possible amount of relevant literature using the Web of Science database. The
terms “hydrodynamic model”, “hydrologic model” and “flood” are used as the search que-
ries. The search results were deduplicated, sorted, and irrelevant entries were removed, and
finally, 230 relevant documents were obtained as analysis data. In these publications, the
USA and China are the wise dominance of studies using such models, and the proportion
of their publications is 42% and 17%, respectively.
Figure 7 shows the yearly distributions of papers published. It is observed that the stud-
ies of external coupling model began in 1990s, and the publications show a steady upward
trend. There has been increasing research on external coupling models, especially in recent
years. The SWMM (Rossman 2015), SWAT (Arnold et al. 1998), HEC (Feldman 2000),
WASP (Ambrose and Wool 2017), EFDC (Hamrick 1992), etc., are the common models
used to develop the external coupling model. Typical applications of the external coupling
model are shown in Table 3.
The external coupling model utilizes hydrographs obtained from hydrologic models
as an input for hydrodynamic models, providing a one-way transition, as shown in Fig. 8.
Usually, the hydrologic model is run first and independently from the hydrodynamic
model. Then, the discharge is transferred using limited boundary nodes to the hydrody-
namic model, which is run with all the required input information and boundary condi-
tions. Finally, the results (i.e., water depth and flow velocity) from the hydrodynamic
model are analyzed and if further changes are necessary to the hydrologic model, the pro-
cess will repeat.
If the format, time, and space steps of the outputs of hydrologic models are inconsist-
ent with that of the inputs of hydrodynamic models, the coupling between hydrologic and

Fig. 7  Counts of publications of external model research by year based on the Web of Science database

13
 Table 3  External coupling of hydrologic and hydrodynamic models
Coupling models Models used Coupling types Study area

Dargahi and Setegn (2011) SWAT​ Directly coupling Lake Tana (Ethiopia)
GEMSS (15,096 ­km2)
Bravo et al. (2012) MGB-IPH Directly coupling Paraguay River (1,095,000 k­ m2)
HEC-RAS
Li et al. (2014) WATLAC Directly coupling Poyang Lake (China) (162,000 ­km2)
MIKE
Feistl et al. (2014) SCS-LR Directly coupling Gardon River (2040 k­ m2)
Natural Hazards (2023) 118:875–902

MASCARET
Wu et al. (2017) HIMS Directly coupling Hongfeng Reservoir Basin (China) (1596 ­km2)
EFDC
Zhang et al. (2017) SWAT​ Directly coupling Poyang Lake basin, (China) (162,000 ­km2)
Delft 3D
Bhola and Leandro (2018) LARSIM Directly coupling Kulmbach
HEC-RAS 2D
Hdeib et al. (2018) SCS-CN curve Directly coupling Awali River basin (301 k­ m2)
1D HEC-HMS
Munar et al. (2018) MGB-IPH Directly coupling Lake Mirim basin (58,400 k­ m2)
IPH-ECO
Li et al. (2019) TOPMODEL Directly coupling upper Pajiang River Basin (PJRB) (468 ­km2)
Mike Flood
Gomes et al. (2021) CAWM IV Directly coupling Capibaribe River Basin (7454 ­km2)
HEC-HMS
Debele and Srinivasan (2006) SWAT​ A separate interface program Cedar Creek watershed (5244 ­km2)
CE‐QUAL‐W2
Betrie et al. (2011) SWAT​ Open MI Blue Nile River Basin (940 ­km2)
SOBEK
Hoch et al. (2018) LISFLOOD-FP BMI Amazon River basin
Delft 3D
Shin et al. (2019) SWAT​ Matlab code Ipjang Reservoir watershed (7.2 ­km2)
EFDC
885

13
 Table 3  (continued)
886

Coupling models Models used Coupling types Study area

Shabani et al. (2021) HEC through a linkage file generated by the Hydrolink Woodbridge watershed (44.5 ­km2)

13
WASP Application Program Interface (API)
Natural Hazards (2023) 118:875–902
Natural Hazards (2023) 118:875–902 887

Fig. 8  Flowchart of external coupling technique of hydrologic and hydrodynamic models

hydrodynamic models is not performed directly, with numerical difficulties arising in this
regard. The coupled approach can show numerical instabilities due to the different spatial
discretization often used in the hydrologic model, where a coarser computational grid is
used, and in the hydrodynamic model, where a more-refined computational grid is often
applied (Chávarri et al. 2013). Therefore, the spatial positions of the input and output
nodes generally do not coincide. To solve these problems, a data file connection method is
proposed for coupling hydrologic and hydrodynamic models. The simulation results of the
hydrologic model are stored in a data file, which can automatically convert the outputs of
the hydrologic model to the input files of the hydrodynamic model to drive the operation of
the hydrodynamic model. For instance, a set of MATLAB codes were developed by Shin
et al. (2019) to convert the SWAT output into the input files of EFDC as the boundary con-
ditions, which showed the potential to be a useful tool for the water environment. Xu et al.
(2007) developed a coupling model, where the HSPF and CE-QUAL-W2 were connected
through the preparation of connection documents.
Besides for data file, a basic model interface (BMI) is also proposed to connect the two
independent models. The BMI is used to store and transmit calculation data for the two
software to be connected (Hoch et al. 2018). Each BMI has several functions that can be
called from external applications like, such as a Python script, and models connected to
a BMI can be updated at a user-specified time step. By design, the BMI uses only simple
data types and is easy to implement in all languages. Peckham et al. (2013) proposed the
BMI as a tool within the Community Surface Dynamics Modeling System (CSDMS) pro-
ject to exchange information between separate models at any given time step. Hoch et al.
(2018) developed a coupling of PCR-GLOBWB and Delft 3D FM using the BMI, which
showed that better results can be obtained through BMI coupling. The BMI is convenient
for developing a one-way coupling of hydrologic and hydrodynamic models.
External coupling models are powerful tools for watershed flood simulation, in particu-
lar on large spatial and temporal scales, due to their convenience in model construction.
However, the location of the boundary points limits the influence of the upland runoff on
downstream waters. The runoff generation on both sides of the river is transferred to lim-
ited points upstream of the mainstream or tributaries of the river network, resulting in an
error in the peak flow rate of the boundary points. Since the flow information is transferred
in one-way from hydrologic to hydrodynamic models, the external coupling cannot capture
the mutual interaction between runoff production and flood inundation. Mass conservation
of water through the coupling interface cannot be guaranteed.

2.3.3 Internal coupling model

Internal coupling is a two-way coupling, in which the governing equations of the hydro-
logic and hydrodynamic models are solved separately, with information at the shared
interface updated and exchanged at each or several computational time steps. The pro-
cess of an internal coupling technique between hydrologic and hydrodynamic models

13
888 Natural Hazards (2023) 118:875–902

Fig. 9  Classification of internal coupling model: a coupling of hydrologic and 1D hydrodynamic models; b
indirect coupling of hydrologic and 2D hydrodynamic models

Fig. 10  Flowchart of internal coupling technique of hydrologic and hydrodynamic models

can be described by the followings: (a) the discharge obtained from the hydrologic
model is transferred to the hydrodynamic model via interface condition; (b) the hydro-
dynamic model uses this discharge to calculate flood information, such as water depth
and flow velocity, and transfer this new information to hydrologic model; (c) hydrologic
model uses the received information to compute updated new results and provides it to
hydrodynamic model; (d) this process is repeated until it reaches the end of the simula-
tion period. The internal coupling model can be further classified into the coupling of
hydrologic and 1D hydrodynamic models, and the indirect coupling of hydrologic and
2D hydrodynamic models, as shown in Fig. 9.

(1) Type-1: coupling of hydrologic and 1D hydrodynamic models

Figure 10 presents the coupling technique of hydrologic and 1D hydrodynamic models,

where the hydrologic and 1D hydrodynamic models are calculated synchronously in time.
The flow discharge rate obtained from the hydrologic model is treated as a mass source of
the 1D hydrodynamic model, while the water depth calculated in the 1D hydrodynamic
model is fed back to the hydrologic model. The coupling of the Mike SHE and Mike11 is
a typical example of the coupling of hydrologic and 1D hydrodynamic models (Thompson
et al. 2004; Danish Hydraulic Institute 2009). Mike SHE and Mike 11 are connected by
the overflow weirs. When the water level in the river is lower than that on the outside, the
water flows into the river through the outside; in contrast, when the water level in the river
is higher than the water level outsides, the water flow from the river to the outside. SWMM
is also an internal coupling of distributed hydrologic model and the 1D hydrodynamic
model (Rossman 2015). Compared with external coupling, internal coupling has improved

13
Natural Hazards (2023) 118:875–902 889

accuracy. However, the application of 1D modeling of overland flow is limited when devel-
oping precise and reliable flood maps in 2D inundation regions.

(2) Type-2: indirect coupling of hydrologic and 2D hydrodynamic models

The coupling of hydrologic and 1D hydrodynamic models cannot accurately simulate the
flood inundation processes in 2D regions, such as lakes, reservoirs, complex flows, and
estuaries where 2D or 3D computations are required. To overcome the lack of 2D hydrody-
namic simulation in type-1, the coupling of hydrologic, 1D, and 2D hydrodynamic models
is proposed.
In this coupling type, the runoff flows into the 1D river or pipes first, and the hydrologic
model is coupled with the 1D hydrodynamic model. And then, the water in 1D rivers or
pipes can overflow into low-lying areas; conversely, the water in low-lying areas can flow
to 1D regions in return. The 1D and 2D hydrodynamic models are coupled in a two-way
manner. This coupling type is an indirect coupling of hydrologic and 2D hydrodynamic
models. For instance, Mike SHE and Mike11 are coupled to form Mike Urban, and Mike11
and Mike21 are dynamically coupled to form Mike Flood. The indirect coupling between
the hydrologic and 2D hydrodynamic models can be developed by coupling Mike Urban
and Mike Flood. The 1D hydrodynamic model is a connection channel between the hydro-
logic and the 2D hydrodynamic models. Some other similar models for indirect coupling
of the hydrologic and 2D hydrodynamic model can be found in Hsu et al. (2000), Seyoum
et al. (2012) , Chen and Huang (2017, 2018), Sindhu et al. (2021).
Compared with type-1, this coupling type has satisfactory and acceptable accuracy and
is widely used. As the 2D hydrodynamic model is only calculated in the local inundation
region, its computational efficiency is greatly improved in comparison with HM2D. How-
ever, in this coupling type, it is assumed that the water first discharges into the 1D rivers or
pipes, and then, flows through 1D rivers or pipes to the 2D regions, as shown in Fig. 11a.
The hydrologic model is not directly coupled with the 2D hydrodynamic model, which is
inconsistent with the actual flood processes. In reality, water may be discharged into both
1D channel and 2D waterbodies simultaneously, and the hydrologic, 1D, and 2D hydro-
dynamic models should be linked directly (Fig. 11b). Direct coupling of hydrologic and
2D hydrodynamic models can reflect the flood processes more truly, which deserves more
attention (Brewer et al. 2018).

2.4 Dynamic bidirectional coupling model

2.4.1 Model framework

Dynamic bidirectional coupling type is defined as a model that joins the hydrologic and
hydrodynamic models into a single modeling framework by combing their code, where
the governing equations of hydrologic and hydrodynamic models are reformulated and
synchronously solved in a single code base. This means that the information exchange
between both portions of the code is performed internally within the same source code
and does not involve the exchange of external input and output files. One example of
this technique is the dynamic bidirectional coupled hydrologic-hydrodynamic models
(DBCM) (Jiang et al. 2021; Shen et al. 2021). The coupling framework of DBCM is
shown in Fig. 12, where fully distributed hydrologic and 2D hydrodynamic models are

13
890 Natural Hazards (2023) 118:875–902

(a)

(b)

Fig. 11  Runoff processes in a indirect coupling of hydrologic and 2D hydrodynamic models and b actual
flood process

Fig. 12  Coupling framework of DBCM

13
Natural Hazards (2023) 118:875–902 891

bidirectionally and dynamically coupled. A fully distributed hydrologic model is used to

simulate runoff generation and routing, which consists of 2D water mass balance (Eqs.
(1–3)) and Manning equations (Eqs. (4–5)). The 2D hydrodynamic model is an efficient
and robust model for predicting flood processes in local inundation regions. The govern-
ing equations of the 2D hydrodynamic model are 2D shallow equations, consisting of
mass and momentum conservation equations (Toro 2001).

Vin+1 − Vin
= Qx in i − Qx out i + Qy in i − Qy out i + Ai qnr i (1)
( ) ( ) ( ) ( )
Δt

m
∑
(2)
( ) ( ) ( n )
Qx in i − Qx out i = − qx Γ ⋅ nx k ΔLk
k=1

m ( )
∑
qny Γ ⋅ ny ΔLk (3)
( ) ( )
Qy in i − Qy out i = −
k
k=1

1∕ 2
h5∕ 3 Sx
qx = (4)
n

1∕ 2
h5∕ 3 Sy
qy = (5)
n
3
(where) the( superscript n and n + 1 is the time step; V is the water volume
3
( of) grid( ­()m );
Qx in i , Qx out i is the inflow and outflow of grid i in x direction ­(m /s); Qy in i , Qy out i
)

is the inflow and outflow of grid i in y direction ­(m3/s); qri indicates runoff rate of grid i
(mm/h), which is rainfall intensity minus infiltration rate; Ai is the area of grid i ­(m2); qx , qy
are the unit discharge stored at cell-center along x and y direction ­(m2/s), with h, u and v
being water depth (m), flow velocity (m/s) in x and y directions, respectively; qxΓ , qyΓ are
the unit discharge at grid boundary in x and y direction, respectively ­(m2/s), which are cal-
culated based on qx , qy; ΔLk is the side length of grid (m); k = 1, 2, 3,…, m is the number
of edges of cell; n is the Manning roughness coefficient; Sx and Sy are water level gradients
along x and y direction, respectively.
In DBCM, the computational domain is divided into non-inundation and inunda-
tion regions, and the area of non-inundation regions is much larger than that of inunda-
tion regions. The hydrologic model is applied to non-inundation areas with small water
depth, whereas the 2D hydrodynamic model is applied to the local inundation areas with
high water depth. When the rain intensity increased, the inundation regions expanded
because of the gradual accumulation of surface water volume. The inflow discharge
positions, flow path, and discharge values subsequently changed, as shown in Fig. 13.
Therefore, a coupling moving interface (CMI) is formed between the inundation and
non-inundation regions, and the hydrologic and 2D hydrodynamic models are coupled
via this CMI.
The solution of the convection flux through the CMI can be considered a local Rie-
mann problem, and special treatment on discharge through the CMI needs to be taken
based on the local flow state using the characteristic theory. A pair of the characteristic
wave at the CMI is calculated:

13
892 Natural Hazards (2023) 118:875–902

Fig. 13  Diagram of CMI moving with accumulated rainfall

√
SL = ui − ghRi
√ (6)
SR = ui+1 + ghLi+1

where SL , SR are the characteristic wave at the CMI, ui , ui+1 are the flow velocity at the grids
i and i + 1; hRi , hLi+1 are the effective water depth at the left and right sides of the CMI.
From Fig. 14, if the characteristic speeds are all positive, the fluxes depend entirely on
the hydrologic model, and the CMI moved from non-inundation to inundation areas. If
the characteristic speeds are all negative, the fluxes depend entirely on the hydrodynamic
model, and the CMI moved from inundation to non-inundation areas. If the characteristic
speeds have a negative and positive value, both the current flow state in the hydrologic and
hydrodynamic models must be taken into consideration, and the CMI is not changed.
In DBCM, the results of hydrologic model affect the 2D hydrodynamic computation,
and the results of hydrodynamic model also affect hydrologic computation, which can take
into account effects due to overflowing in the floodplain, backwater effects at the conflu-
ences. Compared with existing coupling models, the DBCM can fully consider the flow
states with both mass and momentum transfer. It had better simulation accuracy than exter-
nal coupling models. Compared with HM2D, it can save computation time and had better
numerical stability, especially in thin-layer water regions (Kim et al. 2012).

13
Natural Hazards (2023) 118:875–902 893

Fig. 14  The flux calculation depends on the a hydrologic model; b hydrodynamic model and c hydrologic
and hydrodynamic models

However, because uniformly structured grids are adopted in DBCM, high-resolu-

tion simulations in a large domain inevitably involve numerous computational nodes
and substantially increased the computational cost. A multigrid DBCM (M-DBCM)
was proposed by Shen et al. (2021) to further improve the computational efficiency
of DBCM. The flood-prone areas were meshed with fine grids, whereas the rest of
the domain remained coarse grids. The hydrologic model was applied to coarse grids,
whereas the hydrologic model was coupled with a hydrodynamic model to detail the
flood on fine grids. Different time steps were adopted for the coarse and fine grids. The
M-DBCM can save computation time while maintaining high numerical accuracy.

13
894 Natural Hazards (2023) 118:875–902

2.4.2 Advantages of dynamic bidirectional coupling model

(1) Reasonable coupling mechanism

In DBCM and M-DBCM, a fully distributed hydrologic model is adopted to couple with
the 2D hydrodynamic model. In a fully distributed hydrologic model, a 2D routing scheme
including 2D water balance and Manning equations is used to calculate the surface flow.
Compared with a distributed hydrologic model where the runoff routing is considered
as 1D way, the 2D routing scheme adopted in the fully distributed hydrologic model is
more consistent with the actual runoff routing, which can achieve satisfactory simulation
accuracy.
It is difficult to directly couple semi-distributed or distributed hydrologic models with
a 2D hydrodynamic model because of the 1D runoff routing scheme adopted in these
models, such as SWMM, and SWAT. Generally, the semi-distributed or distributed hydro-
logic model is coupled with a 1D hydrodynamic model first and then, coupled with the
2D hydrodynamic model. Therefore, the indirect coupling of hydrologic and 2D hydro-
dynamic models is developed, which is not consistent with the actual runoff path. What is
more, although the 2D surface flow is simulated based on girds in certain fully distributed
hydrological models, such as Mike SHE, and CASC2D, it also fails to develop the direct
coupling of hydrologic and 2D hydrodynamic models.
Based on the fully distributed hydrologic model, a direct coupling of hydrologic and
2D hydrodynamic models is developed in DBCM. The hydrologic and 2D hydrodynamic
model is dynamically and bidirectionally coupled by means of CMI. The computational
regions of hydrologic and 2D hydrodynamic models change over time. In the dynamic
bidirectional coupling approach, overland flow can be discharged from the non-inundation
to 2D inundation regions, and the effects owing to the overflow of the floodplain and back-
water effects at the river confluences are also considered. The position of coupling inter-
face does not keep fixed in advance during the calculation process. The location where
the runoff enters the inundation region varies dynamically, and the flood level can also
submerge the original inflow points and regenerate a new coupling interface. Such an alter-
nation mechanism is close to natural flow processes. And the mass and momentum conser-
vation through the CMI can be ensured.

(2) High efficiency and simplicity

Researchers and professionals have been devoted to developing various coupling models
of hydrologic and hydrodynamic models for watershed flood simulation, but few studies
reported such a dynamic bidirectional coupling model due to the complication of refor-
mulating and simultaneously solving governing equations in a single code base (Liu et al.
2019). For such a coupling model, researchers may worry about the complexity and com-
putational efficiency.
In DBCM, the hydrologic and 2D hydrodynamic models are coded in one software
operated synchronously in time. The CMI is proposed to couple the hydrologic and 2D
hydrodynamic models. If the range of reservoirs, lakes, and other water-bearing areas
changes little, the coupling interface of hydrologic and 2D hydrodynamic models can also

13
Natural Hazards (2023) 118:875–902 895

be fixed. Therefore, a coupling fixed interface (CFI) is developed to couple hydrologic

and 2D hydrodynamic models, and the calculation of the DBCM can be further simpli-
fied when the CMI is changed to CFI. The mass and momentum conservation can also be
ensured through the CFI.
In DBCM, the 2D hydrodynamic model is only applied to relatively small local regions,
and a time-explicit scheme is used to solve these two models, which can reduce the com-
putational time and complexity. Besides, a multi-grid technology can be used to discretize
the computational domain to further improve computational efficiency (Shen et al. 2021).
The computational domain is divided into coarse and fine grids. Different time steps are
adopted in coarse and fine grids. Therefore, the computational efficiency of DBCM can be
greatly improved compared with the HM2D.

(3) Complete functions

Two separately running models are coupled in external and internal coupling models. In
external coupling models, the computations obtained from the hydrologic model are used
as input in the hydrodynamic model. Since the two separate models are coupled, the data
transfer between these models needs manual processing. This coupling compromises its
timely data delivery for emergency responders. In internal coupling models, separately
running models are coupled iteratively using information exchange. In DBCM, the gov-
erning equations of hydrologic and hydrodynamic models are reformulated and simultane-
ously solved in a single code base, which can save computation time and resources.
The DBCM is fully functional and has advantages over other coupling types. It includes
the functions of external and internal coupling models; conversely, the external and inter-
nal coupling types do not have the functions of DBCM, because the CMI is unique to the
DBCM. The DBCM can be converted to other models by changing input parameters. For
example, if DBCM does not take the impacts of the hydrodynamic models on hydrologic
model into account, it can be converted to an external coupling model. Conversely, the
external coupling models do not have the functions of DBCM, because the two separate
models are used and the fixed boundary nodes are adopted in external coupling models.
If the surface flow over the entire watershed is calculated by hydrologic model, the
DBCM can be converted to hydrologic model. If the 2D hydrodynamic model is used to
simulate surface flow over the entire watershed, the DBCM can be converted to HM2D.
But the HM2D cannot be converted to DBCM, due to the no calculation of hydrological
runoff routing in the hydrodynamic model.

2.4.3 Current challenges of DBCM

(1) Improvement of hydrologic model

In the hydrologic model applied to DBCM, only the overland flow processes, i.e., runoff
generation and routing, are calculated. In a flash flood, overland flow is the most impor-
tant factor to be considered, and evaporation, vegetation interception, interflow, and under-
ground runoff can be neglected due to the short rainfall and calculation duration. How-
ever, a complete hydrologic model should include surface flow, interflow, and underground

13
896 Natural Hazards (2023) 118:875–902

runoff. Regarding the surface hydrologic model, the ignored interflow and underground
runoff processes likely lead to underestimations in flood simulation. In future works, the
interflow and underground runoff could be calculated in the hydrologic model.

(2) Adding a 1D hydrodynamic model to DBCM

In DBCM, the hydrologic and 2D hydrodynamic models are used to simulate the flooding
process in non-inundation and inundation regions, respectively. However, in a watershed,
rivers are usually narrow and the width of the river is often tens of meters or even less. To
reflect the flow process in rivers, it needs to be divided into really small grids. It is well-
known that small grids mean a high computational burden and resources. Many researchers
have devoted themselves to solving this problem and proposed indirect coupling of hydro-
logic and 2D hydrodynamic models. But in existing coupling types, the hydrologic and
2D hydrodynamic models are not coupled directly, which is not consistent with the flood
evolution process.
In reality, the runoff routing can directly enter the 1D river channel and 2D regions,
and the hydrologic model should be directly coupled with 1D and 2D hydrodynamic mod-
els. In future works, the 1D hydrodynamic would be added to the DBCM. Therefore, the
direct dynamic bidirectional coupling of hydrologic and 1D, 2D hydrodynamic models can
be developed in the DBCM, as shown in Fig. 15. Besides, in urban flood simulation, the
1D hydrodynamic model can be used to simulate the runoff routing in pipe network, and
the dynamical bidirectional coupling of hydrologic, 1D pipe, 1D river and 2D hydrody-
namic models can also be developed. Therefore, the application of DBCM can be further
extended, and it can be used for simulating both urban and watershed flood processes.

3 Conclusions and future research

Many numerical models have been developed to simulate floods caused by extreme rain-
fall events in the past few decades. The hydrologic models have been widely used in
simulating runoff generation and routing, which are divided into lumped, semi-distrib-
uted, distributed, and fully distributed hydrologic models based on spatial representa-
tion and routing schemes. The hydrologic models require less computational time at the

Fig. 15  Improved DBCM by

adding 1D hydrodynamic model

13
Natural Hazards (2023) 118:875–902 897

price of representing less detailed physical processes, while the hydrodynamic models
in many cases are capable of detailing the flood inundation process with computational
limitations in large-scale applications.
Substantial research has been conducted to improve the computational efficiency and
accuracy of flood simulation models by coupling hydrologic and hydrodynamic mod-
eling (Wang et al. 2021). The external coupling model may be useful as an important
attempt to simulate the flooding process, as it requires less effort and computational
resources than the internal coupling model and HM2D. However, it neglects the interac-
tions between hydrologic and hydrodynamic models. Also, the hydrologic and hydrody-
namic models are coupled by fixed limited coupling nodes, which is not consistent with
the flooding process. Despite the use of an internal coupling model (i.e., coupling of
hydrologic and 1D hydrodynamic model, indirect coupling of hydrologic and 2D hydro-
dynamic model), some limitations are still encountered, such as limiting the information
exchange at a fixed boundary, forcing water flows into the 1D river and 2D inunda-
tion regions successively. Although the HM2D can solve these problems and is a state-
of-the-art approach for simulating flood processes, it has computational limitations,
especially in large-scale applications. To address the current limitations of the existing
models, a dynamic bidirectional coupling model (DBCM) was proposed, which has the
following advantages:

(1) A fully distributed hydrologic model is adopted, which is coupled with a 2D hydrody-
namic model with ease.
(2) The hydrologic and hydrodynamic models are coupled by means of CMI, and the inter-
actions between these two models are also considered. Such a coupling mechanism is
close to the natural flood evolution processes. Therefore, the DBCM has acceptable
simulation accuracy compared with the external and internal coupling models.
(3) As the 2D hydrodynamic model is used to simulate the flooding process in local inun-
dation regions, the DBCM can save more computation time and resources compared
with HM2D.
(4) The DBCM is fully functional, which includes the functions of external and internal
coupling models; conversely, the external and internal coupling types do not have the
functions of DBCM, because the CMI is unique to the DBCM.

The DBCM has potential development if further improvement is made. This model
can be further improved by adding a 1D hydrodynamic model to it. The flow in a narrow
river can be simulated using a 1D model, which may save more time than using a 2D
model. Besides, adding interflow and underground flow modules can further improve
the hydrologic model.
Acknowledgements This study was supported by the National Science Foundation of China (Grant No.
52179068) and the Key Laboratory of Hydroscience and Engineering (Grant No. 2021-KY-04). The authors
thank the anonymous reviewers for their valuable comments.

Funding This study was supported by the National Science Foundation of China (Grant No. 52179068) and
the Key Laboratory of Hydroscience and Engineering (Grant No. 2021-KY-04).

Declarations
Competing interests All authors declare that no conflict of interest exists.

13
898 Natural Hazards (2023) 118:875–902

References
Arnold JG, Srinivasan R, Muttiah RS, Williams JR (1998) Large area hydrologic modeling and assess-
ment part I: model development. J Am Water Resour As 34:73–89. https://​doi.​org/​10.​1111/j.​1752-​
1688.​1998.​tb059​61.x
Ambrose RB, Wool TA (2017) WASP8 stream transport model theory and user’s guide. U.S. Environ-
mental Protection Agency, Office of Research and Development. https://​www.​epa.​gov/​sites/​produ​
ction/​files/​201805/​docum​ents/​strea​mtran​sport-​user-​guide.​pdf
Audusse E, Bristeau MO (2007) Finite-volume solvers for a multilayer Saint-Venant system. Int J Appl
Math Comput Sci 17(3):311–320. https://​doi.​org/​10.​2478/​v10006-​007-​0025-0
Ada MPA, Suidan MT, Shuster WD (2010) Modeling techniques of best management practices: rain bar-
rels and rain gardens using EPA SWMM-5. J Hydrol Eng 15(6):434–443. https://​doi.​org/​10.​1061/​
(ASCE)​HE.​1943-​5584.​00001​36
Arnold JG, Allen PM, Volk M, Williams JR, Bosch DD (2010) Assessment of different representations
of spatial variability on SWAT model performance. Trans ASABE 53(5):1433–1443. https://​doi.​
org/​10.​13031/​2013.​34913
Beasley DB, Huggins LF, Monke EJ (1980) ANSWERS: a model for watershed planning. Trans ASAE
23(4):938–944. https://​doi.​org/​10.​13031/​2013.​34692
Becknell BR, Imhoff JC, Kittle JL, Donigian AS, Johanson RC (1993) Hydrological simulation pro-
gram—Fortran user’s manual for release 12. Us EPA
Beven K, Freer J (2001) A dynamic TOPMODEL. Hydrol Process 15(10):1993–2011. https://​doi.​org/​10.​
1002/​hyp.​252
Bingner RL, Theurer FD (2002) AnnAGNPS Technical processes: documentation Version 2. Available
at www.​sedlab.​olemi​ss.​edu/​AGNPS.​html.
Betrie GD, van Griensven A, Mohamed YA, Popescu I, Mynett AE, Humme S (2011) Linking SWAT
and Sobek using open modeling interface (OPENMI) for sediment transport simulation in the Blue
Nile River Basin. Trans Asabe 54(5):1749–1757
Beven KJ (2012) Rainfall-runoff modelling: the primer, 2nd edn. Wiley-Blackwell, Hoboken
Bottcher AB, Whiteley BJ, James AI, Hiscock JG (2012) Watershed assessment model (WAM): model
use, calibration, and validation. Trans ASABE 55(4):1367–1383. https://​doi.​org/​10.​13031/​2013.​
42248
Bravo JM, Allasia D, Paz AR, Collischonn W, Tucci CEM (2012) Coupled hydrologic-hydraulic modeling
of the Upper Paraguay River Basin. J Hydrol Eng 17:635–646. https://​doi.​org/​10.​1061/​(ASCE)​HE.​
1943-​5584.​00004​94
Bhola PK, Leandro J, Disse M (2018) Framework for offline flood inundation forecasts for two-dimensional
hydrodynamic models. Geosciences (switzerland) 8(9):346. https://​doi.​org/​10.​3390/​geosc​ience​s8090​
346
Brewer SK, Worthington TA, Mollenhauer R, Stewart DR, Mcmanamay RA, Guertault L, Moore D (2018)
Synthesizing models useful for ecohydrology and ecohydraulic approaches: an emphasis on integrat-
ing models to address complex research questions. Ecohydrology 11(7):e1996. https://​doi.​org/​10.​1002/​
eco.​1966
Bulti DT, Abebe BG (2020) A review of flood modeling methods for urban pluvial flood application. Model
Earth Syst Environ 6(3):1293–1302. https://​doi.​org/​10.​1007/​s40808-​020-​00803-z
Brendel CE, Dymond RL, Aguilar MF (2021) Modeling storm sewer networks and urban flooding in Roa-
noke, Virginia, with SWMM and GSSHA. J Hydrol Eng 26(1):05020044. https://​doi.​org/​10.​1061/​
(ASCE)​HE.​1943-​5584.​00020​21
Bates PD (2022) Flood inundation prediction. Annu Rev Fluid Mech 54:287–315. https://​doi.​org/​10.​1146/​
annur​ev-​fluid-​030121-​113138
Chávarri E, Crave A, Bonnet MP, Mejia A, Da Silva JS, Guyot JL (2013) Hydrodynamic modelling of the
amazon river: factors of uncertainty. J S Am Earth Sci 44:94–103. https://​doi.​org/​10.​1016/j.​jsames.​
2012.​10.​010
Chen W, Huang G, Han Z (2017) Urban stormwater inundation simulation based on SWMM and diffusive
overland-flow model. Water Sci Technol J Int As Water Pollut Res 76(12):3392. https://​doi.​org/​10.​
2166/​wst.​2017.​504
Chen W, Huang G, Han Z, Wang W (2018) Urban inundation response to rainstorm patterns with a coupled
hydrodynamic model: a case study in Haidian Island, China. J Hydrol 564:1022–1035. https://​doi.​org/​
10.​1016/j.​jhydr​ol.​2018.​07.​069
Cea L, Lopez-Nunez A (2021) Extension of the two-component pressure approach for modeling mixed free-
surface-pressurized flows with the two-dimensional shallow water equations. Int J Numer Meth Fluids
93(3):628–652. https://​doi.​org/​10.​1002/​fld.​4902

13
Natural Hazards (2023) 118:875–902 899

Costabile P, Costanzo C (2021) A 2D-SWEs framework for efficient catchment-scale simulations: hydro-
dynamic scaling properties of river networks and implications for non-uniform grids generation. J
Hydrol 599(6402):126306. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2021.​126306
Downer CW, Ogden FL (2004) GSSHA: model to simulate diverse stream flow producing processes. J
Hydrol Eng 9(3):161–174. https://​doi.​org/​10.​1061/​(ASCE)​1084-​0699(2004)9:​3(161)
Debele B, Srinivasan R, Parlange JY (2006) Coupling upland watershed and downstream waterbody
hydrodynamic and water quality models (SWAT and CE-QUAL-W2) for better water resources
management in complex river basins. Environ Model Assess 13(1):135–153. https://​doi.​org/​10.​
1007/​s10666-​006-​9075-1
Danish Hydraulic Institute (2009) The MIKE SHE user and technical reference manual. Danish Hydrau-
lic Institute, Copenhagen, Denmark
Dargahi B, Setegn SG (2011) Combined 3D hydrodynamic and watershed modeling of Lake Tana, Ethi-
opia. J Hydrol 398:44–64. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2010.​12.​009
Evans E, Ashley R, Hall J, Penning-Rowsell E, Sayers P, Thome C, Watkinson A (2004) Foresight:
future flooding volume II: managing future risks. Department of Trade and Industry, London
Feldman AD (2000) Hydrologic modeling system HEC-HMS, technical reference manual. Davis, CA,
USA: U.S. Army Corps of Engineers, Hydrologic Engineering Center, HEC
Feistl T, Bebi P, Dreier L, Hanewinkel M, Bartelt P (2014) A coupling of hydrologic and hydraulic
models appropriate for the fast floods of the Gardon river basin (France). Nat Hazard 14(11):2899–
2920. https://​doi.​org/​10.​5194/​nhess-​14-​2899-​2014
Felder G, Zischg A, Weingartner R (2017) The effect of coupling hydrologic and hydrodynamic models
on probable maximum flood estimation. J Hydrol 550:157–165. https://​doi.​org/​10.​1016/j.​jhydr​ol.​
2017.​04.​052
Garcia-Navarro P, Murillo J, Fernandez-Pato J, Echeverribar I, Morales-Hernandez M (2019) The shal-
low water equations and their application to realistic cases. Environ Fluid Mech 19(5):1235–1252.
https://​doi.​org/​10.​1007/​s10652-​018-​09657-7
Gomes MMD, Verçosa LFD, Cirilo JA (2021) Hydrologic models coupled with 2D hydrodynamic
model for high-resolution urban flood simulation. Nat Hazards 108:3121–3157. https://​doi.​org/​10.​
1007/​s11069-​021-​04817-3
Gwapedza D, Nyamela N, Hughes DA, Slaughter AR, Mantel SK, Waal B (2021) Prediction of sediment
yield of the Inxu river catchment (South Africa) using the MUSLE. Int Soil Water Conserv Res
9(1):37–48. https://​doi.​org/​10.​1016/j.​iswcr.​2020.​10.​003
Hamrick JM (1992) A three-dimensional environmental fluid dynamics computer code: theoretical and
computational aspects. In: Special report 317, The College of William and Mary. Virginia Institute
of Marine Science, Gloucester Point, VA
Hsu MH, Chen SH, Chang TJ (2000) Inundation simulation for urban drainage basin with storm sewer
system. J Hydrol 234(1–2):21–37. https://​doi.​org/​10.​1016/​S0022-​1694(00)​00237-7
Hunter NM, Bates PD, Horritt MS, Wilson MD (2007) Simple spatially-distributed models for predict-
ing flood inundation: a review. Geomorphology 90(3–4):208–225. https://​doi.​org/​10.​1016/j.​geomo​
rph.​2006.​10.​021
Henonin J, Russo B, Mark O, Gourbesville P (2013) Real-time urban flood forecasting and modelling—a
state of the art. J Hydroinf 15(3):717–736. https://​doi.​org/​10.​2166/​hydro.​2013.​132
Hou J, Liu F, Tong Y, Guo K, Li D (2020) Numerical simulation for runoff regulation in rain garden
using 2D hydrodynamic model. Ecol Eng 153(2):105794. https://​doi.​org/​10.​1016/j.​ecole​ng.​2020.​
105794
Hoch JM, Haag AV, Dam AV, Winsemius HC, Beek LPHV, Bierkens MFP (2018) Assessing the impact
of hydrodynamics on large-scale flood wave propagation—a case study for the Amazon Basin.
Hydrol Earth Syst Sci 21(1):117–132. https://​doi.​org/​10.​5194/​hess-​21-​117-​2017
Hdeib R, Abdallah C, Colin F, Brocca L, Moussa R (2018) Constraining coupled hydrological-hydraulic
flood model by past storm events and post-event measurements in data-sparse regions. J Hydrol
540(565):160–176. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2018.​08.​008
Jiang C, Zhou Q, Yu W, Yang C, Lin B (2021) A dynamic bidirectional coupled surface flow model for
flood inundation simulation. Nat Hazard 21(2):497–515. https://​doi.​org/​10.​5194/​nhes-​21-​497-​2021
Kniselw G (1980) CREAMS, a field scale model for chemicals, runoff, and erosion from agricultural
management systems. Conservation Research Report USDA, Washington D C. vol 1980, p 44
Kuiry SN, Sen D, Bates PD (2010) Coupled 1D–quasi-2D flood inundation model with unstructured
grids. J Hydraul Eng 136(8):493–506. https://​doi.​org/​10.​1061/​(ASCE)​HY.​1943-​7900.​00002​11
Kim J, Warnock A, Ivanov VY, Katopodes ND (2012) Coupled modeling of hydrologic and hydrody-
namic processes including overland and channel flow. Adv Water Resour 37:104–126. https://​doi.​
org/​10.​1016/j.​advwa​tres.​2011.​11.​009

13
900 Natural Hazards (2023) 118:875–902

Lindstrom G, Pers C, Rosberg J, Stromqvist J, Arheimer B (2010) Development and testing of the HYPE
(Hydrological Predictions for the Environment) water quality model for different spatial scales.
Hydrol Res 41(3–4):295–319. https://​doi.​org/​10.​2166/​nh.​2010.​007
Liang Q (2010) Flood simulation using a well-balanced shallow flow model. J Hydraul Eng 136(9):669–
675. https://​doi.​org/​10.​1061/​(ASCE)​HY.​1943-​7900.​00002​19
Li YL, Zhang Q, Yao J, Werner AD, Li XH (2014) Hydrodynamic and hydrological modeling of the
Poyang Lake catchment system in China. J Hydrol Eng 19(3):607–616. https://​doi.​org/​10.​1061/​
(ASCE)​HE.​1943-​5584.​00008​35
Liang Q, Smith LS (2015) A high-performance integrated hydrodynamic modelling system for urban
flood simulations. J Hydroinf 17(4):518–533. https://​doi.​org/​10.​2166/​hydro.​2015.​029
Liu Q, Qin Y, Zhang Y, Li Z (2015) A coupled 1D–2D hydrodynamic model for flood simulation in
flood detention basin. Nat Hazards 75(2):1303–1325. https://​doi.​org/​10.​1007/​s11069-​014-​1373-3
Liu Z, Zhang H, Liang Q (2019) A coupled hydrological and hydrodynamic model for flood simulation.
Hydrol Res 50(2):580–606. https://​doi.​org/​10.​2166/​nh.​2018.​090
Li WJ, Lin KR, Zhao TTG, Lan T, Chen XH, Du HW, Chen HY (2019) Risk assessment and sensitivity
analysis of flash floods in ungauged basins using coupled hydrologic and hydrodynamic models. J
Hydrol 572:108–120. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2019.​03.​002
Li Z, Chen MY, Gao S, Luo XY, Gourley JJ, Kirstetter P, Yang TT, Kolar R, McGovern A, Wen YX,
Rao B, Yami T, Hong Y (2021) CREST-IMAP v10: a fully coupled hydrologic-hydraulic mod-
eling framework dedicated to flood inundation mapping and prediction. Environ Model Softw
141(1):105051. https://​doi.​org/​10.​1016/j.​envso​ft.​2021.​105051
Marsik M, Waylen P (2006) An application of the distributed hydrologic model CASC2D to a tropical
montane watershed. J Hydrol 330(3–4):481–495. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2006.​04.​003
Munar AM, Cavalcanti JR, Bravo JM, Fan FM, da Motta-Marques D, Fragos CR (2018) Coupling large-
scale hydrological and hydrodynamic modeling: toward a better comprehension of watershed-shal-
low lake processes. J Hydrol 564:424–441. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2018.​07.​045
Nguyen P, Thorstensen A, Sorooshian S, Hsu K, AghaKouchak A, Sanders B, Koren V, Cui Z, Smith M
(2016) A high resolution coupled hydrologic hydraulic. J Hydrol 541:401–420. https://​doi.​org/​10.​
1016/j.​jhydr​ol.​2015.​10.​047
Parajuli PB, Nelson NO, Frees LD, Mankin KR (2009) Comparison of AnnAGNPS and SWAT model
simulation results in USDA-CEAP agricultural watersheds in south-central Kansas. Hydrol Process
23:748–763. https://​doi.​org/​10.​1002/​hyp.​7174
Peckham SD, Hutton E, Norris B (2013) A component-based approach to integrated modeling in the
geosciences: the design of CSDMS. Comput Geosci 53:3–12. https://​doi.​org/​10.​1016/j.​cageo.​2012.​
04.​002
Pilotti M, Milanesi L, Bacchi V, Tomirotti M, Maranzoni A (2020) Dam-break wave propagation in
alpine valley with HEC-RAS 2D: experimental cancano test case. J Hydraul Eng 146(6):05020003.
https://​doi.​org/​10.​1061/​(ASCE)​HY.​1943-​7900.​00017​79
Rankinen K, Lepisto A, Granlund K (2002) (2002) Hydrological application of the INCA model with
varying spatial resolution and nitrogen dynamics in a northern river basin. Hydrol Earth Syst Sci
6(3):339–350. https://​doi.​org/​10.​5194/​hess-6-​339-​2002
Rinsema JG (2014) Comparison of rainfall runoff models for the Florentine Catchment. University of
Tasmania. Retrieved from http://​essay.​utwen​te.​nl/​66526/1/​Rinse​ma_​Jan_​Gert.​pdf
Rossman LA (2015) Storm water management model user’s manual version 5.1; EPA/600/R-14/413b;
U.S. Environmental Protection Agency: Cincinnati, OH, USA
SCS (1972) Hydrology. Section 4 in National Engineering Handbook. USDA Soil Conservation Service,
Washington
Singh VP (1995) Computer models of watershed hydrology highlands ranch. Water Resources Publica-
tions, CO
Spaeth KE, Pierson FB, Weltz MA, Blackburn WH (2003) Evaluation of USLE and RUSLE estimated
soil loss on rangeland. J Range Manag 56(3):234–246. https://​doi.​org/​10.​2307/​40038​12
Singh J, Altinakar MS, Yan D (2011) Two-dimensional numerical modeling of dam-break flows over
natural terrain using a central explicit scheme. Adv Water Resour 34(10):1366–1375. https://​doi.​
org/​10.​1016/j.​advwa​tres.​2011.​07.​007
Seyoum SD, Vojinovic Z, Price RK, Weesakul S (2012) Coupled 1D and noninertia 2D flood inunda-
tion model for simulation of urban flooding. J Hydraul Eng ASCE 138(1):23–34. https://​doi.​org/​10.​
1061/​(ASCE)​HY.​1943-​7900.​00004​85
Sitterson J, Knightes C, Parmar R, Wolfe K, Muche M, Avant B (2017) An Overview of Rainfall-Runoff
Model Types. EPA EPA/600/R-14/152 www.​epa.​gov/​resea​rch

13
Natural Hazards (2023) 118:875–902 901

Shin S, Her Y, Song JH, Kang MS (2019) Integrated sediment transport process modeling by cou-
pling soil and water assessment tool and environmental fluid dynamics code. Environ Model Softw
116(JUN):26–39. https://​doi.​org/​10.​1016/j.​envso​ft.​2019.​02.​002
Shen Y, Jiang C, Zhou Q, Zhu D, Zhang D (2021) A multigrid dynamic bidirectional coupled surface flow
routing model for flood simulation. Water 13:3454. https://​doi.​org/​10.​3390/​w1323​3454
Sindhu K, Singh A, Rao KHVD, Rao VV, Mohammood V (2021) 1D and 2D model coupling approach for
the development of operational spatial flood early warning system. Geocarto Int 37(15):4390–4405.
https://​doi.​org/​10.​1080/​10106​049.​2021.​18863​35
Shabani A, Woznicki SA, Mehaffey M, Butcher J, Wool Tim A, Whung PY (2021) A coupled hydrody-
namic (HEC-RAS) and water quality model (WASP) for simulating flood-induced soil, sediment, and
contaminant transport. J Flood Risk Manag. https://​doi.​org/​10.​1111/​jfr3.​12747
Toro EF (2001) Shock-capturing methods for free-surface shallow flows. John Wiley
Thompson JR, Sørenson HR, Gavin H, Refsgaard A (2004) Application of the coupled MIKE SHE/MIKE
11 modelling system to a lowland wet grassland in southeast England. J Hydrol 590(293):151–179.
https://​doi.​org/​10.​1016/j.​jhydr​ol.​2004.​01.​017
Timbadiya PV, Patel PL, Porey PD (2014) A 1D–2D coupled hydrodynamic model for river flood prediction
in a coastal urban floodplain. J Hydrol Eng. https://​doi.​org/​10.​1061/​(ASCE)​HE.​1943-​5584.​00010​29
Tansar H, Babur M, Karnchanapaiboon SL (2020) Flood inundation modeling and hazard assessment
in lower ping river basin using mike flood. Arab J Geosci 13(18):934. https://​doi.​org/​10.​1007/​
s12517-​020-​05891-w
UN Office for Disaster Risk Reduction (2020) The human cost of disasters: an overview of the last 20 years
(2000–2019).
Vacondio R, Aureli F, Ferrari A, Mignosa P, Dal Palu A (2016) Simulation of the january 2014 flood on the
secchia river using a fast and high-resolution 2d parallel shallow-water numerical scheme. Nat Hazards
80(1):103–125. https://​doi.​org/​10.​1007/​s11069-​015-​1959-4
Vacondio R, Dal Palu A, Ferrari A, Mignosa P, Aureli F, Dazzi S (2017) A non-uniform efficient grid type
for GPU-parallel Shallow Water Equations models. Environ Model Softw 88:119–137. https://​doi.​org/​
10.​1016/j.​envso​ft.​2016.​11.​012
Wang JP, Liang Q (2011) Testing a new adaptive grid-based shallow flow model for different types of flood
simulations. J Flood Risk Manag 4(2):96–103. https://​doi.​org/​10.​1111/j.​1753-​318X.​2011.​01094.x
Wu B, Wang G, Wang Z, Liu C, Ma J (2017) Integrated hydrologic and hydrodynamic modeling to assess
water exchange in a data-scarce reservoir. J Hydrol 555:15–30. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2017.​
09.​057
Wang J, Yun X, Pokhrel Y, Yamazaki D, Zhao Q, Chen A, Tang Q (2021) Modeling daily floods in the
Lancang-Mekong River Basin using an improved hydrological hydrodynamic model. Water Resour
Res 57:e2021WR029734. https://​doi.​org/​10.​1029/​2021W​R0297​341
Xu Z, Godrej AN, Grizzard TJ (2007) The hydrological calibration and validation of a complexly linked
watershed-reservoir model for the Occoquan watershed. Va J Hydrol 345(3–4):167–183. https://​doi.​
org/​10.​1016/j.​jhydr​ol.​2007.​07.​015
Yamazaki D, Sato T, Kanae S, Hirabayashi Y, Bates PD (2014) Regional flood dynamics in a bifurcating
mega delta simulated in a global river model. Geophys Res Lett 41(9):3127–3135. https://​doi.​org/​10.​
1002/​2014g​l0597​44
Yu C, Duan J (2014) Two-dimensional hydrodynamic model for surface-flow routing. J Hydraul Eng
140(9):04014045. https://​doi.​org/​10.​1061/​(ASCE)​HY.​1943-​7900.​00009​13
Yu C, Duan JG (2017) Simulation of surface runoff using hydrodynamic model. J Hydrol Eng (ASCE)
22(6):04017006. https://​doi.​org/​10.​1061/​(ASCE)​HE.​1943-​5584.​00014​97
Zhang X, Long W, Xie H, Zhu J, Wang J (2007) Numerical simulation of flood inundation processes by
2D shallow water equations. Front Arch Civ Eng China 1(1):107–113. https://​doi.​org/​10.​1007/​
s11709-​007-​0011-5
Zhao WJ, Sun W, Li ZL, Fan YW, Song JS, Wang LR (2013) A Review on SWAT model for stream flow
simulation. Adv Mater Res 726–731:3792–3798. https://​doi.​org/​10.​4028/​www.​scien​tific.​net/​amr
Zhang LH, Jin X, He CS, Zhang BQ, Zhang XF, Li JL, Zhao C, Tian J, DeMarchi C (2016) Comparison of
SWAT and DLBRM for hydrological modeling of a mountainous watershed in Arid Northwest China.
J Hydrol Eng 21(5):04016007. https://​doi.​org/​10.​1061/​(ASCE)​HE.​1943-​5584.​00013​13
Zhang L, Lu JZ, Chen XL, Liang D, Fu XK, Sauvage S, Perez JMS (2017) Stream flow simulation and
verification in ungauged zones by coupling hydrological and hydrodynamic models: a case study of
the Poyang Lake ungauged zone. Hydrol Earth Syst Sci 21(11):5847–5861. https://​doi.​org/​10.​5194/​
hess-​21-​5847-​2017
Zhang C, Wang L, Zhu H, Tang H (2020) Integrated hydrodynamic model for simulation of river-lake-sluice
interactions. Appl Math Model 83:90–106. https://​doi.​org/​10.​1016/j.​apm.​2020.​02.​019

13
902 Natural Hazards (2023) 118:875–902

Zhang H, Wu W, Hu C, Hu C, Liu S (2021) A distributed hydrodynamic model for urban storm flood risk
assessment. J Hydrol 600:126513. https://​doi.​org/​10.​1016/j.​jhydr​ol.​2021.​126513

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under
a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted
manuscript version of this article is solely governed by the terms of such publishing agreement and applicable
law.

13