An urban flooding treatment model based on GIS techniques

An urban flooding treatment model based on GIS techniques

Yangbo Chen
Department of Water Resources and Environment, Sun Yat-Sen University, Guangzhou, 510275, China, [email protected]

INTRODUCTION

Urban flooding is a serious natural disaster in the world which

not only causes serious damage, disturbs normal life and
working conditions, but also pollutes the city and causes
sanitary problems.
Urban flooding has long been one of the top issues among
hydrological community, and many models have been
proposed, such as SWMM, Hydroworks and Mouse. In recent
years, the GIS technique is becoming popular with its
advantage in analysing complex terrain characteristics.
This paper presents an urban flooding forecasting model
based on the GIS technique. This model has several parts,
including model pre-processing, evaportranspiration
calculation, runoff production and runoff routing. The model
pre-processing part defines several terms and sets up the model
structure; the evaportranspiration calculation part is to
calculate the evaportranspiration according to the vegetation
type; the runoff production part is to calculate the runoff
produced on the urban area; while the runoff routing part is
used to calculate the runoff flow on the terrain surface and in
the drainage pipelines. The model simulation can be done using
the ArcGIS software.

the above definition, every inlet can be regarded as a RCP,

but there will then be too many RCPs in a real urban area, and
some of them are very close to each other. So, from the
practical point of view, only parts of these inlets need to be
selected as RCPs to simplify the model computation, and others
can be ignored.
Pipelines and Runoff Routing Network: a pipeline is defined
as the line that connects two RCPs, and it is obvious that every
RCP will connect to another RCP unless this RCP is the outlet
of the whole urban area. The pipeline may be an open channel
or an underground pressure pipeline. The runoff flowing to
one RCP will flow to its connecting RCP via the pipeline.
All the pipelines in an urban area constitute the Runoff
Routing Network(RRN) which definitely decides the runoff
flow of the whole urban area, that is the runoff produced on
the surface will first flow to a RCP, and then flow via one
pipeline to its connecting RCP, and finally to the last RCP of

MODEL PRE-PROCESSING
Terms and definitions

To set up the model, some terms and definitions are presented,

as follows:
Runoff collecting point: in a urban drainage system, there
usually exists lots of inlets which connect to the drainage
system, and the runoff produced on the surface will first flow
to one of these inlets and then flow to the outlet of the urban
area via the drainage system. In this paper these inlets are
defined as Runoff Collecting Points(RCPs).
There is one special RCP in an urban area, the outlet of the
urban area: all runoff produced in the urban area will flow
finally to this RCP via the Runoff Routing Network, as shown

Fig. 1 Runoff Routing Network

Hydrology: Science & Practice for the 21st Century. Volume II

2004 British Hydrological Society

361

Yangbo Chen
the Runoff Routing Network, i.e., the outlet of the urban area.
Sub-basin derivation

Sub-basin: the whole urban area will be divided into a number

of sub-basins that can be defined as a closed basin, as shown
in Figure 1; the sub-basin will have the following
characteristics:
(1) There will have one and only have one runoff collecting
point in the sub-basin
(2) All the surface runoff produced in the sub-basin will first
flow to the RCP in it
(3) The RCP on one sub-basin should be connected to another
RCP in one of the neighboring sub-basin
(4) The runoff produced in one sub-basin should flow to
another sub-basin via the pipeline.
Sub-basin derivation: the sub-basin will be derived from the
Digital Elevation Model(DEM), the DEM used in this paper
is a square-grid network model, which divides the whole
terrain into a number of square grids as shown in Figure 2.
Based on the DEM, the following parameters can be derived.
Runoff Flow Direction: runoff produced at each grid is
assumed to flow to only one of its eight neighbouring grids
(Moore et al., 1991) as shown in Figure 3, that means the
runoff produced in one grid will only flow to the neighbouring
grid that has the steepest slope, or the runoff in one grid will
only flow to the neighbouring grid that has the lowest elevation
among the eight neighboring grids. Based on this assumption,
the runoff flow direction of every grid can be decided by using
the DEM.
There is one special grid, the grid with RCP in it, which is
called RCP grid. For the RCP grid, the flow direction is 0,
that means the runoff produced on or flow to RCP grid will
not flow to its neighbouring grid but to its connecting RCP
via the pipeline.
Runoff Flow Network: after deriving the flow directions of
all the grids, a runoff flow network can be formed, as shown

Fig. 2 Digital Elevation Model

362

in Figure 4. The runoff flow network definitely decides the

runoff flow path on the sub-baisn.
Runoff Flow Accumulation: the runoff flow accumulation
of the grid is defined as the number of the up grids with its
runoff flow to this grid; the runoff flow accumulation will be
used to derive the sub-basin.
Sub-basin derivation: the sub-basin can be derived from
the runoff flow network based on the value of runoff flow
accumulation. First set a threshold value of the runoff flow
accumulation as F0, such as 400, then the grids with its runoff
flow accumulation as F0 is selected as the last grid of one subbasin, all the grids with runoff flow to this grid to form the
sub-basin, and the first sub-basin is decided. After that, the
runoff flow accumulation of all the remaining grid is calculated
again, but this time, all the grids already included in the subbasin will be removed, and cannot be included in the
calculation. Repeating the above procedure, the second, third
and n-th sub-basin can be derived as shown in Figure 1, and
the whole urban area is divided into n sub-basins.
The flow direction, flow accumulation calculation and subbasin derivation can be done by using ArcGIS software
directly.
Evaportranspiration calculation

The evaportranspiration is calculated on the sub-basin. The

land covers of the sub-basin are classified into three types:
no-infiltration surface, soil surface (with no vegetation) and
vegetated surface. According to the land cover classification,
different evapotranspiration schemes are applied.
No-infiltration surface: it is assumed that there was no
vegetation on the surface, such as with a concrete surface,
and the evaportranspirtation is assumed to be zero.
Soil surface: the soil surface evaporation is calculated as the
following equation:

Ei,t ESi if Pi,t ! ESi

Wi,t 2

Ei,t Pi,t  (ESi  Pi,t )(WM )

i

Fig. 3 Runoff Flow Direction

if

Pi,t  ESi

(1)

Fig. 4 Runoff Flow Network

An urban flooding treatment model based on GIS techniques

where Ei,t is the i-th sub-basin evaporation on the soil surface
at time t, ESi is potential evaporation of i-th sub-basin on soil
surface, Pi,t is the precipitation of i-th sub-basin at time t, Wi,t
is the soil average storage of i-th sub-basin at time t, and WMi
is the average soil storage capacity of i-th sub-basin.
Vegetated surface: the vegetated surface evaportranspiration
includes two parts: the canopy evaporation and transpiration.
Canopy evaporation takes place during wet periods. In this
period the canopy intercepts precipitation that will be depleted
via evaporation. The canopy evaporation is calculated as
follows:

U i ,t 23
MIN U i ,t , EPi (
)
UM i

E i ,t

k x LAI i

ECi (

Wi ,t
WM i

(3)

)2

Aj
j 1

If the precipitation is measured from the digital weather radar,

then the precipitation on every grid, pi , can be measured. As
all the grids have the same area, so the sub-basin average
precipitation is the arithmetical average that can be calculated
as

1 n
pi
ni1

(4)

where ECi is potential transpiration of i-th sub-basin.

Runoff production

Runoff Production is to calculate the runoff produced on the

sub-basin. The input data of this calculation are the sub-basin
average precipitation measurements that have fallen to the
surface of the sub-basin and can be derived from the measured
rainfall from rain gauges or estimated from the digital weather
radar, while the saturation module is employed to calculate
the runoff.
Sub-basin average precipitation

If the precipitation is measured from the rain gauges, then the

Thiesson Polygon needs to be drawn based on the geographical
position of the rain gauges. If the area of i-th polygon with ith rain gauge in it can be measured as Ai , then the sub-basin
average precipitation can be calculated as

The runoff produced on the sub-basin is calculated in two

categories, the runoff on the no-infiltration surface and on the
infiltration surface.
Runoff on no-infiltration surface: the precipitation falling
to the no-infiltration surface will be fully converted to surface
runoff, which can be calculated as following:

RN i ,t

w p
i

(5)

i 1

where p i is the precipitation of i-th rain gauge, while wi is

the area weight of i-th rain gauge which can be calculated as

Pi ,t x FN i

(8)

Where RN i ,t , Pi ,t and FN i are runoff produced on the noinfiltration surface of i-th sub-basin, precipitation falling on
the no-infiltration surface of i-th sub-basin and the area of the
no-infiltration surface of i-th sub-basin, respectively.
Runoff on infiltration surface: the runoff produced on the
soil and vegetated surface will be controlled by the saturation
module, i.e. runoff will be produced only after the soil is
saturated; this means that the precipitation falling on the
infiltration surface will first fill the soil storage, and only after
the soil storage is full, will precipitation falling to the surface
be converted to runoff.
If the sub-basin can be divided into a number of grids with
different soil storage capacity, then the storage capacity
distribution function can be written as (Zhao, 1984):

W
x 1  1 
Wm

(9)

Where W is the average storage of whole sub-basin, Wm is the

average storage capacity of the whole sub-basin, and x is the
percentage of the whole sub-basin that is saturated when the
sub-basin average storage is W ; b is a parameter.
Here we define net precipitation PN i ,t , which is the
difference of precipitation and evaporation, i.e.

(7)

Runoff production model

where LAIi is the Leaf Area Index of i-th sub-basin, and k is a

coefficient that takes the value of 0.2 mm (Xu et al., 1994).
The transpiration takes place when intercepted water is
depleted, which is calculated as:

E i ,t

(6)

(2)

where EPi is potential evaporation over canopy of i-th subbasin, Ui,t is the intercepted water of i-th sub-basin at time t,
and UMi is the interception capacity of i-th sub-basin, which
can be calculated as:

UM i

Ai

wi

PN i ,t

Pi ,t  E i ,t

(10)

Then the runoff generated on the infiltration surface can be

determined as (Todini, 1996):

363

Yangbo Chen
1

If

0 d PN i ,t  (b  1)W m (1 

W b 1
)
Wm
then
1

Qti 't

W b 1 PNi ,t 1 b 1

PNi ,t  Wm  W   Wm [(1 
) 
] (11)

W
W
b
1

m
m

RI i , t

W b1
then
)
Wm
PN i ,t  (Wm  W )

if PN i ,t t (b  1)Wm (1 

RI i ,t

if PN i ,t d 0 then

RI i ,t

(12)

f x 't

(14)

RI i ,t  RGi ,t

(15)

RN i ,t x FN i  RS i ,t x FS i

(16)

where FNi, FSi are the areas of no-infiltration surface and

infiltration surface of i-th sub-basin, respectively.
Runoff routing model

Runoff routing includes two parts, the surface runoff routing

and Pipeline Runoff Routing. The former calculates the routing
of surface runoff produced on the sub-basin from the surface
to the RCP that it flows into, while the later calculates the
runoff flow in the pipelines.
Surface Runoff Routing Model

The Nash linear cascade reservoir method is employed to

simulate the surface runoff routing, the Instantaneous Unit
Hydrograph(IUH) is as follows
t

u (0, t )


1
t
( ) n 1 e k
K x *(n) K

(17)

where n is the number of reservoirs, K is the time constant, n

and K are model parameters.
In this paper, a mathematical procedure (Todini, 1996) is
used to make the runoff routing calculation. The equation for
this calculation can be written as:

364

(18)

where 't is the time length of one stage, Qti is the discharge
of i-th reservoir at time t, Q00 is the surface runoff of the subbasin, and Q kn is the discharge at the sub-basin outlet, i.e., the
runoff at RCP of i-th sub-basin. By using the above equation,
starting from Q00 , the runoff at RCP at time t  't ,
t  2't , t  3't ,, t  m't can be determined.

For the pipeline runoff routing, the flow can be regarded as

an one-dimensional unsteady open channel flow which is
controlled by the following St. Venan equations:

wA wQ
wt  wx 0

wh S  S
0
f
wx

(19)

In this paper, the kinematical wave approximation method

is adopted to find the solution.
CONCLUSION

The surface runoff produced on the i-th sub-basin is as

Ri ,t

Pipeline Runoff Routing Model

where RGi ,t is the underground runoff, f is the steady

infiltration rate. Then surface runoff is determined as

RS i ,t

(13)

where RI i ,t is the runoff produced on the infiltration surface

of i-th sub-basin at time t.
In this paper, the runoff is divided between surface runoff
and underground runoff; the underground runoff is governed
by the infiltration, so it can be determined as:

RGi ,t

2k  't i
't
Qt 
Qti1't  Qti 1
2k  't
2k  't

This paper presents a model for urban flood simulation based

on the GIS technique, which includes several sub-models such
as pre-processing, evaportranspiration calculation, runoff
production and runoff routing. The model simulation can be
done in the ArcGIS software directly, while the DEM is a
prerequisite for the model simulation.
ACKNOWLEDGMENT

This study was supported by the National Natural Science

Foundation of China, funding no. 50179019.
REFERENCES
Moore, I.D., Grayson, R.B. and Ladson, A.R. 1991. Digital terrain
modeling: A review of hydrological, geomorphologic and
biological applications. Hydrol. Process.,14, 330.
Xu, L., Lettenmaier, D.P., Wood E.F. and Burges, S. J. 1994. A simple
hydrologically based model of land surface water and energy fluxes
for GSMs., J. Geophys. Res., 99(D7), 14, 415428.
Chen, Y., Hu, J. and Yu, J. 2003. A flash flood forecast model for the
Three Gorges Basin using GIS and remote sensing data. In:
Weather Radar Information and Distributed Hydrological
Modelling, Y. Tachikawa et. al. (eds). IAHS Publ. no. 282.
Todini, E. 1996. ARNO Rainfall-Runoff Model. J. Hydrol., 175, 339
382.
Zhao, R.J. 1984. Watershed Hydrologic Modeling. Water Resources
and Electric Power Press.
EPA SWMM website, http:// ccee.oregonstate.edu/swmm/0
Hydroworks swebsite, http://www. wallingfordsoftware.com/0
Mouse website, http://www.bossintl.com/html/mouse_details.html/