Tech. J. River Res. Inst.

14 (1): 22-28, 2018 (December), ISSN: 1606-9277

FLOOD HAZARD MAPPING OF SURMA RIVER BASIN IN SYLHET

CITY
G. M. Munna1*, M. J. B. Alam1, M. M. Uddin1, and M. T. Rahman1

Abstract
As flood is one of the major disasters of Bangladesh and does immense damage to the lives and livelihood of locals
living near or in the flood prone zones, it is necessary to study all possible water bodies and their nearby areas. The
popularity of using software based data management tools to justify existing vulnerabilities and the overall status of
flood and flood hazard in Bangladesh has been observed in recent years while many studies of flood hazard have
been found in the literature. It was therefore aimed in this current study to determine the status of flood throughout
the years 2006 to 2015 in the district of Sylhet. The discharge capacity of Surma River has been calculated for 25
year return period and a flood hazard map has been generated using GIS which shows the river has not sufficient
capacity to carry all the water during peak flow which results in water logging in Sylhet city. Although the overall
capacity of the Surma River is much larger than the total yearly discharge and runoff, caution is necessary for a
flash flood. At the end, the digital elevation model has been used to understand the elevation of the study area.

Keywords: flood, geo-statistics, GIS map, hazard, ordinary kriging, Surma River

Introduction
Floods are one of the highly devastating natural hazard maps can be of various use. Flood hazard
hazards in South-Asia. It is the most common maps are designed to increase awareness of the
natural disaster of Bangladesh. The country is likelihood of flooding among the public, local
deltaic in shape and is situated at the Ganges- authorities and other organizations. They also
Brahmaputra-Meghna (GBM) basin which is the encourage people living and working in flood-
third largest fresh water outlet of the world’s prone areas to find out more about the local flood
oceans. The country has to bear the enormous risk and to take appropriate action (Environment
pressure of this flow whereas only 7% of the Agency, 2010). Flood hazard maps can be used
total catchment of the GBM basin lies within the by developers to determine if an area is at risk of
territory of Bangladesh. This unique flooding, and by insurers to determine flood
geographical feature makes Bangladesh more insurance premiums in areas where flood
susceptible to regular flooding. Bangladesh is insurance exists. The creation of flood maps
expected to be one of the most affected countries usually combines topographic data with historic
of global climate change. So, the flood or modeled information on extreme sea levels
characteristics are likely to be changed with the and wave heights. This allows the determination
changing climate. Characteristics study of the of the water level at the coast under extreme
flood is fundamental for planning the proper conditions and shows how this water could flood
adaptation strategy and effective flood inland. This is likely to involve the deployment
management. Sylhet city is vulnerable to of storm surge and wave models. Geographic
flooding because of having heavy rainfall during Information Systems (GIS) are frequently used to
the monsoon season. It is located in the north- produce flood hazard maps. They provide an
eastern zone of Bangladesh. The area of Sylhet effective way of assembling information from
city is 26.5 km2 with a population about 479,837 different maps and digital elevation models
(BBS, 2011). Built out of hilly areas this city is (Sanyal and Lu, 2003). Using GIS, the extent of
situated in an area that is higher than sea level, flooding can be calculated by comparing local
from 1 meter to 7 meters (Tanjil, 2016). As the elevations with extreme water levels. Flood
city is on higher land than most another part of Hazard maps will give information on the spatial
the country it has very low chances of long and temporal flood depth and water logging
duration flood. But the heavy rainfall throughout situation at different climate condition.
the rainy season and most of the year, chances of
a flash flood are high. In this manner, flood

1
Dept. of Civil and Environmental Engg., Shahjalal University of Science and Technology, Sylhet.
* Corresponding Author (E-mail: [email protected])
 Tech. J. River Res. Inst. 14 (1): 22-28, 2018 (December), ISSN: 1606-9277

Methodology
Selection of study area Water level data
The study area is Sylhet city which is the main The daily water level data were collected from
city of Sylhet district. River Surma, the primary BWDB for the years 2006 up to 2015 (10 years).
river of Sylhet flows through it. As one of the This dataset was then converted to monthly water
basic requirements of the study area is a Digital level data by selecting the peak/maximum water
Elevation Model (DEM) and SRTM Digital level of a month. This gives a better overview of
Elevation Model of the study area, was used in the river Surma water level. These data set then
taking the decision of selection of the low laying plotted in a graph to show the water level across
flood pone area. In a broader sense, the different time. The Menn-Kendall analysis was
investigated area is lack of any remarkable also carried out to see if the river showed any
stream of Bangladesh, but the area is well trend.
drained by a network of locally important
streams most of which are both structurally Rainfall data
lithologically controlled and dendritic in pattern.
The daily rainfall data was collected from
The relatively major streams are fewer in number
BWDB for the year’s 2006 up to 2015 (10years).
and are of perennial type, that is they flow even
This dataset was then converted to monthly
in the dry season, but during rainy season they
rainfall data by selecting the peak/maximum
flow with their full strength and become able to
rainfall of a month. This gives a better overview
carry large boulders to distant places whereas the
of the Sylhet city rainfall. Then the data was
minor streams are large in number and of
converted into a format (.xlsx) suitable for use in
intermitted type, that is they are seasonal in their
the Arc map software. Rainfall was then
flow, and water ceases to flow during the dry
converted to Runoff and was used for Kriging in
spell.
the Arc map software.
The basic idea is to find out how much water
Runoff calculation
spills out of the river Surma. For that purpose,
Data of the annual rainfall, discharge of river The expected volume of monthly runoff for 50%
Surma and the capacity of the river are probability was estimated for Surma river using
necessary. the total Table 1) monthly rainfall depth in
Sylhet station for the period (2006-2015)

Table 1. Expected runoff depth of monthly rainfall for 50% possibility

Equivalent
Monthly Rainfall Monthly Rainfall Monthly Runoff*
Probability Runoff depth
(mm) (MCM) (MCM)
(mm)
50% 210 87.15 40.96 98.7

*Using runoff coefficient method Where A is the catchment area and V is the
R=k P Eq. (1) runoff volume.
Where,
R = runoff depth (mm) The study area had a catchment area about 1 km2
K = coefficient of runoff (Sogreah, 1978) and the runoff depth was calculated in mm. To
P = Rainfall depth (mm) simplify, the following was obtained with the
unit (𝑚3 )
Runoff volume calculation 𝑅
V= ∗ 1 ∗ 1000 ∗ 1000 = 1000𝑅
1000
Runoff volume is the total rainfall depth of a
catchment area. Therefore, runoff volume is
obtained by the following formula:

V= R * A Eq. (2)

Technical Journal, RRI, (December), 2018 Page 23

 Tech. J. River Res. Inst. 14 (1): 22-28, 2018 (December), ISSN: 1606-9277

Digital Elevation Model (DEM) And k1 k2 are constants that have different values
based on the shape of the reservoir
Digital Elevation Model is one of the essential
Following formula is used from the dam design
data for flood study of an area. DEM data of the
manual (Lawrence and Cascio, 2004).
study area were collected from the NASA Shuttle
Radar Topographic Mission (SRTM). The C = 0.25*1*D*W*T
NASA Shuttle Radar Topographic Mission
(Fowler, 1977), where k1 = 0.25 and k2 = 1, for
(SRTM) digital elevation data (DEMs) were
valley cross section shapes.
downloaded from the SRTM FTP server
The study area has a roughly estimated 11.3km
(ftp://e0srp01u.ecs.nasa.gov/srtm/version2/) for
throwback which was later reduced to 10km for
the study area. The resolution of DEM is 90
the calculation of capacity. This provides a
meter. The DEM data were further processed
marginal volume left out of calculation which
using ArcGIS 10.0 to fill in the no-data voids or
gives a better safety check.
cells. The processing involved the production of
The formula was then simplified as below with
vector contours and the re-interpolation of these
the unit m3
derived contours back into a raster DEM using
ILWIS contour interpolation tool. DEM was used 𝐶 = [0.25 ∗ 1 ∗ 𝐷 ∗ 𝑊 ∗ 𝑇(= 10) ∗ 1000]
to develop flood inundation maps and land types = 2500𝐷𝑊
classification. Difference between water level Width and depth (cross section) of the river was
data obtained from the interpolated water level obtained from BWDB for 3 stations, namely
surface of different return periods and land S#29 (Sheikhghat), S#30 (Mendibag) and S#33
surface values have been considered as (Kanaighat). Among these, the width in the S#30
inundation depth in the study area. The DEM is station was almost double compared to the other
also used for the preparation of a vulnerability two which increased the over-all capacity of the
map for different flood risk elements. study area.

Discharge data Geostatic analysis steps

The daily discharge data were collected from The methodology of this study begins with the
BWDB for the year, 2006 upto 2015 (9years). selection of the study area. The Discharge data of
This dataset was then converted to monthly the Surma river and the Rainfall data of Sylhet
discharge data by selecting the peak/maximum city were obtained from BWDB. This data was
Discharge of a month. This gives a better then processed into the monthly maximum
overview of the river Surma discharge. Then the dataset. The geostatic analysis is done under the
data was converted into a format (.xlsx) suitable kriging method using Arc GIS platform. Various
for use in the Arc map software. kriging methods were used to find better visual
representation.
Reservoir capacity calculation
Geo-statistics
In order to understand the total capacity of River
Surma, calculation of the total volume of the The theory of Geo-statistics has been expressed
study area is necessary. Calculating the reservoir in many textbooks, including those of Issaks and
capacity requires estimating the shape of the Srivastava (1989) and Goovaerts (1977). Here a
reservoir. As the reservoirs are irregular both in brief discussion of the geo-static methods that
cross sections and in long sections it is not easy were used in the study is presented. In geo-
to calculate. In many cases reservoirs are statics, a semi-variogram is used to quantify the
estimated from the reservoir width, the differences between sampled data values as a
throwback and maximum impounded water function of their separation distance, h. In
depth (Lawrence and Cascio, 2004). The formula practice, the experimental semi-variogram *(h),
is based on the equation below with different is calculated as follows
values for the two constants.
1
𝑦∗ℎ = ∑𝑁(ℎ) ) 2
𝑖=1 [𝑧(𝑥𝑖 − 𝑍(𝑥𝑖 + ℎ)] Eq. (4)
2𝑁(ℎ)
C = k1*k2 *D*W*T Eq. (3)
Where, N(h) is the number of sample pairs that
Where, C = Capacity, D = Depth, W = Width, T are separated by a vector h, and Z(x) and z(xi+h)
= Throwback are the values of the variable z at locations of xi

Technical Journal, RRI, (December), 2018 Page 3

 Tech. J. River Res. Inst. 14 (1): 22-28, 2018 (December), ISSN: 1606-9277

and xi+h, respectively. However, for kriging 𝑦(ℎ) = 𝐶0 + 𝐶 when, h >a; Eq. (6)
analysis, an appropriate theoretical model should
be used to fit the experimental data. The most Where, C0 is the y-axis intercept (the nugget
widely used models include the spherical, effect), 𝐶0 + 𝐶 is the “sill”, which is near the
exponential and Gaussian models. The spherical sample variance, and a represents the range of
model used in the study is defined as follows: influence.

𝑦(ℎ) = 𝐶0 + 𝐶[
3ℎ 1 ℎ
− ( )3 ] where, h ≤ a; Among many kriging procedures, Ordinary
2𝑎 2 𝑎 Kriging was chosen due to the lack of sufficient
Eq. (5) data.

Spatial Analysis

Exploratory spatial data analysis

Trend analysis
No Data Global and local outliers (Trend exists) Yes

Yes No
Transform Trend Removal
Kriging

Flood hazard Map

Fig. 1. Flow chart of the geo-static analysis steps

Ordinary Kriging (OK) observations. The values of ƛ𝑖 are weighted to

obtain a sum of unity, and the error variance is
OK assumes that the mean is stationary but
minimized as follows:
unknown. In addition, the OK estimator is known
as the best linear unbiased estimator (BLUE) and 𝑛(𝑢)
{∑𝑗=1 ƛ𝑖 𝛾 (𝑥𝑖 , 𝑥𝑗 ) − 𝜇 = 𝛾(𝑥𝑖 , 𝑥𝑗 )} 𝑗=
is defined as follows (Journel and Huijbregts,
𝑛(𝑢)
1978) 1, … , 𝑛(𝑢) ∑𝑗=1 ƛ𝑗 =1 Eq. (8)
𝑛(𝑢) 𝑛(𝑢)
𝑧 ∗ (𝑥0 ) = ∑𝑖=1 ƛ𝑖 𝑧(𝑥𝑖 ) 𝑤𝑖𝑡ℎ ∑𝑖=1 ƛ𝑖 = 1
Eq. (7) Where, μ is the Lagrance co-efficient for
minimizing the OK estimation variance,
Where 𝑧 ∗ (𝑥0 ) is the OK estimator at location 𝛾(𝑥𝑖 , 𝑥𝑗 ) is the average semivariogram value
𝑥0 , 𝑧(𝑥𝑖 ) is the observed value of of the variable between the observed values and 𝛾(𝑥𝑖 , 𝑥𝑗 )
at location 𝑥𝑖 , ƛ𝑖 is the weight assigned to the represents the average semivariogram value
known values near the location to be estimated between the location xi and the location to be
and n(u) is the number of neighboring estimated. The OK estimation variance (or

Technical Journal, RRI, (December), 2018 Page 4

 Tech. J. River Res. Inst. 14 (1): 22-28, 2018 (December), ISSN: 1606-9277

standard deviation) can be used as a measure of

the estimation uncertainty as follows:
𝑛(𝑢)
𝜎 2∗ (𝑥0 ) = ∑𝑖=1 ƛ𝑖 𝛾 (𝑥𝑖 , 𝑥0 ) + 𝜇 Eq. (9)
Results and discussion
There are several statistical tests available for
testing stationary-ness of time series (Hirsch Et
Al 1993). In the present study, the regression test
for linear trend has been carried out for the
annual water level series from 2006 to 2014 at
Kanaighat SW-266 station, Sylhet SW-267
station of Surma-Meghna River and Islampur
SW-332 station of Dhala River. These tests start
from a null hypothesis (H0) that the observations
are samples from a stationary process. The
likelihood of this hypothesis is evaluated based
on the value of a test statistic, a property of the
data set. A large deviation of the test statistic
from the stationary value is unlikely to be
coincidental. The P-value is the probability that
the deviation of the test statistic from the
homogeneous case is coincidental. If the p value
is less than the significance level α (alpha) =
0.05, H0 is rejected. Rejecting H0 indicates that Fig. 2. The average capacity of Surma River
there is a trend in the time series while accepting
H0 indicates no trend was detected. The p-value Here we can see that on an average, the river has
is a percentage. It tells you how likely it is that a capacity of about 15 thousand m3 with a
the coefficient for that independent variable maximum of about 20 thousand m3 in the
emerged by chance and does not describe a real Machimpur region.
relationship. A p-value of .05 means that there is
Water discharge through River Surma
a 5% chance that the relationship emerged
randomly and a 95% certainty that there is Originating from the Barak River of eastern
indeed a positive trend. This is the “Significance India, Surma river’s discharge data of three
F” value in Excel; some other statistical different stations are featured in the following
programs call it by other names. This measures Fig. 3, where discharge rate decreases from east
the likelihood that the model as a whole to north-west.
describes a relationship that emerged at random,
rather than a real relationship. As with the p-
value, the lower the significance F-value, the
greater the chance that the relationships in the
model are real.

Arc-Map analysis
Results were produced in the form of maps.
These maps include Capacity of Surma river,
Discharge pattern in Surma river, Rainfall
affecting the existing flow and Digital Elevation
Model (DEM) of the study area.

River capacity: From the equation, the

corresponding capacity at that point of river
Surma in m3 can be obtained and plot them in the
arc map software. After that, Geostatic analysis
was carried out with Kriging which gave the Fig. 3. Water discharge (2006-14) pattern in the
following map (Fig. 2) River Surma.

Technical Journal, RRI, (December), 2018 Page 5

 Tech. J. River Res. Inst. 14 (1): 22-28, 2018 (December), ISSN: 1606-9277

Discharge of the river with additional rainfall

water Again Fig. 6 shows a comparison between the
models created in the Arc map with the DEM. It
Rain water runoff has been induced in the can be noted that there are low lying areas quite
following Fig. 4 along with discharge. Here the next to the river.
total discharge is lessened by the rainfall.

Fig. 4. Water discharge with rainfall (2006-2014)

induced pattern in the river Surma Fig. 6. DEM compared with the study area of
Surma River
The following map (Fig. 5) contains information
about the elevation of the study area with respect Flood hazard mapping
to mean sea level. As the study area is mostly in
The total maximum peak discharge through
between -101inch to 291inch or -3 meters to 7.5
Surma River at Sylhet City Corppration (SCC)
meters. It is comparatively on higher land than
is calculated as 321.35 m3/s. Based on discharge
the rest of the country, which helps in flushing
a flood hazard map (for 25 year return period)
out the excess discharge through the river or
has been generated using GIS (Fig. 7)
through the smaller water ways in the study area.

Fig. 5. DEM of the study area Fig. 7. Flood hazard map for Sylhet City

Technical Journal, RRI, (December), 2018 Page 6

 Tech. J. River Res. Inst. 14 (1): 22-28, 2018 (December), ISSN: 1606-9277

Conclusion
There is a decreasing trend of water level in similar trend with slight shifting to the western
Islampur SW332 station and there is no linear part. The years 2012-14 drastically shifted
trend that could be detected at the 5% toward the west and to the middle region of the
significance level in Kanaighat SW 266 and study area as well as the heart of the city. From
Sylhet SW267 stations. River capacity is the DEM, low lying lands near the study area are
sufficiently large enough to keep the flow within in the primary risk of flash floods.
the channel throughout most of the year.
From flood inundation mapping it is clear that
Rainfall increases overall precipitation count in the water bodies in SCC are not sufficient
the river throughout the year. A hypothetical ten enough to carry all the water during the peak
times the regular rainfall induced map shows a flow in near future. So, water logging will be a
significant rise in the precipitation count of the major suffering to the people. Moreover, some
river which shows that the river is quite parts of the city will be inundated due to flooding
sufficiently large to support the excess water. of Surma River. Taking some flood control
management and designing a proper storm water
The tri-yearly analysis shows the discharge trend drainage system facility in all across Sylhet City
changes in recent years. The years 2006-08 Corporation must be done in order to get rid of
shows the high discharge near the Eastern region these vicious sufferings.
of the study area. The years 2009-11 follows a

References
Aziz, F., Tripathi, N. K., Mark, O. and Kusangi, M. https://www.preventionweb.net/files/670_72351.pdf
(2002). A Customized Flood Evacuation System to (Accessed 21 February 2016).
Alleviate Flood by Integrating Hydrodynamic Model
Mike11and GIS. 1st Asia–Pacific DHI Software Islam, M. S. (2007). Flood vulnerability of natural
Conference, Bangkok, Thailand. resources using Remote Sensing and GIS. M.Sc.
Thesis, BUET.
Bhuyian, S. R. (2014). Flood Hazard and Vulnerability
Assessment in a Riverine Flood Prone Area: A Case Mirza, Q. M. M., Dixit, and Nishat, A. (2003). Nat.
Study. Unpublished M.Sc. thesis, BUET, Dhaka. haz. J. 28: 7.

Blaikie, P., Cannon, T., Davis, I. and Wisner, B. Rasid, H., and Pramanik, M. (1990). Visual
(1994). At Risk – Natural Hazards, People’s interpretation of satellite imagery for monitoring floods
Vulnerability and Disasters (Online), London, in Bangladesh. J. Env. Mgt. 14(6): 815-821.

Blasco, F., Bellan, M. F. and Chaudhury, M. U. (1992). Smith, D. I. (1994). Flood damage estimation – A
Estimating the extent of floods in Bangladesh using review of urban stage damage curves and loss
SPOT data. Rem. Sen. of Env. 39: 167-178. functions. Water S. A. 20 (3): 231-238.

Davision, C. M., Robinson S. and Neufeld V. (2004). Tanjil, R. (2016). Flood Hazard Mapping of Surma
Mapping the Canadian Contribution to strengthening river in Sylhet city. Unpublished B.Sc. Thesis,
National Health Research Systems in Low and Middle Shahjalal University of Science and Technology,
Income Countries: A concept paper. Sylhet.

Gain, A. K. (2008). Flood Risk in the Eastern part of UNESCO and UNDP (2008). Tsunami risk assessment
the Dhaka. Unpublished M. Sc. Thesis, IWFM, and mitigation for Indian Ocean: study report.
Bangladesh University of Engineering and Technology.

Technical Journal, RRI, (December), 2018 Page 7