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

net/publication/351436885

Development of an urban flood model for Bengaluru city, Karnataka, India

Article  in  Current Science · January 2021

DOI: 10.18520/cs/v120/i9/1441-1448

CITATIONS READS

0 424

21 authors, including:

P. P. Mujumdar Shubha Avinash

Indian Institute of Science Indian Institute of Science
135 PUBLICATIONS   4,447 CITATIONS    5 PUBLICATIONS   22 CITATIONS   

SEE PROFILE SEE PROFILE

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

Urban Flood Model for Bangalore City View project

UPSCAPE : Upscaling Catchment Processes for Sustainable Water Management in Peninsular India View project

All content following this page was uploaded by P. P. Mujumdar on 09 May 2021.

The user has requested enhancement of the downloaded file.

 GENERAL ARTICLES

Development of an urban flood model for

Bengaluru city, Karnataka, India
P. P. Mujumdar*, M. S. Mohan Kumar, G. S. Sreenivasa Reddy, Shubha Avinash, Ila Chawla,
G. S. Kaushika, Aisha Sharma, Saimy Davis, Tom Thomas, Claris Ann Thomas,
Vanishri S. Jahagirdar, H. M. Arpitha, K. Parashuram and Anjan Kumar

In the recent past, a steep increase in the frequency and intensity of flash floods in urban regions
has resulted in significant damage to life and property, necessitating a holistic approach to address
the problem. An integrated urban flood model for Bengaluru city, Karnataka, India has been deve-
loped with a modular approach, which includes real-time rainfall and weather monitoring, and a
rainfall forecast system feeding into a comprehensive hydrologic flood model. An output manage-
ment and information dissemination channel for timely disaster preparedness and response is inte-
grated in the framework, with active involvement of the government. The framework and outcomes
of the system are summarized and presented here.

Keywords: Disaster preparedness, flash floods, rainfall forecast, urban flood model, weather monitoring.

WITH a surge in urbanization in India over the last few and Geokno along with Bruhat Bengaluru Mahanagara
decades, many moderately sized and populated cities are Palike (BBMP), Bangalore Development Authority
now turning into megacities. Bengaluru, Karnataka in (BDA) and Bangalore Water Supply and Sewerage Board
particular, transformed from being the garden city to the (BWSSB) as the stakeholders. The framework of the
Silicon Valley of India in a short span of about three dec- UFM project is based on the outline from the Integrated
ades, especially since the early 2000s1. Increase in the Urban Flood Management for India (IUFM) project (2013–
developed areas in the city and the associated changes in 2016; https://sites.google.com/view/itraiufm), wherein
the land surface properties have resulted in an increase in the controlled watershed in the IISc campus along with
the overland flows, leading to severe flood events during other sub-catchments in Bengaluru city were considered
the last decade. There has also been an increase in short- as pilot study areas.
duration high-intensity precipitation events in the city The present project aims at developing a comprehen-
that cause floods. Occurrence of such heavy flood events, sive, near-real time flood forecast model for Bengaluru
year after year, causing huge disruption to normal life and city. This involves the development of a precision fore-
resulting in significant economic loss, necessitates an cast system using a high-resolution numerical weather
integrated approach to address the urban flooding pro- prediction model setup exclusively for the city. To deve-
blem in a holistic manner. lop an efficient forecast system, a well-established net-
An integrated urban flood model (UFM) for Bengaluru work of water-level sensors, integrated two-dimensional
city is being developed by the Indian Institute of Science flood model, which along with the water levels provides
(IISc), Bengaluru, in collaboration with the Karnataka information on the spread of flood waters (vis-à-vis flood
State Natural Disaster Monitoring Centre (KSNDMC), inundation), an automated information dissemination sys-
Bengaluru in a time-bound project, to mitigate the tem for preparing and mitigating flood-related damages
impacts of floods in the city. The other partners in this and to utilize flood water as a resource is proposed. One
project are the Centre for Study of Science and Techno- of the critical aspects for the development of the model is
logy Policy (CSTEP), Spatika Information Technologies to collect data from various agencies, and use them
appropriately for simulating floods in the city.
P. P. Mujumdar, M. S. Mohan Kumar, Ila Chawla, G. S. Kaushika, The UFM for Bengaluru city comprises six compo-
Aisha Sharma, Saimy Davis, Tom Thomas, Claris Ann Thomas and nents, viz. (i) establishment of sensor and rain gauge
Vanishri S Jahagirdar are in the Interdisciplinary Centre for Water network and data aggregation; (ii) precise rainfall forecast
Research (ICWaR), Indian Institute of Science, Bengaluru 560 012,
India and G. S. Sreenivasa Reddy, Shubha Avinash, H. M. Arpitha,
system; (iii) comprehensive flood modelling; (iv) flood
K. Parashuram and Anjan Kumar are in the Karnataka State Natural information dissemination and feedback; (v) innovative
Disaster Monitoring Center (KSNDMC), Bengaluru 560 064, India. technologies for flood mitigation and management; and
*For correspondence. (e-mail: [email protected]) (vi) documentation and packaging of flood management

CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021 1441

GENERAL ARTICLES

Figure 1. Framework of the urban flood model for Bengaluru city, Karnataka, India.

of about 900 m amsl. The natural terrain of the city has

three major outward-draining valleys, viz. Hebbal Valley
(draining from northwest to east), Koramangala–
Challghatta (KC) valley (draining from south to northeast)
and Vrishabhavathi Valley (draining from north to south;
Figure 2). The valleys include a number of lakes/storage
tanks which are interconnected through natural streams.
Rapid urbanization and encroachments have caused the
lakes to shrink and have reduced their storage/infiltration
capacity2. Study of the land-use pattern indicates that
about 30% of the central part of Bengaluru is impervious,
which results in the generation of higher run-off as over-
land flow during an intense rainfall event1,3.
Bangalore region receives about 850 mm of average
annual rainfall, of which about 80% is received during
the southwest and northeast monsoon seasons. It is, how-
ever, important to note very high spatial variation of the
extreme precipitation events. Figure 3 shows a spatial
map of mean return levels (mm) for a 10-year return
period for annual maximum precipitation. It can be seen
that the mean return levels on the western part of Benga-
luru city are higher by 10% compared with the other parts
for annual maxima4,5.

Establishment of sensor and rain gauge network

Figure 2. Location and major valleys in Bengaluru (BBMP area) with and data aggregation
stormwater and natural drains and lakes (source: Hebbar et al.2).
To monitor the flow during flood events, a network of
system. Figure 1 shows the framework of the project. The 100 automatic rain gauges (ARGs) and 25 water-level
city of Bengaluru (urban) has an area of 709 sq. km, sensors (WLS), across Bengaluru have been installed and
administered by BBMP within the BDA area of are maintained by KSNDMC in collaboration with
1245 sq. km (ref. 1). It is located at an average elevation BBMP. The real-time rainfall data are obtained at a
1442 CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021
 GENERAL ARTICLES

temporal resolution of 15 min from the ARGs through a 5 km spatial resolution according to the requirements of
Subscriber Identity Module (SIM) based communication KSNMDC. The real-time data from ARG sensors and
system to the server at KSNDMC, with one rain gauge automatic weather stations (AWS) were shared five times
for every 7 sq. km. The telemetric WLSs are set up on the between 5 am–7 am daily with the SAC WRF model
storm water drains (SWDs) at various significant loca- through ftp servers to capture the real-time, location-
tions based on the identified flood-vulnerable areas, and specific weather conditions. This facilitates an increased
accessibility to and safety of the sensors. The WLS accuracy in forecasts based on the observed data. In
transmit data at 15 min temporal resolution, which is crit- general, the SAC forecasts are seen to perform well for
ical during high flood events. The networks are equipped light-to-moderate rainfall range. A finer-resolution WRF
with alert systems to keep the authorities informed if the model is being setup at IISc using boundary conditions
level crosses a threshold, helping them to prepare in case given by the US National Centers for Environmental Pre-
of flood events. Figure 4 a shows locations of the diction (NCEP) with a resolution of 25 km, including two
installed WLS and ARG network. nests at 9 and 3 km. The WRF model forecasts at 3 km
The flood modelling requires collection of fine- resolution are now being tested with the observed rainfall
resolution data for mapping the flow patterns. Contour data from the installed ARGs. This high-resolution WRF
map of 1 m spatial resolution provided by BDA has been model will be integrated with the system in subsequent
processed to obtain the digital elevation model (DEM). versions of the model.
Details regarding lakes and stream, watersheds, build-
ings, roads and railways were also obtained from BDA.
Comprehensive flood modelling
The SWD network and the administrative boundaries
were obtained from BBMP. The flood-vulnerable zones
The UFM flood model has been developed using the
classified were obtained from BBMP and other govern-
storm water management model (SWMM) of the United
mental bodies through KSNDMC. Figure 4 b shows the
States Environmental Protection Agency (EPA). SWMM
SWD network and flood-vulnerable areas identified in
is a dynamic rainfall–run-off simulation model designed
Bengaluru city.
specifically for urban regions, and is used to compute the
run-off generated from rainfall after accounting for vari-
Rainfall forecast ous losses from the system. The input datasets required
for the SWMM model set-up, viz. ARG locations and
The rainfall forecasts were obtained from Weather rainfall data, sub-catchments, SWD network (location
Research and Forecasting (WRF) model data available and channel geometry), and impervious area in the form
with the Space Applications Centre (SAC), Indian Space of buildings and roads, topographical data and DEM data
Research Organisation (ISRO), Ahmedabad. The fore- (obtained from contour maps) were obtained from
casts were obtained for 12, 24, 36, 48, 60 and 72 h daily KSNDMC, BDA and BBMP. The nodes, drains, ARG
at 12:30 pm for all the 100 ARG locations with up to locations and sub-catchments were pre-processed, con-
nected and imported from ArcGIS to SWMM. Initial flow
in the drains was assumed to be at 25% of the total capa-
city and maximum flow is the full capacity of the drains.
The surface roughness coefficient (Manning’s n) was
assigned for concrete-lined SWDs as 0.015, and for other
surfaces based on the respective land-use types and the
literature reports6. Lakes act as major sinks for the flood
waters generated by run-off and the existing lakes are
represented as point storage units in the model2. The pre-
cipitation input to the model was added from the installed
rain gauge network. The simulated peak run-off, run-off
volume and critical and flood-vulnerable locations were
obtained using rainfall data of 15 min resolution generat-
ed during a historical extreme event as an input to the
model. The flow was routed kinematically based on
the continuity and momentum equations, which allow
flow to vary spatially and temporally. The results were
obtained in the form of hydrographs for each junction/
node in the model. The outcomes of the SWMM serve as
Figure 3. Spatial map of mean return levels (mm) of annual maxi-
mum precipitation with a 10-year return period for Bengaluru (BDA inputs to the HEC-RAS model to obtain flood inundation
boundary) (Source: Rupa and Mujumdar5). maps.

CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021 1443

GENERAL ARTICLES

Figure 4. (a) Location of automatic rain gauges (ARGs) and water-level sensors (WLS) in Bengaluru
city (BBMP area), and (b) SWD network (source: BBMP) and flood-vulnerable areas (source: KSNDMC).

Figure 5. Comparison of observed and simulated water levels for a location in the Hebbal Valley, Bengaluru city.

Model calibration and validation the 25–26 of June 2020, with a recorded cumulative rain-
fall of 185 mm in 24 h with an hourly peak of 136 mm.
The model was calibrated by repetitive runs with verifica- This latter cumulative peak was the highest recorded rain-
tion based on the recorded water-level data from the WLS fall for Bengaluru since 1891. The model-simulated
network for high flood events that occurred in 2016. inundation depths at the flood vulnerable areas (Figure
Figure 5 shows a comparison of the observed (WLS) and 4 b) were validated with the recorded depths at the
simulated (SWMM model) water levels for a location in respective locations for these two events. Figure 6 shows
Hebbal Valley, for the event that occurred on 8–9 the model outcomes (spread and depth of flood/
September 2020. The model was then run for two high inundation) for KC and Vrishabhavathi valleys.
flood events that occurred on 7–8 May 2019 and 25–26
June 2020.
The city witnessed widespread heavy rainfall on 7 and Model outcomes: near real-time flood mapping and
8 May 2019. The Doddabidrekal ARG station recorded a flood forecasting
cumulative rainfall of 150 mm in 24 h, with an observed
hourly maximum of 98 mm. Similarly, a high-intensity The near real-time flood inundation maps were prepared
rainfall event was observed in the Vrishabhavathi valley on and despatched based on the outcome of the flood model

1444 CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021

 GENERAL ARTICLES

Figure 6. Flood inundation map of (a) KC Valley on 7 May 2019 and (b) Vrishabhavathi Valley on 25 June 2020.

with real-time observed rainfall from the ARGs as inputs, weather information to the citizens of Bengaluru, a
and were correlated with the WLS data for various loca- mobile application, ‘Bengaluru Megha Sandesha’ has
tions. The rainfall forecasts, which are customized based been developed by KSNDMC as a product of the UFM
on the observations from the AWS and ARGs for better project. The in-built features of the application and the
accuracy were procured daily from SAC. The flood information provided by it, on an experimental basis, are
forecasts were obtained with a lead time of 12–72 h, by a first of its kind in the country. The information is based
executing the flood model for the forecasted rainfall data. on real-time data from 100 ARGs (Figure 4) and 12
AWSs installed by KSNDMC across Bengaluru. Rainfall
forecast is based on location-specific outputs from the
Flood information dissemination and feedback
WRF model of SAC. The data include rainfall, surface
temperature, relative humidity, cloudiness, wind speed
The model outcomes and relevant information were dis-
and wind direction. The urban flood model outcomes
seminated as deliverables in three stages. (1) The rainfall
serve as an input to the flood inundation section and
and flood forecast warning system. (2) The post-flood
flood forecast section of the app. The flood-warning sys-
damage alert system. (3) Bengaluru Megha Sandesha
tem is designed to display the occurrence of floods, based
mobile application and Varunamitra dynamic web portal.
on the region-wise flood-vulnerable zones identified and
Based on the forecasted rainfall and outcomes of the
their corresponding threshold rainfall that can initiate the
flood model for the occurrence of a high rainfall event,
flood scenario. The WLS located at various sensitive
the flood and rainfall forecast system provides relevant in-
SWDs aid in the in situ verification of floods. Training
formation for sending alerts to the concerned organizations
has been conducted for relevant technocrats from various
through WhatsApp and messaging services. Details of loca-
zones of Bengaluru to utilize the data from the applica-
tion and intensity of rainfall and the occurrence of floods
tion for flood scenarios. The application enables users to
were communicated through focused chat groups to rele-
look for safer alternative routes based on high-rainfall re-
vant bureaucratic authorities for disaster preparedness and
gions en-route. This is linked to Google Maps for easy
evacuation, if necessary. The system is currently in place
navigation. Figures 7 and 8 show a flow chart with vari-
and has been employed by various governmental bodies.
ous components of the Bengaluru Megha Sandesha flood
The system is also equipped to alert a network of
application and the interfaces for various sections respec-
trained disaster-relief workforce and the relevant authori-
tively. The application is available for free downloading
ties on the inundation and other damages (blockage of
on Android platforms.
roads/electrical transmissions, etc.), through focused
messaging/WhatsApp groups. This information helps in
quick addressal of these issues with minimal inconve- Varunamitra web portal
nience to the citizens.
Varunamitra (www.varunamitra.karnataka.gov.in) is a
Bengaluru Megha Sandesha app dedicated, dynamic web portal for weather information
in the BBMP area. It includes details on the zone-wise
To directly disseminate rainfall and flood forecast infor- static flood vulnerable areas, daily forecasts for rainfall,
mation along with the location-specific near-real time floods, etc. and real-time details of rainfall and weather

CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021 1445

GENERAL ARTICLES

Figure 7. Components of Bengaluru Megha Sandesha flood application.

Figure 8. Bengaluru Megha Sandesha application dashboard and the interface showcasing near real-time rainfall
and flood inundation, and the safe route for navigation.

information from the network of the ARG and AWS. Documentation and packaging of flood
Similar to the Bengaluru Megha Sandesha application, management system
the forecasts are obtained from the WRF model of SAC.
Figure 9 shows the real-time rainfall and flood dash- The project components, data, flood model and mitigation
boards. and management solutions are being thoroughly docu-
mented and packaged. The packaging aims at providing
Innovative technologies for flood mitigation and streamlined solution with an efficient design for an inte-
management grated urban flood management practice. At present, it is
a pilot project for integrated urban flood management
The urban flood model is also being extended to assess solution in India.
the possibility of reuse of flood water through innovative The model is being constantly improved incorporating
technologies, such as the use of low impact developments changes in land-use patterns (e.g. encroachment of lakes
(LIDs). A few LID options include use of porous pave- affecting the inter-lake connectivity, reduction in per-
ment technology and porous bed channels for ground- vious area, etc.) and fine-resolution DEM data available
water recharge, and rainwater harvesting. These are from NRSC. Procurement and assimilation of very fine-
included in the advanced version of the model, being resolution LiDAR terrain data of up to 20 cm resolution
designed to assess its impact in mitigating floods and is also in progress, and will be integrated in the sub-
considering floods as resources and not calamities. sequent version of the model. Laboratory experimentation

1446 CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021

 GENERAL ARTICLES

Figure 9. Real-time rainfall and flood information dashboard in the Varunamitra web portal.

Figure 10. Channel blockage study of Hulimavu sub-basin in Bengaluru. a, Laboratory set-up of field channel replica for flow with blockage.
b, SWMM modelling for simulated flood. Depth of flow increases upstream of the blockage and reduces downstream.

and SWMM modelling of blockages in the drainage sys- Integrated urban flood modelling at ICWaR
tem due to solid waste, service lines and other obstruc-
tions have been carried out as a case study on a channel The Interdisciplinary Centre for Water Research
in the Hulimavu sub-basin (Figure 10). Based on this case (ICWaR), IISc and KSNDMC, have collaborated on a
study, modelling of the floods with channel blockages for project entitled ‘Urban Flood Model for Bangalore City’ to
the whole Bengaluru city is under progress. The WRF mitigate and manage flood events in urban areas of Benga-
forecast model at a spatial resolution of 1 sq. km is being luru city. The project duration is three years (2018–2021)
developed for better forecasts. Experimentation and and is funded by the Department of Science and Techno-
demonstration of LID applications on the ground, and logy (DST), Ministry of Science and Technology, Gov-
also their quantification through modelling are being ernment of India (GoI) under demand driven convergent
carried out as a part of this project. water solutions in mission mode. Details of the project

CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021 1447

GENERAL ARTICLES

are available at https://sites.google.com/site/urbanflood- 5. Rupa, C. and Mujumdar, P. P., Quantification of uncertainty in spatial
modelbangalore. return levels of urban precipitation extremes. J. Hydrol. Eng., 2018,
23(1), 04017053.
A number of projects are in progress at ICWaR, to ad- 6. Huber, W. C. (n.d.), Storm Water Management Model (SWMM)
dress the problem of urban floods. A project on urban bibliography. Athens, Ga.: Environmental Research Laboratory, Of-
modelling and development of multi-sectorial simulation fice of Research and Development, US Environmental Protection
lab and science-based decision support framework to ad- Agency, 1985.
dress urban environmental issues is in progress under the
National Supercomputing Mission funded by DST. The ACKNOWLEDGEMENTS. The work is funded through the follow-
Urban Climate Downscaling Portal (http://ucdp.icwar. ing projects: ‘Urban Flood Model for Bengaluru City’, sponsored by
iisc.ac.in/) has been developed by ICWaR, in collabora- the Department of Science and Technology, Government of India (GoI)
tion with UNESCO to provide a platform with informa- (project no. DST/TM/WTI/DD/2k17/62(G)2), and ‘Advanced Research
tion pertaining to scientific assessments on climate in Hydrology and Knowledge Dissemination’, sponsored by the Minis-
try of Earth Science, GoI (Project No. MoES/PAMC/H&C/41/2013-
change and downscaling techniques, specifically for PC-II). We thank KSNDMC and SAC, Ahmedabad for making availa-
urban areas, with the target audience as students, ble the data used in this study. The authors also thank CSTEP, Spatika
researchers, policy makers and practitioners worldwide. Information Technologies, Geokno, BBMP, BDA and BWSSB for their
role in the execution of this project. We thank the Chairman, Advisory
Committee for the UFM project, Shri S. V. Ranganath (Former Chief
1. Ramachandra, T. V. and Mujumdar, P. P., Urban floods: case study Secretary, Government of Karnataka) and members of the Advisory
of Bangalore. J. Disaster Dev., 2009, 3(2), 1–99. Committee; Dr R. K. Dave, Dr Veena Srinivasan, Prof. B. S. Murty and
2. Hebbar, R. et al., Spatio-Temporal Analysis of Lakes of Bengaluru Shri S. Vishwanath for their valuable suggestions. We also thank Dr
(S-TALAB), 2018. Chandra Rupa, Mr Kanneganti Bhargav Kumar, Mr T. Suresh Kumar,
3. Ramachandra, T. V., Vinay, S. and Bharath H. Aithal, Frequent Ms Sridevi Priya and Ms Srivani, Ms. Lubna for their role in the UFM
floods in Bangalore: causes and remedial measures, ENVIS Techni- project.
cal Report 123, Environmental Information System, CES, Indian
Institute of Science, Bengaluru, 2017.
4. Rajulapati, C. R., Gupta, H. and Mujumdar, P. P., Diurnal variability Received 11 November 2020; revised accepted 5 January 2021
of hydrological variables in urban areas. Urban Climate, 2020, 33,
100669. doi: 10.18520/cs/v120/i9/1441-1448

1448 CURRENT SCIENCE, VOL. 120, NO. 9, 10 MAY 2021

View publication stats