INDIA S FIRST REAL TIME INTEGRATED URBAN …...Overland flow 2. River/channel flow 3. Sewer flow...
Transcript of INDIA S FIRST REAL TIME INTEGRATED URBAN …...Overland flow 2. River/channel flow 3. Sewer flow...
INDIA’S FIRST REAL-TIME INTEGRATEDURBAN FLOOD FORECASTING SYSTEMFOR THE CITY OF CHENNAI
National conference on Flood Early Warning for Disaster Risk Reduction 30-31 May 2019, Green Park Hotel, Hyderabad, India
Presented by
Subhankar KarmakarProfessorCentre for Env. Sci. and Engg. (CESE)Associated Faculty member: IDP in Climate Studies (DST-CoECS)Associated Faculty member: Centre for Urban Science and Engineering (C-USE) Indian Institute of Technology Bombay, Mumbai
RECENTLY COMPLETED RESEARCH PROJECTS ON FLOOD RISK MAPPING
A comprehensive mapping of flood risk in changing climate: An application to Jagatsinghpur District, Orissa
Collaborators: IIT Bombay and National Remote Sensing Centre, Hyderabad
Sponsored by: STC-IITB-ISRO (2014-18, Completed)
Main deliverables: Projection of extreme rainfall events rainfall from GCM outputs, using
statistical and robust downscaling techniques. Combined 1D and 2D hydrodynamic flood modelling using MIKE
FLOOD Evaluation of Social vulnerability of the entire study area based on the
available demographic information and recent census data To generate flood risk map for the entire study area aggregating
information from flood lines, social vulnerability and LU/LC.
Near-Real-Time Urban Flood Forecasting System Collaborators: IIT Bombay, University of Maryland, Northeastern
Univ., and CDAC, PuneSponsored by: MoES, GoI (2014-2018, Completed)
Main deliverables: Improving global forecast system of extreme precipitation events
with regional statistical model: Application of quantile-basedprobabilistic forecasts over Mumbai
Optimal selection of canopy layers, physics options for WRFparameterization for an urban catchment: An application to aflood forecasting system for Mumbai, India
Generation of flood inundation maps under extreme rainfallscenarios
Design of an expert system for flood forecasting and management for the city of Chennai
Collaborators: IIT Bombay, IIT Madras, IISc Bangalore, Anna University, National Remote Sensing Centre (Hyderabad), ICMOM (Chennai),
Sponsored by: Office of the Principal Scientific Advisor, GoI (2016-19, Completed)
Main deliverables: Design of an real time expert system of flood forecasting 3-D visualization of flood propagation
Riverine
Urban
Urban
CHENNAI FLOOD FORECAST TEAM
30 Scientists from 8 Institutes First of its kind in India Institutes
IIT Bombay IIT Madras IISc Bangalore Anna University NCCR (formerly ICMAM), MoES IMD, MoES INCOIS, MoES NCMRWF, MoES
3<Rivers+Estuary+Coast+Urban infrastructure+Water structures>
OVERVIEW OF THE PROBLEM AND COMPLEXITY
CHALLENGES Simulations of urban precipitation: challenging task due to
multiple factors Fine-resolution urban DEM: either classified or not
available The Storm Water Drainage (SWD) for Chennai is a
complex system and the lack of necessary data like invert level, crown etc.
The complex building layer for an urbanized city like Chennai to be modelled using the Flexible Mesh approach.
Forecasted precipitation: spatial bias 3-way coupled flooding modelling: instability Lack of integration between multiple factors High computing time requirements for simulations of flood
inundation and velocity
5
Chennai Municipal Area (CMA)
LiDAR extent
Greater Chennai Corporation (GCC)
Figure. Selection of study area (as per the available data) for flood modelling
Selection of study area
HEC-HMS & HEC-2DIIT MADRAS
Hydrological Modelling
HydraulicModelling
• Chennai Metropolitan Area (CMA) excluding the portion covered by the LiDAR data• The discharge from the hydrological modelling is used as input data to MIKE FLOOD
MIKE FLOODIIT BOMBAY
• The LiDAR extent (orange shaded portion) is considered as the study area• A comprehensive 1D-2D flood modeling framework (flexible mesh)
Overall Framework
For Chennai: Forecasts Available from NCMRWF• 44 ensemble forecasts released on everyday around 3 PM with 0 hrs
GMT initial condition• Deterministic Forecasts released every 6 hours with updating the
initial condition• Challenges:
• Modeling Uncertainty• Understanding forecasts at a
fine resolution• Qualitative assessment of
forecasts based on users’ choice
Year Dates Ensembles2007 Nov 1-Dec 31 232008 Nov 1-Dec 31 232009 Nov 1-Dec 31 232010 Nov 1-Dec 31 232011 Nov 1-Dec 31 232012 Nov 1-Dec 31 232013 Nov 1-Dec 31 232014 Nov 06-Dec 31 112015 Nov 01-Dec 27 11
HINDCAST DATA AVAILABILITY (NCMRWF)
RG Stations in and around Chennai
Station ID Station Name Location Years43278 Minambakkam 13°, 80° 11' 2007, 2008,
2010, 2012, 2013, 2015
43279 Nungambakkam 13° 4', 80° 15' 2007, 2010, 2011, 2013(Dec), 2014, 2015
HOURLY DATA AVAILABILITY
SL Station Location Data
1 TIRUPATI 13° 40', 79° 35' 2009-14
2 KANCHIPURAM 12° 48', 79° 42'
2007, 2009, 2010-14
3 MINAMBAKKAM 13° 4', 80° 15' 2007-2014
4 TIRUTTANI 13° 12', 79° 36'
2007,2009-10, 2012-14
5 TIRUVANNAMALAI 12° 18', 79° 06'2009-2014
6 NUNGAMBAKKAM 13° 4', 80° 15'
2007,2010,2011-14
Daily Rainfall Data Availability (Year 2000 onwards)
Rainfall forecast: training algorithm
44 Ensemble• Ensemble 1• Ensemble 2• Ensemble 3
• Ensemble 44
Maxima
Mean
Minima
Quantile Regression
Station Level Data
Initial Condition: 20171031
Nungambakkam
Date: 2015/12/01
Initial Condition: 20180221
Minambakkam
Initial Condition: 20171031
Flood Data Bank
• Cases• Return Periods (2, 5,10, 25, 50, 75, 100 and 200 years)• Storm Duration (1, 3, 6, 12, 18, 24 hours)• Past Condition (Wet, Normal and Dry)• Tide Conditions (6 conditions, HH and LL with 3 types of
cyclones)• Simulations Completed: 796
Forecast Lead Time
Day 1 Day 2 Day 3
Day 03 PM IST
Meteorological Forecasts Released
Day 05 PM IST
Max Flood Inundation for Next
3 days from Data Bank
44 ensemble forecast with 0 hrs GMT IC
Tide Forecast
U/S Hydrologic Forecasts
Past Conditions
Algorithms (0-72 Hours Prediction)
Peak Rainfall
Total Rainfall
Highest Tide
Past Condition
Flood Inundation Level
Forecasted Rainfall
80pc
85pc
90pc
95pc
99pc
Forecasted Tide
Past Condition:Based on past 3 days rainfall
Look up table
Search in Data Bank
Flowchart with Real Time Flood Simulations at 6 hrs Interval with Single Member Forecast
Forecasted Inundation (Different Percentile)
95 and 99 PC Showing Flood Inundation
Yes No
Real time simulations at ICMAM at 6 hour interval
No Real time simulation
Tide-Surge modeling
Scenarios
WORK GROUPWORK GROUP Dr. K SrinivasaMr. PLN Murty
Prof. Manasa R. BeheraMs Maneesha SebastianMr. Gowrishankar C
Computational Domain
Discretization with unstructured mesh
Bathymetry and bottom friction
Computational Domain
Discretization with unstructured mesh
ADCIRC+SWAN (Ocean Circulation +
Wave Model)
Storm Surge
Model computed storm tide (surge + tide) ANIMATION
PAST PROGRESS
Tide-Surge modelingComputational
Domain
Discretization with unstructured mesh
Bathymetry and bottom friction
Computational Domain
Discretization with unstructured mesh
Scenarios
ADCIRC+SWAN (Ocean Circulation +
Wave Model)
Storm Surge
Maximum water level for Cyclone
VALIDATION
Tide-Surge modelingComputational
Domain
Discretization with unstructured mesh
Bathymetry and bottom friction
Computational Domain
Discretization with unstructured mesh
Scenarios
ADCIRC+SWAN (Ocean Circulation +
Wave Model)
Storm Surge
SCENARIOS
CYCLONE CATEGORY
Cyclonic Storm (CS)Severe Cyclonic Storm (SCS)
Very Severe Cyclonic Storm (VSCS)
TIDE LEVEL
Highest High Water Spring (HHWS)
Lowest Low Water Spring (LLWS)
SCENARIOS
1. HHWS-CS 4. LLWS-CS
2. HHWS-SCS 5. LLWS-SCS
3. HHWS-VSCS 6. LLWS-VSCS
U/S Hydrologic Modeling
• Used Model: HEC-HMS
Flood Modeling Framework A complete three-way hydrodynamic flood modeling in a coastal urban catchment
1D river flow
2D overland flow
River-overland flow interaction (1D-2D)
Tide-river interaction
Sewer-overland flow interaction (1D-2D)
1D sewer flow
Sewer -river interaction (1D-1D)
Schematic diagram of a coastal urban catchment
Rain
SW treatment plant
1. Overland flow2. River/channel flow3. Sewer flow
MIKE FLOODMIKE FLOODMIKE 11 MIKE 21 MIKE URBAN
• MIKE 11 is a fully dynamic one-dimensional modelling tool for bothsimple and complex river and channelsystems flood simulation.
• Tidal and storm surge studies may beperformed using MIKE 11 in rivers andestuaries.
• MIKE 11 has in-built rainfall-runoffeditor which provides facility ofparameterization, data inputting, evenediting of final results.
• various models in-built in MIKE 11 areNorth American Mesoscale Model (NAMModel), Unit Hydrograph Module(UHM), Soil Moisture Accounting Model(SMAP), Urban rainfall-runoff model [a.Time/area method b. Non- linearreservoir (Kinematic wave) method],Flood estimation handbook model (FEHModel).
• MIKE 21 Flow Model is a modeling system for2D free-surface flows.
• MIKE 21 is used in wide range of hydraulic andrelated phenomenon.
• Modelling of tidal hydraulics• Storm surges
• MIKE 21 is having capability to model floodingscenarios considering tidal impact in river.
• This module is commonly called as TidalAnalysis and Prediction Module [MIKE 21 usermanual, 2007].
• This tool analyse time series of hourly tideheights for a specified period of time usingAdmiralty method and Institute of OceanSciences (IOS) Method.
• Majorly this tool is used for:• Analysis of tidal heights• Analysis of tidal currents• Prediction of tidal heights• Prediction of tidal currents
• MIKE URBAN is a complete integration ofGIS and water modelling especially for urbanwater modelling applications.
• Some important applications are:• Storm water management• Prediction of local flooding• Contingency planning and risk
assessment• Water quality assessment
• SWMM and EPANET is in-built in MIKEURBAN for storm water & sewer modellingand water distribution modellingrespectively.
• MIKE URBAN collection system-rainfall-runoff module includes MOUSE engine forrainfall-runoff computation.
• In-built Rainfall-runoff models in MIKEURBAN are: (a) Time-Area method (b)Kinematic wave model (c) Linearreservoir model (d) Unit Hydrograph model.
Input: DEM, Rainfall time series, Tidal time series , Bathymetry, Drainage network layout, River cross section, Time series of river stage data
Output: Discharge-velocity relationship curve, Flood inundation map
Research component: No past effort on urban flood modelling considering extreme rainfall and tidal influence using MIKE Flood
Urban Flood Modeling
Performance of Flood Model
We acquired the maximum observed depth for407 points during the December 2015 floodevent in Chennai. These 407 validation pointswere used to validate the MIKE FLOOD model.
Real Time Data Collection
10 Automatic Weather Stations (AWS)5 Automatic Rain Gauges (ARG)6 Radar type Automatic Waterlevel recorders
ForecastDay
80percentile
85percentile
90percentile
95percentile
99percentile
01.12.2015
How does the model work for 2015 flood?
InitialCondition
ForecastDay
80percentile
85percentile
90percentile
95percentile
99percentile
31.10.2017
01.11.2017 No Flood No Flood
02.11.2017 No Flood No Flood No Flood
03.11.2017 No Flood No Flood
A Recent Example
Visualization of Flood
Acknowledgements
• Organizers: National Remote Sensing Centre (NRSC), ISRO & Central Water Commission (CWC), MoWR RD & GR Under National Hydrology Project
• Office of PSA, GoI• IIT Madras, IISc Bangalore, Anna University, INCOIS, NCCR, IMD• Greater Chennai Corporation• Centre for Environmental Science and Engineering /// Department of
Civil Engineering /// IDP in Climate Studies /// Centre for Urban Science and Engineering; Indian Institute of Technology Bombay
• Ph.D. students• Danish Hydraulic Institute (DHI), New Delhi and Pune
Thank you<[email protected]>