IUKWC Workshop Nov16: Developing Hydro-climatic Services for Water Security – Session 1 – Item 3 - SC_Kar
Hydrological Information System for Flood Warning
(A New Initiative of MoES )
Sarat C. Kar
National Centre for Medium Range Weather Forecasting (NCMRWF), Noida, India
Indo-UK Workshop on Developing Hydro-Climate Services for Water
Security: 29 November 2016 -1 December 2016, IITM, Pune
Recent extreme events have exposed the vulnerability
of our society to cope with such situations.
Weather forecasts of such events- Flood causing
potential of the forecasted rainfall.
River Basins in India Flood Prone Regions in India
Due to global warming, it is believed
that there is an intensification of
global water cycle.
However, this intensification is not
clearly evident in regional context
(south Asia and adjoining
Surface hydrology and ground water
exhibit significant interannual
variability over this region due to
interannual variations in the summer
The western and central Himalayas
receive large amount of snow during
winter seasons during the passage
of western disturbances.
• The work of flood forecasting and warning in India is entrusted with the Central
Water Commission (CWC) of Ministry of Water Resources (MoWR).
• The activity of flood forecasting by CWC comprises of level forecasting and inflow
• Level forecasting is done using regression models developed using observed level at
point A with that of level at point B on the river. The skill is reasonably good during
normal flow period.
• The Hydromet Division of IMD at New Delhi provides meteorological support for
flood warning and flood control operations to CWC through its Flood Meteorological
Flood Forecasting in India
Climatology of Snow in Dec, Jan, Feb and Mar
Snow Accumulation and melt Variability over Himalayas
Spatial pattern of snowmelt
(monthly) over the Himalayas
Sarita Tiwari et al (2015) PAAG
Annual Cycle of Snowfall and rainfall based on
station data in Satluj basin
Simulated Discharge using various types of Input datasets
Sarita Tiwari et al (2016) HSJ, under review
Climatology and Interannual Variability of
Rainfall in Narmada basin
Observed Runoff (daily-10 years mean) at
3 stations on Narmada
Rajat Sharma et al (2015) IJEE
Terrestrial Water Storage from GRACE Satellites
For proper estimation Evaporation,
consistent forcing to hydrology model
(especially precipitation, Soil
moisture) etc) and proper modeling
approach is required.
Need of a high resolution Indian Land
data Assimilation System along with
additional land observations
Evaporation from GLDAS
Evap in GLDAS
CWC Gauge Location
Rainfall-Runoff Simulations of extreme monsoon rain events
in Subarnmarekha River basin
Aradhana et al (2016) NatHaz (under review)
Extreme Precipitation & Flooding in Kashmir September 2014
Jammu & Kashmir experienced one of the worst floods in the past 60 years due to unprecedented and
intense rains. The Jhelum River and its tributaries were in spate and caused extensive flooding in
24 hrly 3B42 Rain
Global Model Forecasts from 03 Sept and 04 Sept 2014
Winds at 600hPa & Precip from WRF model simulations (Microphysics Expt)
Sarat Kar and Sarita Tiwari (2016) NatHaz
River Gauge Height
and Discharge in
IMD 0.25degree Obs
Rain From 06Sep to
IITM ERPS at 1degree
11 members T382GFS
11 members T382 CFS
11 members T126 GFS
11 members T126 CFS
MoES (through IMD) provides forecasts at various space and time scales, many instances of extreme rainfall
having potential of causing floods and flooding events are missed.
Mainly due to model limitation of capturing actual magnitude, spatial and temporal distribution of rainfall As
different scales are involved, does a rainfall forecast at a model grid point tell if it has potential to cause flood?
Development of Meteorological Support for Floods and
Water Cycle Studies in India
Proposal submitted to Ministry of Earth Sciences
3-year Action Plan (FY 2017-2020)
IMD, New Delhi
• While the MoES through IMD provides weather forecasts at various space/time scales
and CWC issues warnings/advisories on floods, many instances of extreme rainfall
having potential of causing floods and flooding events are missed causing loss to
economy and life.
• This is mainly due to the model limitation of capturing the actual magnitude of rainfall,
its spatial and temporal distribution.
• Moreover, the meteorological forecasts are not readily usable by various stake holders
such as CWC. Therefore, there is an urgent need to improve and customize
meteorological forecasts specifically for floods.
• The real time flood forecasting is one of the most effective non- structural
measures for flood management.
• For developing an efficient system for improving meteorological supports for flood
forecasting in India, a coordinated action plan with Ministry of Earth Sciences (IMD,
IITM, NCMRWF), and various stake holders is required.
MoES : 7-8 March 2014 in association with the office of the Cabinet Secretary. The meeting noted that at
present, there is no operational integrated forecasting and warning system for floods over India.
The brainstorm meeting recommended to develop a state-of-the-art forecasting system for warning of
floods for different river basins and that MoES may take up this initiative first as a research problem.
MoES: November 26 2014 meeting on flood warning systems. Dr Chidambaram, Principal Scientific Advisor
to PM chaired the meeting. Several strategies to develop a flood warning system in collaboration with
academics and operational groups were discussed.
NCAOR, Goa: April 13-14, 2016 In additions to topics on polar science, scientists deliberated on strategies
for further studies on water cycle and flood forecasting, Modelling of flood especially for glacial lake
outburst floods (GLOF) in Himalayas, floods in mountainous terrain and urban floods.
IITM Pune: September 15 2015 recommended that much of the R&D work may be carried out by research
and academic organizations who have pats experience of hydrological/flood modeling.
Problems to be addressed- Gap Areas
• Many instances of extreme rainfall having potential of causing floods and flooding events are missed causing loss
to economy and life.
• it is not known now if a rainfall forecast at a model grid point has potential to cause flood. There are no
customized skillful (medium-range, extended-range) forecasts for floods at the moment.
• Therefore, there is an urgent need to improve and customize meteorological forecasts specifically for floods.
• Global water cycle is changing due to human interventions. It is not known how in the regional context (the
Indian context) this change is occurring.
• There has been a shift in our water usage (from more surface water use to more ground water use in recent
years). It is not known how this change can be sustained in view of climate change and change in water cycle.
Therefore, the main problems to be addressed through this project are
to develop a meteorological support system for flood warning and forecasting and
monitoring and understanding 3-dimensional features of water cycle and its change over
Jhelum River Basin Satluj River Basin
Koshi River Basin
Brahmani River Basin
Krishna River Basin
Narmada River Basin
Chambal River Basin
•Integration of a suite of diagnostic and prediction models over all spatial and temporal scales for flood
warning and forecasting system (IMD, NCMRWF, IITM)
•Develop and operationalize hydro-meteorological information system including all available archived (past)
and real-time hydro-meteorological data. (IMD, NCMRWF)
•Preparation and verification of probabilistic QPF from ensemble/post processed rainfall forecasts inmeso-
scale, medium and extended-range for all the major river Basins in India (IITM, NCMRWF and IMD)
•Statistical downscaling of Model forecasted QPF to river sub basin scale and post processing of model
rainfall forecast for improved prediction of QPF at river sub basin scale. (NCMRWF, IITM and IMD)
•Preparation of streamflow forecasts at gauge stations using hydrology models and validate streamflow
models for major river basins. (NCMRWF, IITM and IMD)
•Implementation of flood models. Testing of flood analysis and forecasting (warning) modeling system using
bias corrected model forecasted QPF. Validate flood forecasting support system for all the major rivers of
India. (IITM, NCMRWF and IMD)
•Monitor and analyze the 3-d structure of water cycle and document the change in water cycle in the
region at river basin scale. Document salient features of hydrological processes at river basin scale
•Development of a regional land data assimilation system (NCMRWF, IMD)
•Application of spaced based gravity observations for terrestrial water storage change (to monitor ground
water) and to develop application systems for flood early warning (NCMRWF)
•Initiate work on modeling of GLOF (NCMRWF)
•Operationalize the flood warning system in IMD (IMD, NCMRWF, IITM)
•Real-time test runs of flood warning modeling system and Implement the flood warning system in IMD
along with a web-based dissemination system. (NCMRWF and IMD)
Development of flood forecasting models, decision support systems and changing water cycle (Academic
and Research organizations through extra-mural funding)
• Different strategies are required for forecasting and warning of floods.
• Extreme rainfall events occurring for a short period need a different strategy,
• floods caused by large scale heavy rainfall persisting for more than 2 days over a region
need a separate strategy.
• Temporal and spatial scale of extreme precipitation is much smaller, its predictability
also becomes low and the forecast skill rapidly decreases with time.
• Therefore, extreme rainfall events cannot be predicted precisely many days in advance.
• However, these events can be predicted at least 24-48 hours in advance.
• Based upon a large-scale weather forecast model output, general areas where the
important small scale systems are likely to form can often be predicted in advance.
• However, predicting the location, timing, and severity depends upon continual
monitoring using dense observations.
• For optimal decision-making, users need to consider the range of likely outcomes using
• Probabilistic forecasts are generated to account uncertainties involved in predicting
quantum of rainfall.
• From the probabilistic forecasts, an ensemble of scenarios for flooding can be
generated with the uncertainty limits.
• Prediction of extreme weather events needs to be a multi-tiered process (first an
advisory followed by a watch and then an outlook) that reflects growing uncertainty as
forecast lead-time increases.
• Large scale floods are caused due to significant synoptic forcing (weather systems)
which has predictability for more than 3 days.
• Therefore, it is relatively easier for forecasting and warning of such events.
The flood forecasting system will use a high resolution numerical weather prediction
model with the state-of-the-art data assimilation to predict quantum of rainfall (QPF)
over the river basins.
Rainfall forecasts from IITM/IMD medium-range/meso-scale models and IITM extended
range system along with that of NCMRWF models shall be input to hydrology models.
The structure of hydrology and flood models should be simple and it should not
have excessive input requirements, but at the same time the forecasted flood must be
as accurate as possible.
For calibrating as well as for initializing these models, we need many observations of
hydro-meteorological parameters like temperature, rainfall, river run-off, evaporation,
and several soil parameters as well as information on inundation area of past floods.
The present observational network over these basins is not sufficient to cater to this
need. Therefore, it is proposed to augment the present hydro-meteorological observing
systems over the river basins by installing many new observational platforms.
• As a first step, raw model forecasts (each ensemble member as well as ensemble mean)
shall be utilized to generate an ensemble of stream flow forecasts.
• These forecasts shall be further subjected to statistical downscaling to basin scale using
stochastic weather generators, Bayesian hierarchical modeling of extremes.
• QPFs from multiple models (GFS, CFS, UKMO, WRF etc) shall be combined using suitable
and advanced statistical methods.
• The work shall be carried out in collaboration with various Indian groups (at IITs, IISc,
Universities, CWC, MoWR, Reservoir/dam authorities etc).
• The project envisages international collaboration with expert agencies and academicians
outside India. Some of the leading organizations are USGS, ERDC, Columbia University
and University of Colorado, USA, Centre for Ecology and Hydrology in IK.
• Will link up with S2S project and World Flood Awareness program led by WMO and
River Area AWS/ARG Agro-AWS Snow
Koshi 74,500 km2 120 7 7 1 1
Narmada 98,796 km2 158 10 10 1
Jhelum 47,528 km2 76 5 8 5 1
Satluj 66,317 km² 106 7 12 7 1
Chambal 31,460 km² 50 3 3 1
Brahmani 39,033 km² 62 4 4
Tungbhadra 71,417 km2 114 7 7
Augmentation of Observing Network:
Proposal for a
Joint MoES-MoWR Centre for River Forecasting (JCRF)
• The JCRF shall have an apex body with Secretary MoES, Secretary, MoWR, Director
General, IMD and Chairman, CWC, member from NDMA, Irrigation/disaster
management Secretary from the concerned state as members
• to formulate strategy and review the activities of the Joint Centre.
• The Apex body shall also have some international and national experts to provide
scientific input/ advice to the Apex body.
• Proposal for such a Centre shall be submitted separately.
• During the Action Plan period (2017-2020), several meetings and workshops shall be
organized to fine tune the proposal for its effective implementation.
The primary tasks of the JCRF shall be:
(i) Daily forecasts for flood and water management
(ii) Seasonal river water availability forecasts for water management
(iii) Flash flood warning support
(iv) Sediment transport
(v) Sustenance of River Ecosystem (may be later stage)
• The JCRF shall integrate short/medium/extended-range and seasonal Forecasts from
IMD/NCMRWF/IITM into hydrological and flood modeling systems.
• Weather and hydrology forecasts will be integrated into daily operations.
• JCRF forecasts shall develop/use various hydrology models including snow models and
• Improved prediction of Floods is important for economy of the country.
• Weather and climate forecasts have to be suitably integrated with hydrology and flood
• Real-time monitoring of 3-d aspects of water cycle can be done with additional
observations and a regional land data assimilation system
• An effective partnership with all stakeholders (MoES- IMD, IITM, NCMRWF with MoWR-
CWC and academic and research organization in India and abroad) is needed.
• The current proposal to MoES on “Meteorological Support for Floods” aims to bridge
the gap between stakeholders and to develop and integrated flood warning system.