Early Warning Services for Management of Cyclones over North Indian Ocean: Current Status and Future Scope

Abstract

Tropical cyclones are the most devastating phenomena among all natural disasters, having taken more than half a million lives all over the world in the last five decades. Cyclones are accompanied by very strong winds, torrential rains and storm surges. The havoc caused by cyclones to shipping in the high seas and coastal habitats along the Indian coasts due to above mentioned adverse weather have been known since hundreds of years. The tropical warm north Indian Ocean (NIO), like the tropical North Atlantic, the South Pacific and the NW Pacific, is a breeding ground for the disastrous TC phenomenon. Historically, in terms of loss to human life, the Bay of Bengal TCs have accounted for deaths ranging from 1,000-300,000. The Bay of Bengal has experienced more than 75 % of the total world-wide TCs causing human death of 5,000 or more in last 300 years (Dube et al. 2013).

Full text

PART II
Tropical Cyclones

Early Warning Services for
Management of Cyclones over
North Indian Ocean: Current
Status and Future Scope
M. Mohapatra*, B.K. Bandyopadhyay,
Kamaljit Ray and L.S. Rathore
India Meteorological Department, Lodhi Road, New Delhi – 110003
* mohapatraimd@gmail.com
1. Introduction
Tropical cyclones are the most devastating phenomena among all natural
disasters, having taken more than half a million lives all over the world in the
last ve decades. Cyclones are accompanied by very strong winds, torrential
rains and storm surges. The havoc caused by cyclones to shipping in the high
seas and coastal habitats along the Indian coasts due to above mentioned
adverse weather have been known since hundreds of years. The tropical warm
north Indian Ocean (NIO), like the tropical North Atlantic, the South Pacic
and the NW Pacic, is a breeding ground for the disastrous TC phenomenon.
Historically, in terms of loss to human life, the Bay of Bengal TCs have
accounted for deaths ranging from a thousand to three hundred thousands.
The Bay of Bengal has experienced more than 75% of the total world-wide
TCs causing human death of 5000 or more in last 300 years (Dube et al.,
2013).
The low pressure systems over the NIO are classied (Table 1) based
on the associated maximum sustained surface wind (MSW) at the surface
level (IMD, 2003). The systems with the intensity of depressions and above
are considered as cyclonic disturbances. The ‘cyclone’ is a generic term
associated with a low pressure system with MSW of 34 knots or more. It
corresponds to the denition of tropical storms over other ocean basins like
Pacic and Atlantic Oceans. Over these basins, it is called as TC/Typhoon

90 M. Mohapatra et al.
when the MSW is 64 knots or more corresponding to very severe cyclonic
storm (VSCS) over the NIO.
Table 1: Classication of cyclonic disturbances over the NIO (since 1999)
Low pressure system Maximum sustained surface winds
Low pressure area (L) < 17 knots
Depression (D) 17-27 kts
Deep depression (DD) 28-33 kts
Cyclonic storm (CS) 34-47 kts
Severe cyclonic storm (SCS) 48-63 kts
Very severe cyclonic storm (VSCS) 64-119 kts
Super cyclonic storm (SuCS) 120 kts and above
The reduction of cyclone disasters depends on several factors including
hazard analysis, vulnerability analysis, and preparedness & planning, early
warning, prevention and mitigation. The early warning is a major component
for the south Asian region due to its socio-economic conditions. The early
warning component includes skill in monitoring and prediction of cyclone,
effective warning products generation and dissemination, coordination with
emergency response units and the public perception about the credibility of
the ofcial predictions and warnings. The entire process of cyclone early
warning system is shown in a schematic diagram (Fig. 1). It is important to
continuously upgrade all the components of early warning based on latest
technology for effective management of TCs.
Fig. 1: Monitoring and forecasting process of tropical cyclone (after Mohapatra
et al., 2013a).

Early Warning Services for Management of Cyclones 91
Considering all these, the initiative has been taken recently to modernize
early warning system for TC and hence to maximize relevance and
effectiveness of the cyclone warning during emergent situations. The above
objective is accomplished through:
(i) Modernization of observational system
(ii) Modernization of cyclone analysis and prediction system
(iii) Updating of Standard Operation Procedure (SOP) for cyclone
monitoring and forecasting
(iv) Institutional mechanism with various NWP modelling centres and
disaster management agencies
(v) Building forecast demonstration projects on land falling cyclones
(vi) Value added warning products generation, presentation and
dissemination and
(vii) Measures for enhancing condence of disaster managers and public in
forecast and warning through forecast verication and preparation of
reports
(viii) Capacity building through training
Climatological Characteristics of Cyclones over the NIO
It is now a well known fact of climatology (IMD, 2008) that about 5 to 6
cyclones occur in a year over the NIO prominently during the pre-monsoon
season (March-April-May) and the post-monsoon season (October-November-
December). It accounts for about seven per cent of the global cyclones. The
maximum frequency is in the two months of May and November. Cyclones
generally move in a northwesterly direction. However, they may recurve
sometimes depending upon the environmental conditions. The Bay of Bengal
cyclones more often strike Odisha-West Bengal coast in October, Andhra
coast in November and the Tamil Nadu coast in December. Over 50 per cent
of the cyclones in the Bay of Bengal strike different parts of the east coast
of India, 30 per cent strike coasts of Bangladesh, Myanmar and Sri Lanka
and about 20 per cent dissipate over the sea itself. Percentage of cyclones
dissipating over the Arabian Sea is higher (60%) as the western Arabian Sea is
cooler. Maximum landfall occurs over Gujarat coast (18% of total cyclones in
Arabian Sea) of India followed by Oman coast. Life period of a cyclone over
the NIO is 5-6 days. It has VSCS (64 knots or more) intensity for 2-3 days as
against six days of global average.
About 50% of the depressions develop into cyclone intensity and only
less than 25% of cyclones further intensify into severe cyclones. While it
takes about 2 to 4 days to develop a low pressure area into a depression, the
intensication from a depression to severe or very severe cyclones can occur
in 24 to 48 hrs.
Several efforts have been made in the IMD to update climatological
records on cyclones of the NIO as it provides useful guidance on early
warning and planning of coastal regions. Recently an electronic atlas has

92 M. Mohapatra et al.
been published for tracks of TCs over the Bay of Bengal and Arabian Sea
(IMD, 2008). It is available in cyclone page of IMD’s website with free access
(www.imd.gov.in).
2. Observational Systems for Cyclones
over the NIO
It is important to correctly determine the location and intensity of the cyclone,
as initial error in location and intensity can lead to increase in error in forecast
location and intensity (Mohanty et al., 2010). Hence, there is a need of dense
observational network over the sea and along the coast. The observational
network for cyclone monitoring consists of land-based surface and upper-
air stations, Doppler Weather Radars (DWRs)/Cyclone Detection Radars
(CDRs), satellites, ships and buoy. The synoptic charts are prepared and
analyzed every three hours to monitor the TCs over the NIO. As the NIO is a
data sparse region and a cyclone genesis takes place in mid-oceanic region,
most of the cyclogenesis (location and time) and intensity are determined with
the satellite observations. When the system comes closer to the coast and lies
within radar range, it is monitored with radar followed by satellite. However,
there are cases of genesis near the coast within the radar range which could
not be detected by satellite. When the cyclone lies close to the coast, coastal
observation is given more weightage followed by radar and satellite. Brief
descriptions on various observational aspects are given below.
2.1 Satellite Based Observations
The geostationary satellite, Kalpana-I provides imageries in visible (VIS),
infrared (IR) and water vapour (WV) channels. In addition INSAT-3A is also
equipped with Charged Coupled Device (CCD) cameras capable of providing
imageries in VIS, Near IR (NIR) and Short-Wave IR (SWIR) channels
with greater resolution. During cyclone situation, data from Kalpana-I are
processed at hourly/half hourly intervals to assess the location and intensity. In
addition to above, the products like outgoing long wave radiation, quantitative
precipitation estimates, sea surface temperatures (SST), cloud motion vectors,
water vapour derived wind vector, and isotherm analysis on enhanced infrared
images are also analyzed on operational mode for cyclone monitoring. The
microwave imageries were introduced for monitoring and guidance in a
subjective manner during later part of 2000s. It was used more objectively to
locate the system centre since 2010 by utilizing the cyclone module available
in cyclone forecasting workstation at IMD, New Delhi. Microwave imageries
are more helpful in predicting the structural characteristics and intensication
in short range, as the characteristics of intensication is rst observed in
microwave imageries unlike VIS and IR imageries (Jha et al., 2013). Apart
from Indian satellites, products from other international satellites are also
used for TC monitoring.

Early Warning Services for Management of Cyclones 93
The sea surface wind as estimated by scatterometer-based satellites
(ASCAT, Windsat and OSCAT) is very useful in locating the centre of the
TC (Uhlhorn and Black, 2003). However, it has the limitations as it provides
only two observations. It also suffers from rain contamination and inability
to measure the wind speed more than 50 knots (28 mps). The OSCAT-based
surface winds are being used since November 2009. Gray’s Parameters (Gray,
1968) including SST & Ocean heat content, convective instability, wind shear,
low level relative vorticity, coriolis parameter and upper level divergence are
monitored for genesis and intensication and these parameters are mostly
estimated by satellite technique. Past studies (Kalsi, 2002; Bhatia and Sharma,
2013) have built up a store-house of knowledge on satellite applications in
cyclone monitoring in the NIO.
2.2 Buoy and Ship Observations
Government of India has established a National Data Buoy Programme
(NDBP) at National Institute of Ocean Technology (NIOT), Chennai. Under
this programme, twelve moored data buoys are deployed currently in the NIO.
The data buoys are tted with sensors to measure air pressure, air temperature,
wind speed and direction and sea surface temperature among other parameters.
These buoys have resulted in better monitoring and reduction of location
and intensity error in association with ship and satellite observations. As per
the guidelines issued to Indian voluntary observing eet (IVOF), synoptic
observations are made at the main standard times: 0000, 0600, 1200 and
1800 UTC by the ships. When additional observations are required, they are
made at one or more of the intermediate standard times: 0300, 0900, 1500
and 2100 UTC. Over the NIO, 186 ships are registered under IVOF. The ship
observations were quite high during pre-satellite period (before 1960s). It
gradually decreased with the advent of polar orbiting satellite in 1960s and
reduced further with the introduction of Indian geostationary satellites during
1980s.
2.3 Radar Observations
There are 11 numbers of S-band radar stations viz. Kolkata, Paradip,
Visakhapatnam, Machilipatnam, Chennai, Sri Harikota, Karaikal, Kochi, Goa,
Mumbai and Bhuj (Fig. 4a). Out of these 11 stations, ve stations (Chennai,
Kolkata, Sriharikota, Visakhapatnam and Machilipatnam) are using DWRs
and the remaining stations have conventional S-band radars. Conventional
radar provides information on reectivity and range only, whereas a DWR
provides velocity and spectral width data along with various meteorological,
hydrological and aviation products which are very useful for forecasters in
estimating the storm’s centre, its intensity and predicting its future movement.
A radar image of a matured TC consists of eye, eye wall, spiral bands, pre-
cyclone squall lines and streamers

94 M. Mohapatra et al.
2.4 Surface Observational Network including Automatic
Weather Stations (AWS)
IMD has a good network of surface observatories satisfying the requirement
of WMO. There are 70 departmental manned surface observatories of IMD
at present all along the coast and over Bay and Arabian Sea islands. There
are 21 pilot balloon observatories and 15 radiosonde/radio wind (RS/RW)
observatories. The meteorological data thus collected all over these stations
are used on real time basis for TC monitoring.
2.5 High Wind Speed Recorders (HWSRs)
The high wind speed recorder for TC monitoring has been installed in
Digha (West Bengal), Puri, Paradip and Gopalpur (Odisha), Visakhapatnam,
Machilipatnam and Nellore (Andhra Pradesh), Chennai (Tamil Nadu), Karaikal
(Puducherry), Mumbai (Maharashtra), Veraval and Dwarka (Gujarat). Further
it is planned to install at eight more places in next two years. It can measure
the wind speed upto 250 kmph.
3. Standard Operation Procedure for Monitoring
of Cyclone
Various kinds of analytical procedure are described in Standard Operation
Procedure (SOP) Manual (IMD, 2003, 2013). A systematic check list is
prepared for identication of location and intensity of cyclone. The procedure
necessarily deals with determination of location and intensity along with
other characteristics like associated MSW, estimated central pressure and
pressure drop at the centre, shape and size, radius of outermost closed isobar,
point and time of landfall, if any or area of dissipation etc. with the available
observations in the storm region.
3.1 Monitoring of Genesis of Cyclone
Genesis parameters are evaluated in following steps to monitor the
cyclogenesis:
Step I: SST, depth of 26°C isotherm and ocean thermal energy
Step II: Conditional instability through a deep and moist atmospheric layer
Step III: Pre-existing disturbance
Step IV: Environmental conditions (vertical wind shear, low level vorticity,
upper level divergence etc,)
Step V: NWP and dynamical-statistical model forecasts for genesis
Based on synoptic, statistical, dynamical-statistical and NWP models
guidance, a consensus decision is taken on genesis of depression and its likely
intensication into TC.

Early Warning Services for Management of Cyclones 95
3.2 Determination of Location of Centre and Intensity
of Cyclone
The location of the centre of the TC is determined based on (a) Synoptic, (b)
Satellite (INSAT/METSAT/microwave) and (c) Radar observations. When
the cyclone is far away from the coast and not within the radar range, satellite
estimate gets more weight, though it is modied sometimes with availability
of ship and buoy observations. When the cyclone comes closer to the coast,
radar estimate gets maximum preference followed by satellite. When cyclone
is very close to coast or over the land surface, coastal observations get the
highest preference followed by radar and satellite observations.
3.2.1 Synoptic Technique
In synoptic technique, the centre of the system is determined by considering
the centroid of the wind distribution at the surface level. In the pressure eld,
the location of lowest mean sea level pressure is considered as the centre of the
system (IMD, 2003). The synoptic technique has got serious limitations over
the open sea due to non-availability of sufcient observations. However, the
AWS stations along coast are very useful as they provide hourly observations
on real time basis (Bhatia et al., 2008). The coastal hourly observations help
not only in correctly analyzing the location, but also in determining the landfall
point and time and hence help in adverse weather warning. For intensity
estimation, the available surface observations are taken into consideration to
nd out MSW and number of closed isobars at the interval of 2 hPa within a
specied region around the system centre (IMD, 2003).
3.2.2 Satellite Technique
In the initial stage (depression/deep depression), the centre is determined
from the centre of the low cloud lines (IMD, 2003). There are four types
of cloud pattern (Dvorak, 1984) in TC. In case of shear pattern, when the
convection lies away from the centre, centre is same as the centre of low cloud
lines. As the system intensies and acquires the banding pattern, the centre is
determined from the banding feature using logarithmic spiral. In the central
dense overcast (CDO) pattern, the centre of CDO is the centre of the system.
In the eye pattern, the centre determination is easier and accurate as it is same
as the centre of the eye of the cyclone.
The intensity classication by satellite technique is based on Dvorak’s
technique (Dvorak, 1984; Velden et al., 2006). The intensity of the tropical
system is indicated by a code gure called T Number based on above pattern
recognition technique. Another feature of the technique is the Current
Intensity number (C.I.) which relates directly to the intensity (in term of wind
speed) of the cyclone. The C.I. number may differ from the T number on
some occasions to account for certain factors which are not directly related
to cloud features. The empirical relationship between C.I. number and the

96 M. Mohapatra et al.
MSW are given in Table 2. Third column of the table gives the pressure depths
(peripheral pressure minus central pressure in hPa) as applicable for Indian Sea
area using the relation Vmax = 14.2 × SQRT (pressure depth) following Mishra
and Gupta (1976). As there is no aircraft reconnaissance in the NIO, Dvorak’s
technique has not been veried and also the pressure wind relationship not
veried. Comparison of satellite-based intensity and the best track estimates
of IMD indicate a difference of about 0.5 T (Goyal et al., 2013). Recently the
microwave imageries and brightness temperatures are also used to determine
central pressure and MSW (Jha et al., 2013). However, this technique has not
been validated over the NIO due to non-availability of aircraft reconnaissance.
Table 2: Maximum sustained wind (MSW) speed and pressure depth in
relation to CI number
C.I. number Max. wind speed (knots) Pressure depth (in mb)
1 25 2.0
1.5 25 3.0
2 30 4.5
2.5 35 6.1
3 45 10.0
3.5 55 15.0
4 65 20.9
4.5 77 29.4
5 90 40.2
5.5 102 51.6
6 115 65.6
6.5 127 80.0
7 140 97.2
7.5 155 119.1
8 170 143.3
3.2.3 Radar Techniques
The eye or the centre of the TC can be derived from a continuous and
logical sequence of observations. The geometric centre of the echo-free area
is reported as the eye location. If the wall cloud is not completely closed,
it is still usually possible to derive an eye location with a high degree of
condence by sketching the smallest circle or oval that can be superimposed
on the inner edge of the existing portion of the wall cloud. When the wall
cloud is not developed fully but a centre of circulation is identiable, this
feature is observed and reported similar to the eye. When the eye or centre is
indistinct or outside the range or the radar beam overshoots the inner eyewall
when it does not extend very high, spiral band overlays are used to estimate
the location of the centre (IMD, 1976). Based on observed winds from DWR,
the intensity can be determined (Raghavan, 2013). As radar based wind are
not available at surface level, the wind observations from these techniques are

Early Warning Services for Management of Cyclones 97
converted to 10-metre wind using the suitable conversion technique like those
used in case of aircraft reconnaissance technique in Atlantic.
The location estimation error has been about 55 km over the sea areas
(Mohapatra et al., 2012a; Goyal et al., 2013). According to Elsberry (2003),
the errors in determining the TC centre over the northwest Pacic Ocean can
be upto 50 km by satellite xes, 20-50 km by radar observations and by about
20 km by aircraft reconnaissance. The induction of DWR has reduced the error
in xing the centre of cyclones in radar range. The landfall point estimation
error has been reduced to about 25 km by 2010 mainly due to installation of
coastal AWS. The average error in MSW estimation has reduced over the
years. It could have been T0.5 (05-20 knots or 3-10 mps) with the introduction
of Dvorak’s classication of intensity since 1974.
4. Cyclone Forecasting System
A variety of observational data was used in India till 1960s to forecast the
track intensity and landfall of TCs. Since 1960s, satellite era, added another
feature. There has been rapid development in objective techniques since 1970s
and especially in recent years for forecasting tracks and intensity of TCs in
the NIO. To summarise, following methods are currently used by IMD for TC
track and intensity forecasting:
(i) Statistical technique—Analogue, Persistence, Climatology, Climatology
and persistence (CLIPER)
(ii) Synoptic technique—Empirical technique
(iii) Satellite techniques—Empirical technique
(iv) Radar techniques—Empirical technique
(v) Numerical weather prediction (NWP) models
(vi) Dynamical statistical models
In the synoptic method, prevailing environmental conditions like wind
shear, low to upper level wind and other characteristics are considered. All
these elds in the NWP model analyses and forecasts are also considered. The
development of characteristic features in satellite and radar observations is
also taken into consideration. While, the synoptic, statistical and satellite/radar
guidances help in short-range forecast (upto 12/24 hrs), the NWP guidance
is mainly used for 24-120 hr forecasts. Consensus forecasts that gather all
or part of the numerical forecast tracks and intensity and uses synoptic and
statistical guidance are utilised to issue ofcial forecast.
4.1 NWP Models
There are three types of NWP models for cyclone forecasts, viz., individual
deterministic models, multi-model ensemble (MME) and single model
ensemble prediction system (EPS). Also there are dynamical statistical models
for the purpose of genesis and intensity prediction.

98 M. Mohapatra et al.
4.1.1 Individual Deterministic Model
• Global Forecast System (GFS)
The Global Forecast System (GFS), adopted from National Centre for
Environmental Prediction (NCEP), at T574L64 (~25 km in horizontal)
resolution (incorporating Grid point Statistical Interpolation (GSI) scheme as
the global data assimilation for the forecast up to seven days) is implemented
operational at IMD, New Delhi on IBM-based High Power Computing
Systems (HPCS). The model is run twice in a day (00 UTC and 12 UTC). The
real-time outputs are made available to the national web site of IMD.
• Non-hydrostatic mesoscale modelling system WRFDA-WRF-ARW
The mesoscale forecast system Weather Research and Forecast WRFDA
(version 3.2) with 3DVAR data assimilation is being operated twice daily to
generate mesoscale analysis at 27 km and 9 km horizontal resolutions using
IMD GFS-T574L64 analysis/forecast as rst guess. Using initial and boundary
conditions from the WRFDA, the WRF (ARW) is run for the forecast up to
three days with double nested conguration and horizontal resolution of 27
km and 9 km and 38 Eta levels in the vertical. The model mother domain
covers the area between lat. 25º S to 45º N and long 40º E to 120º E and child
covers whole India.
• Quasi-Lagrangian Model (QLM)
The QLM, a multilevel ne-mesh primitive equation model with a horizontal
resolution of 40 km and 16 sigma levels in the vertical, is being used for
cyclone track prediction in IMD. The integration domain consists of 111 × 111
grid points centred over the initial position of the cyclone. The model includes
parameterization of basic physical and dynamical processes associated with
the development and movement of a cyclone by (i) merging of an idealized
vortex into the initial analysis and (ii) imposition of a steering current over
the vortex area with the use of a dipole. The initial elds and lateral boundary
conditions are taken from the IMD GFS T574L64. The model is run twice
a day based on 00 UTC and 12 UTC initial conditions to provide six hourly
track forecasts valid up to 72 hours.
• Hurricane WRF Model (HWRF)
Recently, under Indo-US joint collaborative programme, IMD adapted
Hurricane-WRF model for cyclone track and intensity forecast over NIO
region. It has nested domain of 27 km and 9 km horizontal resolution and
42 vertical levels with outer domain covering the area of 800 × 800 and
inner domain 60 × 60 with centre of the system adjusted to the centre of the
observed cyclone. The model has special features such as vortex initialization,
coupling with ocean model to take into account the changes in SST during the
model integration, tracker and diagnostic software to provide the graphic and
text information on track and intensity prediction for real-time operational
requirement. The model is run on real time twice a day based on 00 UTC and
12 UTC initial conditions to provide six hourly track and intensity forecasts

Early Warning Services for Management of Cyclones 99
valid up to 120 hours. The model uses IMD GFS-T574L64 analysis/forecast
as rst guess.
• Other Models
IMD also makes use of NWP products prepared by some other operational
NWP Centres like, European Centre for Medium Range Weather Forecasting
(ECMWF), GFS-USA, GFS-NCMRWF, Japan Meteorological Agency
(JMA), United Kingdom Meteorological Ofce (UKMO), Global Tropical
model, Meteo-France etc.
4.1.2 Multi-Model Ensemble (MME)
The MME technique (Kotal and Roy Bhowmik, 2011) is based on a statistical
linear regression approach. The predictors selected for the ensemble technique
are forecast latitude and longitude positions at 12-hour interval up to 120-hour
of ve operational NWP models. The 12-hourly predicted cyclone tracks are
then determined from the respective mean sea level pressure elds using a
cyclone tracking software. A collective bias correction is applied in the MME
by applying multiple linear regression based minimization principle for the
member models WRF(ARW), QLM, GFS(IMD), GFS(NCEP), ECMWF and
JMA. There is also facility in cyclone module of forecasters’ work station to
develop MME using equal weightage to individual model tracks available in
cyclone module.
4.1.3 Ensemble Prediction System (EPS)
As part of WMO Programme to provide a guidance on cyclone forecasts in
near real-time for the ESCAP/WMO member countries, IMD implemented
JMA supported software for real-time forecast over NIO during 2011.
The Ensemble and deterministic forecast products from ECMWF (50+1
members), NCEP (20+1 members), UKMO (23+1 members), MSC (20+1
members) and JMA (20+1 members) are available near real-time for NIO
region. These products include: Deterministic and ensemble track forecasts,
strike probability maps and strike probability of cities within the range of 120
kms four days in advance. The super-ensemble has also been developed based
on above ensembles.
In India, NCMRWF runs the global ensemble forecasting system (GEFS)
conguration consisting of four cycles corresponding to 00Z, 06Z, 12Z 18Z
and 10-day forecasts are made using the 00 UTC initial condition. A T190L28
control that is started with T574L64 analysis is run out to 10 days with 20
perturbed forecasts. The ensemble spread is a measure of the difference
between the members and is represented by the standard deviation with
respect to the ensemble mean. On an average, small (high) spread indicates
a high (low) forecast accuracy. The ensemble spread is ow-dependent and
varies for different parameters. It usually increases with the forecast range,
but there can be cases when the spread is larger at shorter forecast ranges than
at longer.

100 M. Mohapatra et al.
4.1.4 Statistical Dynamical Model for Cyclone Genesis and
Intensity Prediction
A genesis potential parameter (GPP) for the NIO has been developed (Kotal
et al., 2013) as the product of four variables, namely vorticity at 850 hPa,
middle tropospheric relative humidity, middle tropospheric instability, and the
inverse of vertical wind shear. The GPP is operationally used for predicting
cyclogenesis at their early development stages. The grid point analysis and
forecast of the genesis parameter up to seven days are generated on real time.
Region with GPP value equal or greater than 30 is found to be high potential
zone for cyclogenesis.
A statistical-dynamical model (SCIP) (Kotal et al., 2008) has been
implemented for real time forecasting of 12-hourly intensity upto 72 hours.
The model parameters are derived based on model analysis elds of past
cyclones. The parameters selected as predictors are: initial storm intensity,
intensity changes during past 12 hours, storm motion speed, initial storm
latitude position, vertical wind shear averaged along the storm track, vorticity
at 850 hPa, divergence at 200 hPa and SST. For the real-time forecasting,
model parameters are derived based on the forecast elds of IMD GFS model.
4.2 Adverse Weather Forecasting
A TC causes three types of adverse weather, viz., heavy rain, gale wind and
storm surge during its landfall.
4.2.1 Heavy Rainfall
The forecast/warning of heavy rainfall includes: (i) time of commencement,
(ii) duration, (iii) area of occurrence and (iv) intensity of heavy rainfall.
The methods for prediction of heavy rainfall include: (i) synoptic, (ii)
climatological, (iii) satellite, (iv) radar and (v) NWP techniques. While NWP
models provide prediction of rainfall for different lead period; satellite and
radar provides quantitative precipitation estimates during past 3/12 hrs. The
intensity and spatial distribution of rainfall estimated by satellite and radar are
extrapolated to issue forecast. In synoptic and climatology method, synoptic
climatology of rainfall intensity and spatial distribution are used. The nal
forecast is the consensus arrived from various methods as mentioned above.
4.2.2 Gale Wind
The forecast of gale wind includes: (i) time of commencement, (ii) duration,
(iii) area of occurrence and (iv) magnitude of gale wind. The methods for
prediction of gale wind include: (i) synoptic, (ii) climatological, (iii) satellite,
(iv) radar, (v) NWP and (vi) dynamical statistical techniques. In the satellite
method, region of maximum reectivity and mesoscale vortices are assumed
to be associated with higher wind. In radar technique, the direct wind
observation are available through uniform wind technique, ppv2 product
and radii velocity measurements. The wind estimates from satellite and radar

Early Warning Services for Management of Cyclones 101
and other observations are extrapolated to forecast the wind. MSW is also
available from other sources like scatteometry wind from satellite, buoy and
ships apart from estimate by Dvorak technique. Though the wind forecasts
by the NWP models are underestimated the initial condition of wind from the
model can be corrected based on actual observations and accordingly model
forecast wind can be modied. The forecast based on dynamical statistical
model also can be utilised in the similar manner.
4.2.3 Storm Surge
Storm surge is the rise of sea water above the astronomical tide due to cyclone.
The storm surge depends on pressure drop at centre, radius of maximum wind,
point of landfall and interaction with sea waves, astronomical tide, rainfall,
river run off, bathymetry, coastal geometry etc. The forecast of storm surge
includes: (i) time of commencement, (ii) duration, (iii) area of occurrence
and (iv) magnitude of storm surge. The methods for prediction of gale wind
include: (i) IMD Nomogram (Ghosh model), (ii) IIT Delhi Model (Dube et al.,
2013) and INCOIS, Hyderabad model. INCOIS model also provides coastal
inundation in terms of aerial extent and height of inundation.
4.3 SOP for Forecasting and Decision Support System (DSS)
An SOP (IMD, 2003, 2013) is followed for analyzing various forecast guidance
available from different sources as discussed in previous sections. There
is well dened road map and check list for this purpose. The TC analysis,
prediction and decision-making process is made by blending scientically
based conceptual models, dynamical and statistical models, meteorological
datasets, technology and expertise. For this purpose, a decision support
system (DSS) in a digital environment is used to plot and analyse different
weather parameters, satellite, radar and NWP model products. In this hybrid
system, synoptic method could be overlaid on NWP models supported by
modern graphical and GIS applications to produce high quality analyses
and forecast products. The cyclone module installed in this system has the
following facilities:
• Analysis of all synoptic, satellite and NWP model products for genesis,
intensity and track monitoring and prediction
• Preparation of past and forecast tracks upto 120 hrs
• Depiction of uncertainty in track forecast
• Structure forecasting (forecast of wind in different sectors of TC)
However, all the data are not still available in cyclone module. For better
monitoring and prediction, additional help is taken of ftp and websites to
collect and analyse:
• Radar data and products from IMD’s radar network and neighbouring
countries
• Satellite imageries and products from IMD and international centres
• Data, analysis and forecast products from various national and international
centres

102 M. Mohapatra et al.
The automation of the process has increased the efciency of system,
visibility of IMD and utility of warning products (Mohapatra et al., 2013a).
4.4 Forecast and Warning Products
4.4.1 Track Forecast Products
Considering recent development in prediction capability, IMD introduced the
objective cyclone track forecast valid for next 72 hrs in 2009 and upto 120
hrs in 2013. The TC forecast is issued six times a day at the interval of three
hours, i.e. based on 00, 03, 06, 09, 12, 15, 18 and 21 UTC observations. The
forecasts are issued about three hours after the above mentioned observation
time. An example of the product during cyclone, Thane is shown in Fig. 2a.
4.4.2 Cone of Uncertainty in Track Forecast
The “cone of uncertainty (COU)”—also known colloquially as the “cone of
death,” “cone of probability,” and “cone of error”—represents the forecast
track of the centre of a cyclone and the likely error in the forecast track
based on predictive skill of past years. The COU in the forecast of IMD has
been introduced with effect from the TC, ‘WARD’ during December, 2009
(Mohapatra et al., 2012c). It is helpful to the decision makers as it indicates
the standard forecast errors in the forecast. A typical example of COU forecast
showing the uncertainty circles for different forecast periods are shown in
Fig. 2b. The observed track lies within the forecast COU in about 60-70% of
the cases like other Ocean basins like northern Atlantic and Pacic Oceans.
4.4.3 Quadrant Wind Radii Forecasting
The cyclone wind radii representing the maximum radial extent of winds
reaching 34kts, 50kts and 64kts in each quadrant (NW, NE, SE, SW) of cyclone
are generated as per requirement of ships. The initial estimation and forecast
of the wind radii of TC is rather subjective and strongly dependent on the data
availability, climatology and analysis methods. The subjectivity and reliance
Fig. 2: (a) A typical example of observed and forecast track of depression which
later on became the very severe cyclonic storm, Thane and (b) a typical graphical
presentation of quadrant wind forecast during cyclone, Thane in 2011.

Early Warning Services for Management of Cyclones 103
on climatology is amplied in NIO in the absence of aircraft observations.
However, recently with the advent of easily accessible remotely sensed surface
and near-surface winds (e.g. Ocean Sat., Special Sensor Microwave Imager
(SSMI), low level atmospheric motion vectors and Advanced Microwave
Sounder Unit (AMSU) retrieval methods, DWR, coastal wind observations
and advances in real time data analysis capabilities, IMD introduced TC wind
radii monitoring and prediction product in Oct., 2010. A typical example of
the quadrant wind radii product is shown in Fig. 2b.
4.5 TC Forecasting Skill Accuracy over NIO
All these initiatives as mentioned in previous sections have resulted in
improved cyclone warning service delivery, timeliness of the warning, and
reduction in loss of lives as the outcome. The trends in forecast performance
during 2003-2012 are presented here to illustrate these facts.
4.5.1 Landfall Forecast Accuracy
The landfall point forecast error has reduced signicantly in recent years. The
12 and 24 hr landfall point forecast errors have decreased at the rate of about
16 and 34 km per year respectively during 2003-2012 (Fig. 3). However, the
rate of decrease is relatively less in case of landfall time forecast error.
Fig. 3: Landfall point and time forecast errors of IMD during 2003-2012.

104 M. Mohapatra et al.
4.5.2 Track Forecast Accuracy
The annual average TC track forecast errors have decreased at the rate of
about 5.1 km/year and 7.2 km/year during 2003-2012 for 12 and 24 hr
forecasts respectively. The 36-72 hrs forecast errors also have decreased as
shown in Fig. 4 (Mohapatra et al., 2013b). The skill in tropical cyclone track
forecast has increased at the rate of 8.2% and 4.1% for 12 and 24 hr forecasts
respectively during 2003-2012. There is also signicant increase in skill of
36-72 hr track forecasts. The average track forecast errors (skill) for 24, 48
and 72 hr lead periods are 136 km (32%), 253 km (42%) and 376 km (50%)
respectively during 2009-2012.
Fig. 4: (a) Average tropical cyclone track forecast error and (b) track forecast
skill during 2003-2012.

Early Warning Services for Management of Cyclones 105
4.5.3 Intensity Forecasting Accuracy
The average absolute error (AE)/root mean square error (RMSE) in intensity
(wind) forecast is about 10 (13), 13 (18) and 19 (24) knots, respectively, for
24-, 48- and 72-h forecasts over the NIO as a whole during 2009–2012. The
skill of intensity forecast is about 44%/48%, 60%/58% and 60%/65% for 24-,
48- and 72-h forecasts during 2009–2012 with respect to AE/RMSE. There is
also improvement in terms of reduction in AE and RMSE of MSW forecast
over the NIO (Fig. 5) like that over the northwest Pacic and northern Atlantic
Oceans during 2005–2012 (Mohapatra et al., 2013c). The skill in intensity
forecast compared to persistence method has signicantly improved by about
6% (10%) and 9% (8%) per year, respectively, for 12- and 24-h forecasts
considering the AE (RMSE).
Fig. 5: Absolute and RMS errors of maximum sustained surface wind forecast issued
by IMD during 2003-2012.
5. Cyclone Warning Organisation
At present, the cyclone warning organization of IMD has three-tier system to
cater to the needs of the maritime states at national, regional and local levels
and to carry out international responsibility.

106 M. Mohapatra et al.
The liaison with the Central Government organisations and other agencies
as well as co-ordination and supervision of cyclone warning activities are done
by Cyclone Warning Division (CWD) at New Delhi. CWD, New Delhi is also
functioning as Regional Specialised Meteorological Centre-Tropical Cyclones
(RSMC-Tropical Cyclones), New Delhi and provides the TC advisories to
WMO/ESCAP Panel countries, viz., Bangladesh, Myanmar, Thailand, Sri
Lanka, Maldives, Pakistan and Oman. It also acts as a TC Advisory Centre
(TCAC) for international civil aviation as per the requirement of International
Civil Aviation Organisation (ICAO) and issues the TC advisories to airport
meteorological ofces over NIO and Pacic region for issue of signicant
meteorological (SIGMET) information to different civil aviation authorities
and airlines.
There are three Area Cyclone Warning Centres (ACWCs) at Chennai,
Mumbai and Kolkata and three Cyclone Warning Centres (CWCs) at
Visakhapatnam, Ahmedabad and Bhubaneswar. The ultimate responsibility
for operational storm warning work for the respective area rests with the
ACWCs and CWCs. Area of responsibility of various ACWCs and CWCs is
shown in Table 3.
Table 3: Area of responsibility of ACWC/CWC
Centre Sea area Coastal area* Maritime State
ACWC Kolkata Bay of
Bengal
West Bengal, Andaman
& Nicobar Islands
West Bengal & Andaman
& Nicobar Islands
ACWC Chennai Tamil Nadu, Puducherry,
Kerala & Karnataka
Tamil Nadu, Puducherry,
Kerala, Karnataka &
Lakshadweep
ACWC Mumbai Arabian
Sea
Maharashtra, Goa Maharashtra, Goa
CWC Bhubaneshwar -Odisha Odisha
CWC Visakhapatnam -Andhra Pradesh Andhra Pradesh
CWC Ahmedabad -Gujarat, Diu, Daman,
Dadra & Nagar Haveli
Gujarat, Diu, Daman,
Dadra & Nagar Haveli
*Coastal strip of responsibility extends upto 75 km from the coast line.
5.1 Bulletins Issued for International Users
• Tropical Weather Outlook for WMO/ESCAP panel countries is issued
once daily at 0600 UTC based on 0300 UTC observation and analysis.
It contains convective activity, meteorological situation over the basin,
observed lows, and their potential of intensication within the next 72
hours.
• Special Tropical Weather Outlook for WMO/ESCAP panel countries
is issued ve times a day (based on 00, 03, 06, 12 and 18 UTC). It contains

Early Warning Services for Management of Cyclones 107
current location and intensity, past movement, convective activity, T
number, estimated central pressure and MSW, sea condition, 120 hrs (00,
06, 12, 18, 24, 36, 48, 72, 96 and 120 hrs) forecast track and intensity (text
and graph) from deep depression stage onwards till the weakening of the
system, storm surge guidance and diagnostic and prognostic features.
• Tropical Cyclone Advisory Bulletin for WMO/ESCAP panel countries
is issued every three hourly (based on 00, 03, 06, 09, 12, 15, 18 and
21 UTC). It contains current location and intensity, past movement,
convective activity, T number, estimated central pressure and MSW, sea
condition, 120 hrs (00, 06, 12, 18, 24, 36, 48, 72, 96 and 120 hrs) forecast
track and intensity (text and graph) from deep depression stage onwards
till the weakening of the system, storm surge guidance and diagnostic and
prognostic features.
• TCAC bulletin for issue of SIGMET by Met. Watch Ofces is issued as
soon as any disturbance over the NIO attains or likely to attain the intensity
of cyclonic storm. These bulletins are issued at six hourly intervals based
on 00, 06, 12, 18 UTC synoptic charts and the time of issue is HH+03
hrs. These bulletins contain present location of cyclone in lat./long., max
sustained surface wind (in knots), direction of past movement and estimated
central pressure, forecast position in Lat./Long and forecast winds in knots
valid at HH+6, HH+12, HH+18 and HH+24 hrs in coded form.
5.2 Bulletins Issued at National Level
• Four-stage warning bulletin
The cyclone warnings are issued to central and state government ofcials in
four stages. The First Stage warning known as “Pre Cyclone Watch” issued
at least 72 hours in advance contains early warning about the development of
a cyclonic disturbance in the north Indian Ocean, its likely intensication into
a cyclone and the coastal belt likely to experience adverse weather. This early
warning bulletin is issued by the Cyclone Warning Division and is addressed
to the Cabinet Secretary and other senior ofcers of the Government of India
including the Chief Secretaries of concerned maritime states. The Second Stage
warning known as “Cyclone Alert” is issued at least 48 hrs in advance of the
expected commencement of adverse weather over the coastal areas. It contains
information on the location and intensity of the storm, likely direction of
its movement, intensication, coastal districts likely to experience adverse
weather and advice to shermen, general public, media and disaster managers.
This is issued by the concerned ACWCs/CWCs and CWD at HQ. The Third
Stage warning known as “Cyclone Warning” issued at least 24 hours in
advance of the expected commencement of adverse weather over the
coastal areas. Landfall point is forecast more precisely at this stage.
These warnings are issued by ACWCs/CWCs/and CWD at three-
hourly interval giving the latest position of cyclone and its intensity,

108 M. Mohapatra et al.
likely point and time of landfall, associated heavy rainfall, strong wind
and storm surge alongwith their impact and advice to general public, media,
shermen and disaster managers. The Fourth Stage of warning known as “Post
Landfall Outlook” is issued by the concerned ACWCs/CWCs and CWD at
HQ at least 12 hours in advance of expected time of landfall. It gives likely
direction of movement of the cyclone after its landfall and adverse weather
likely to be experienced in the interior areas.
At CWD, New Delhi, the bulletins are issued from the stage of depression
onwards. During the stage of depression/deep depression, it is issued based on
00, 03, 06, 12 and 18 UTC observations. When the system intensies into a
cyclonic storm, these bulletins are issued at 00, 03, 06, 09, 12, 15, 18 and 21
UTC (every three hourly interval) based on previous observations. The cyclone
warnings are sent on real time basis to the Control Room in the Ministry of
Home Affairs, Government of India, besides other ministries and departments
of the central government, Doordarshan and All India Radio (AIR) at New
Delhi and other electronic and print media and concerned state governments.
Different colour codes are being used since post-monsoon season of 2006
at different stages of the cyclone warning bulletins (cyclone alert-yellow,
cyclone warning-orange and post landfall outlook-red), as desired by the
National Disaster Management.
• DGM’s Bulletin for high govt. ofcials
DGM’s bulletin for high govt. ofcials is issued once a day. It summarises past
24 hrs development in terms of track and intensity and past 24 hrs weather.
Other contents are same as that of bulletin for India coast as discussed in
previous section.
• Tropical Cyclone (TC) vital bulletin
The TC vital contains the vital components required to create a synthetic
cyclone in NWP model. It contains the location, intensity, radius of maximum
wind, radii of 28, 34, 50 and 64 knots wind threshold in four different quadrants
of the cyclone. It is issued four times a day based on 00, 06, 12 and 18 UTC
observation from the deep depression stage.
5.3 Bulletins Issued at Regional and Local Levels
Following user specic bulletins are issued by all ACWCs as per their area of
responsibility:
• Four-stage warning for designated govt. ofcials up to district level
• Audio warnings through cyclone warning dissemination systems along the
coast (installed in disaster managers ofces)
• All India Radio bulletin
• Press bulletin
• Warnings for shermen through All India Radio
• Sea area bulletins

Early Warning Services for Management of Cyclones 109
• Coastal weather bulletin
• Warning for Indian navy
• Warnings for port and sheries ofcials
• Warning for aviation
The CWCs issue all the above bulletins for their area of responsibility
except the sea area bulletin.
6. Warning Dissemination Mechanism
Cyclone warnings are disseminated to various users through telephone, fax,
e-mail, All India Radio, Television and other print and electronic media.
These warnings/advisories are uploaded in the website of IMD (www.imd.
gov.in). Also cyclone warning bulletins are disseminated by SMS to state and
national disaster management authorities. In addition to the above network,
IMD also disseminate warnings to the concerned ofcials and people using
broadcast capacity of INSAT satellite. This is a direct broadcast service
of cyclone warning in the regional languages meant for the selected areas
affected or likely to be affected by the cyclone. There are 352 Cyclone
Warning Dissemination System (CWDS) stations along the Indian coast;
out of these 101 digital CWDS are located along Andhra coast. The ACWCs
and CWCs are responsible for originating and disseminating the cyclone
warnings through CWDS. The bulletins are generated and transmitted every
hour in three languages viz., English, Hindi and regional language. In case
of emergency, police wireless and telecommunication lines of railways and
aviation authorities are also used.
The Cyclone Advisories bulletin for WMO/ESCAP panel countries and
international airports are disseminated through global telecommunication
system (GTS), e-mail and through ftp and TC vitals for research community
and NWP modelling are disseminated through e-mail and ftp.
7. Future Plans
ESSO-IMD continuously expands and strengthens its activities in relation
to observing strategies, forecasting techniques, disseminating methods
and research relating to different aspects of TCs to ensure most critical
meteorological support to disaster managers and decision makers not only in
India but also to the WMO/ESCAP panel countries. It has the following future
plans to further improve the cyclone warning system.
(i) It is planned to further improve the observational network including buoys
over the NIO, DWR and AWS along the coast through modernisation
programme during 12th Five Year Plan.
(ii) Under INSAT-3D programme, there is an advanced imager with six
imagery channels (VIS, SWIR, MIR, TIR-1, TIR-2, and WV) and a

110 M. Mohapatra et al.
nineteen-channel sounder (18 IR and one Visible) for derivation of
atmospheric temperature and moisture proles. It will provide 1 km.
resolution imagery in visible band, 4 km resolution in IR band and 8 km
in water vapour channel. This new satellite will provide much improved
capabilities for cyclone monitoring. In preparation for the reception and
processing of this data, SAC-ISRO has installed a data reception and
processing system to process the data on real time mode and provide
products with respect to cyclone monitoring.
(iii) The FDP on landfalling cyclones over the Bay of Bengal has been taken
up to minimise the error in prediction of TC track and intensity forecasts
and hence adverse weather. During pre-pilot phase (15 Oct-30 Nov 2008-
12), several national institutions participated for joint observational,
communicational and NWP activities resulting in improved forecast and
delivery of services (Mohapatra et al., 2013a). With possible manned
and unmanned aircraft reconnaissance during 2013-15, it will help in
improving TC track and intensity forecasting and hence the adverse
weather warning, as demonstrated in Atlantic and Pacic Ocean basins.
(iv) Currently an effort is underway in which high resolution HWRF model
with the support from NCEP, USA is being used in track and intensity
predictions. However, only atmospheric component is operational at
present and effort will be made to operationalise the coupled model with
inclusion of ocean component, which is being customized by INCOIS,
Hyderabad.
(v) Attempt will be made to assimilate more observational data, especially
remotely sensed satellite and DWR data as it has become necessary to
provide adequate and realistic observations for frequent initialization of
NWP models for short to medium range forecasting of track, intensity
and associated adverse weather.
(vi) With the completion of ongoing modernisation programme and other
initiatives as mentioned above, the cyclone forecast error is likely to
reduce by about 20% by 2015 and by 40% by 2020 from the base year of
2010 according to vision document of Ministry of Earth Sciences, Govt.
of India.
REFERENCES
Bhatia, R.C. and Sharma, A.K., 2013. “Recent advances in observational support from
space-based systems for tropical cyclones”. Mausam, 64: 97-104.
Bhatia, R.C., Das, Sunit, Mohapatra, M. and Roy Bhowmik, S.K., 2008. “Use of
Satellite and AWS data for weather Prediction”. IMD Met. Monogr. Synop. Met.
No. 6/2008. India Meteorological Department, New Delhi.
BMTPC, Govt. of India, 2006. “Vulnerability Atlas of India”. BMTPC, Govt. of India,
New Delhi.

Early Warning Services for Management of Cyclones 111
Dube, S.K., Rao, A.D., Poulose, Jismy, Mohapatra, M. and Murty, T.S., 2013. “Storm
surge inundation in south Asia under climate change scenario”. In: Monitoring
and prediction of tropical cyclones over Indian Ocean and climate change.
U.C. Mohanty, M. Mohapatra, O.P. Singh, B.K. Bandyopadhyay and L.S. Rathore
(eds). Capital Publishing Co., New Delhi.
Dvorak, V.F., 1984. “Tropical cyclone intensity analysis using satellite data”.
Technical Report (NOAA TR NESDIS 11), National Oceanic and Atmospheric
Administration, National Environmental Satellite, Data, and Information Service.
Goyal, Suman, Mohapatra, M. and Sharma, A.K., 2013. “Comparison of best track
parameters of RSMC, New Delhi with satellite estimates over north Indian
Ocean”. Mausam, 64: 25-34.
Gray, W.M., 1968. Global view of origin of tropical disturbance and storms”. Mon.
Wea Rev., 96: 669-700.
IMD, 1976. “Weather Radar Observations Manual”. India Meteorological Department,
New Delhi.
IMD, 2003. “Cyclone Manual”. India Meteorological Department, New Delhi.
IMD, 2008. “Track of storm and depressions over the Indian Seas during 1891-2007,
Cyclone e-Atlas of IMD”. India Meteorological Department, New Delhi.
IMD, 2013. “Cyclone Warning Services: Standard Operation Procedure”. Cyclone
Warning Division, India Meteorological Department, New Delhi.
INCOIS, Hyderabad, 2012. “Coastal vulnerability atlas of India”. INCOIS, Hyderabad.
Jha, T.N., Mohapatra, M. and Bandyopadhyay, B.K., 2013. “Estimation of intensity
of tropical cyclone over Bay of Bengal using microwave imagery”. Mausam, 64:
105-116.
Kalsi, S.R., 2002. “Use of satellite imagery for tropical cyclone intensity analysis and
forecasting”. Met. Monograph, Cyclone Warning Div. No. 1/2002.
Kotal, S.D. and Roy Bhowmik, S.K., 2011. “A multimodel ensemble technique for
cyclone track prediction over north Indian Sea”. Geozika, 28: 275-291.
Kotal, S.D., Roy Bhowmik, S.K., Kundu, P.K. and Das, A.K., 2008. “A statistical
cyclone intensity prediction model for Bay of Bengal”. J Earth Syst Sci, 117:
157-168.
Kotal, S.D. and Bhattacharya, S.K., 2013. “Tropical cyclone Genesis Potential
Parameter (GPP) and its application over the north Indian Sea”. Mausam, 64:
149-170.
Mishra, D.K. and Gupta, G.R., 1976. “Estimates of maximum wind speed in tropical
cyclones occurring in the Indian Seas”. Ind. J. Met. Geophys., 27: 285-290.
Mohanty, U.C., Osuri, Krishna K., Routray, A., Mohapatra, M. and Pattanayak, Sujata,
2010. “Simulation of Bay of Bengal Tropical Cyclones with WRF model: Impact
of initial and boundary condition”. Marine Geodesy, 33: 294-314.
Mohapatra, M., Bandyopadhyay, B.K. and Tyagi, Ajit, 2012a. “Best track parameters
of tropical cyclones over the North Indian Ocean: A review”. Nat. Hazards, 63:
1285-1317.
Mohapatra, M., Nayak, D.P. and Bandyopadhyay, B.K., 2012b. “Evaluation of Cone
of Uncertainty in Tropical Cyclone Track Forecast over north Indian Ocean issued
by India Meteorological Department”. Tropical Cyclone Research and Review, 2:
331-339.
Mohapatra, M., Mandal, G.S., Bandyopadhyay, B.K., Tyagi, Ajit and Mohanty, U.C.,
2012c, “Classication of cyclone hazard prone districts of India”. Nat Hazards,
63: 1601-1620,

112 M. Mohapatra et al.
Mohapatra, M., Sikka, D.R., Bandyopadhyay, B.K. and Tyagi, Ajit, 2013a. “Outcomes
and Challenges of Forecast Demonstration Project (FDP) on Landfalling Cyclones
over the Bay of Bengal”. Mausam, 64: 1-12.
Mohapatra, M., Nayak, D.P., Sharma, R.P. and Bandyopadhyay, B.K., 2013b.
“Evaluation of ofcial tropical cyclone track forecast over north Indian Ocean
issued by India Meteorological Department”. J. Earth Syst. Sci., 122: 589-601.
Mohapatra, M., Bandyopadhyay, B.K. and Nayak, D.P., 2013c. “Evaluation of
operational tropical cyclone intensity forecasts over north Indian Ocean issued
by India Meteorological Department”. Nat Hazards, DOI 10.1007/s11069-013-
0624-z.
Raghavan, S., 1997. “Radar observations of tropical cyclone”. Mausam, 48: 169-188.
Raghavan, S., 2013. “Observational aspects including weather radar for tropical
cyclone monitoring”. Mausam, 64: 89-96.
Uhlhorn, E.W. and Black, P.G., 2003. “Verication of remotely sensed sea surface
winds in hurricanes”. J. Atmos. Ocean. Technol., 20: 99-116.
Velden, C.S. et al., 2006. “The Dvorak tropical cyclone intensity estimation technique:
A satellite-based method that has endured for over 30 years”. Bull. Amer. Met.
Soc., 87: 1195-1210.