For reliable cyclone detection

Print edition : August 03, 2002

The India Meteorological Department hopes to set up a nationwide network of Doppler weather radars to predict with better accuracy the time of landfall of cyclones, storms and related weather systems.

THE 'super cyclone' that struck the Orissa coast on October 29, 1999 is a grim reminder that the cyclone-warning systems and the techniques of cyclone detection, monitoring, tracking and landfall prediction that are currently in use in India are far from adequate (Frontline, November 26, 1999). The system that the India Meteorological Department (IMD) uses to monitor the formation and various stages of evolution of cyclones is based on a combination of satellite data and conventional observations made using buoys, ships and radars and their analyses.

The imagery (infrared and visible) from the Very High Resolution Radiometer (VHRR) of the geostationary satellite INSAT is predominantly used for space-based information. Satellite-based analysis mainly includes (1) intensity estimation; (2) mean wind flow estimation; and (3) animation of sequences of satellite images and extrapolation of the apparent motion of the cloud system.

When the cyclonic system comes nearer to the Indian coastline (less than 400 km), its subsequent development and movement is monitored by a chain of ten S-band Cyclone Detection Radars (CDRs) set up by the IMD to cover the entire coastal belt. The likely movement of the storm is predicted with the help of track prediction models and by reference to historical data of tropical cyclones over the Bay of Bengal and the Arabian Sea. Depending on the severity of the cyclone, warnings are issued from the Area Cyclone Warning Centres (ACWCs) in Kolkata, Chennai and Mumbai and the Cyclone Warning Centres (CWCs) at Visakhapatnam, Bhubaneswar and Ahmedabad. The cyclone warning process is coordinated by the Weather Central in the Weather Forecasting office of IMD, Pune, and the Northern Hemispheric Analysis Centre (NHAC) at IMD, New Delhi.

The failure of the VHRR aboard INSAT-2E, which was launched in April 1999, resulted in the IMD making an erroneous landfall prediction and consequently issuing flawed warnings on the morning of October 29 that probably resulted in the large number of casualties. The incident highlighted the need for a terrestrial cyclone detection, monitoring and tracking system more reliable than the current network of CDRs and warning systems.

Besides, according to satellite meteorology experts, satellite-based analysis has certain limitations. Tropical cyclones undergoing rapid intensity changes tend to be underestimated. Similarly, being a gross technique, features specific to a given storm tend to get overlooked. To overcome these problems, meteorologists have attempted to integrate INSAT's imagery with data from other satellites such as METEOSAT, GOES and ERS-1. However, such techniques are still in the research stage and it will be some time before they are perfected and operationalised.

At present microwave weather radar networks, particularly Doppler weather radars (DWRs), integrated with other data sources, notably satellites, have emerged the world over as the most powerful tools for the surveillance and monitoring of severe weather systems such as cyclones and for the estimation of precipitation over a large area. The best-known and perhaps the largest network is that of the NEXRAD WSR-88D of the United States, a countrywide network of 160 S-band Doppler radars established in 1991. In India, the existing network of CDRs is based on old technology. Indeed, the first S-band CDR became operational at Visakhapatnam in 1970. The first Indian made S-band CDR, built by Bharat Electronics Ltd. (BEL), was commissioned in Mumbai in 1980.

The IMD has plans to replace its CDR network with modern DWRs in a phased manner. Although the decision to set up the DWR network was taken nearly a decade ago and the money allocated, the first DWR units are being installed only now. Two DWRs, imported from Gematronik Gmbh, a German firm, at a cost of Rs.13 crores each, have been installed at the Regional Meteorological Centres in Chennai and Kolkata. The Chennai DWR is already operational while the one in Kolkata is undergoing tests.

The decision to import DWRs, even though indigenous capability to design and fabricate such radars existed, was taken in view of the urgent need for them. The inordinate delay in their arrival is apparently due to the refusal by the U.S. and Japanese suppliers, who were identified originally in preference to Gematronik, to part with the source code of the system software. Access to source code is essential in case modifications are needed in the system software to adapt it to the Indian context. According to reliable sources, although Gematronik Gmbh has agreed to part with the source code it will do so as and when required and its use will be under the company's supervision.

Interestingly, the parallel effort at indigenous development, undertaken by the Indian Space Research Organisation (ISRO) as a multi-institutional effort about six years ago, has also succeeded and the first indigenous DWR, fabricated by BEL, has been installed at the Sriharikota launch complex, SHAR, after tests in Bangalore. It is now undergoing certain calibration tests. Its performance is reported to be satisfactory and the data products are comparable in quality to those of the imported systems. According to G. Viswanathan, project director, Radar Development Cell (RDC), ISRO, Bangalore, this project cost a total of Rs.9 crores.

The IMD proposes to instal three more systems in the next phase, two Gematronik systems and one to be built by BEL based on the ISRO design. They are proposed to be installed at Paradip in Orissa and Machilipatnam and Visakhapatnam in Andhra Pradesh. The decision to import DWRs even after the successful indigenous development would seem to defy logic. According to S.N. Srivastava, Deputy Director-General, IMD, the Gematronik systems are state-of-the-art ones and the indigenous effort is still to be perfected in terms of algorithms to obtain primary data products and derived products.

An interesting story lies behind the IMD's decision to import DWRs. With the State governments being allowed to seek loans and credits from international financial institutions directly, the Andhra Pradesh government secured World Bank funding to purchase DWRs under its Hazard Mitigation and Emergency Cyclone Recovery Project and floated in 2000 a tender for three S-band DWRs for the Andhra coast. The tender had effectively ruled out the ISRO/BEL DWR because bidding was open only to "eligible source countries" as defined in the World Bank's guidelines. According to informed sources, the World Bank funding now stands lapsed because the process of procurement could not be completed in the specified two-year period. Now that there are no World Bank conditions on the purchase and that government funds are used, all three DWRs could have been indigenous ones, says Viswanathan. "We have decided to place a bulk order for five or six future systems with BEL," says V. S. Ramamurthy, Secretary, Department of Science and Technology (DST), IMD's apex department.

According to Viswanathan, at present BEL's projected cost for a single system is Rs.11 crores but with bulk orders this should come down. Viswanathan has proposed to set up a countrywide network of 32 DWRs, just like NEXRAD, to monitor severe weather systems across the country. This proposal is under active consideration, says Ramamurthy.

SO, what is a Doppler radar? The term radar is the acronym for 'radio detection and ranging', and tropical cyclones are particularly suited for radar observation and monitoring because of the spatial and temporal scale of the system as well as the organised pattern of precipitation in cyclones which the radar can detect. Conventional weather radars operate at wavelengths of 3 to 10 cm; the Indian CDR network operates at 10 cm wavelength (or 3 gigahertz frequency). At these wavelengths the radiation from the radar is scattered back by precipitation and cloud particles.

Weather radars usually operate in the pulsed mode in that they transmit short pulses of radio waves at a rate of about 1,000 pulses a second, with each pulse lasting only about a millionth of a second. Therefore, after each pulse, there is a short time period during which the radar does not transmit. During this time the radar listens to the "reflected echo" from the cloud or any other weather system. This signal is received by the antenna and analysed by the radar signal processor. Based on the variations in amplitude (or power received), the phase or frequency shift and the change in polarisation state of the reflected electromagnetic waves, the radar quantifies the different characteristics of the precipitating systems.

The radar has two modes of rotation - one that changes the angle of elevation (angle of the beam axis with respect to the ground), and the other (rotation around the vertical) that changes the angle of azimuth (angle measured with respect to the north). This enables two types of atmosphere scans: a series of two-dimensional maps called plan-position indicator (PPI), which are scans around the vertical with the elevation angle fixed; and the range-height indicator (RHI), which are scans by varying the elevation (up-down direction) with the azimuth fixed. The radar displays are usually in the form of surveillance scans obtained by rotating the radar for the full angular range or sector scans with the rotation for a chosen range. Radar displays are colour-coded to denote intensity variations. The power returned by a particle to the radar is proportional to the sixth power of the particle diameter. The sum of the sixth powers of the drop diameters in unit volume is called 'reflectivity'. This is one of the primary data product that a radar provides. By the time a weather radar beam hits a storm, it is typically several hundred metres wide. It thus reflects off a volume of cloud, which contains many condensed droplets or particles. Weather radars thus give a "bulk" picture of a storm. By continuous or frequent observation, the movement of the precipitating weather system or individual precipitation cells can be tracked.

A Doppler radar is like a normal radar in that it can measure not only the reflectivity but also the component of the velocity of the system towards or away from the radar (the radial velocity). This is derived from the fact that as precipitation moves towards or away from the antenna, the returning radar pulse will change in frequency. The frequency change is called Doppler shift, and hence the name. It follows from the Doppler principle, which says that the frequency becomes higher as the source approaches us and lower as it recedes from us. We experience this when a train whistle or siren approaches and passes us. The Doppler shift, which is directly proportional to the radial velocity and inversely proportional to the signal wavelength, can be used to determine the wind speed.

The frequency shift is usually of the order of a few kilohertz for atmospheric targets. In a signal frequency of the order of a few GHz, this is too small to be measured directly. The Doppler signal is usually derived by a comparison of the transmitted signal and the returned signal. As the duration of a single pulse is extremely short, about 0.5-2 microseconds, it is not possible to determine the Doppler frequency directly from this one pulse. It is done by measuring the 'phase shift' over several consecutive pulses. The phase of a wave indicates its position relative to a reference position. The phase shift between two instances refers to the shift in the position of the wave relative to the reference position.

Now imagine at time T1 the radar sends out a pulse and it returns a target distance d; at time T2 another pulse is sent towards the same target and it returns a distance d+a because the target is moving. This will result in a phase shift of the returned signal. By knowing the phase shift, the wavelength and the time interval between the pulses, the velocity of the target can be computed. The mathematical technique of Fast Fourier Transform (FFT) or Pulse-Pair Algorithm is used for the calculation of the Doppler frequency from a series of pulses.

If the target is moving sideways so that its distance from a radar does not change, there will be no Doppler shift, because the system has zero radial velocity. Indeed, in precipitating systems velocity in the vertical direction would be much greater than the radial and DWR cannot measure that. However, using three radars located close to each other, it is possible to determine all the three velocity components and obtain a three-dimensional true wind field of any precipitating system. Typically, however, in an operational context, only data set from a single radar is available. A technique known as Velocity-Azimuth Display (VAD), which enables the derivation of vertical wind profile in a limited area around the radar is used in such situations.

However, there are some other limitations of radar observations. The range up to which a ground-based radar can observe is limited only by the earth's curvature (400 km). The detectability of an atmospheric system depends on its distance and height. The maximum range would also depend upon the temperature and humidity, with which radiowave propogation varies. Although shorter wavelengths produce stronger echoes, wavelengths of less than 10 cm are attenuated by rainfall in the path of the signal. In view of these, a network of radars, data-linked in real time, is required to devolve the various sets of observations and obtain more accurate data.

Not everything that the radar captures is from the atmospheric system that is being observerd. Although the transmitted signal is a narrow beam (a couple of degress wide), and that too from a radar that is usually mounted atop a building or tower, by the time it reaches the target it spreads out and gets scattered by other objects in its path. Also, the transmitted pulse pattern has sidelobes of lower intensity sticking out sideways to the main central beam which strike objects in the radar's vicinity. Thus the antenna picks up return signals from a variety of objects besides the target - buildings, trees, hills, towers and so on. This is known as ground clutter and is removed by digital filtering techniques. With the dvelopment of Doppler radars the problem has become relatively simpler as gound clutter usually consists of stationary objects while meteorological systems are in motion which DWRs can detect.

Doppler radars have a further limitation, known as the Doppler Dilemma. This limits their range and velocity determination depending on the pulse repeat frequency (PRF) or the frequency at which pulses are sent out from the radar. When the PRF is high, the radar has less time to sample the atmosphere between pulses. If the radar has less time to sense, it cannot detect objects that are further away from it as compared to a low PRF. Thus, a high PRF has a smaller unambiguous range as compared to a low PRF. The Doppler dilemma states that there is an inverse relationship between the unambiguous range and the unambiguous velocity that a DWR can measure. That is, the greater the PRF, the shorter the unambiguous maximum range and the larger the unambiguous maximum velocity that can be determined, and vice versa.

Velocities that are beyond the range limit will be "folded" back into the range so that a strong outbound velocity beyond the observable range will be interpreted as a strong inbound velocity within the observable range. There are various software algorithms in existence to remove such ambiguities, but none ensures full resolution. There are hardware techniques as well that allow a choice of appropriate combination of PRFs to minimise the ambiguity. Besides these, the indigenous effort proposes to incorporate a coding technique that tags every pulse that is sent out. The coding pattern enables the removal of the ambiguity.

DWRs with computer-controlled operation and digital processing provide three primary products: reflectivity, mean (precipitation particle) velocity, and velocity spectrum. Reflectivity is correlated by statistical techniques with the rate of precipitation through a non-linear empirical relationship to map the distribution of instantaneous rain rates, the rainfall intensity and the cumulative rainfall amounts. Reflectivity can also be related to the liquid water content in the system and the fall speed of precipitation particles. The spread or width of the velocity spectrum can be used to infer turbulence and wind shear in the system, which are useful in recognizing signatures of thunderstorms, tornadoes, microbursts and tropical cyclones.

Modern DWRs have in-built algorithms to obtain a variety of derived products from the three main products such as the probability of severe weather, hail, turbulence and icing, and tropical cyclone eye parameters such as the magnitude and radius of maximum winds. There are tracking algorithms for the detection of development and movement of thunderstorms and tropical cyclones. Therefore, from a user perspective the performance of the indigenous DWR as compared to the off-the-shelf systems of specialised manufacturers such as Gematronik, will be judged in terms of hardware as well as software. Interestingly, the IMD does not seem to have evolved its own test plan for the imported system so that all the systems could be evaluated against a common benchmark. Besides, the Test Plan Document provided by Gematronik was apparently misplaced by the IMD during transit from Germany.

The critical elements in the hardware include the stability of the pedestal on which the huge and heavy 9-metre radar dish is mounted atop a building so that there is no jitter, the radar has a good pointing accuracy and the base does not yield under the large reactionary torque force when the radar comes to a halt; low transmission loss in the radome (in which the radar is housed to prevent damage in severe weather conditions), frequency stability of the transmitter and the permissible pulse forms and PRFs and reliability of the signal processor. With regard to software, it will be in terms of the range of in-built algorithms and the associated range of useful derived products that the system is able to provide.

The indigenous DWR consists of a high-power (one MW) coherent klystron-based transmitter with a pencil beam of 0.95 degree width and a low noise receiver with high dynamic range. The emitted signal shape has very low side lobes. The entire radar is controlled by a single operator using a personal computer-based radar controller. The system is also designed to support the transmission and reception of signals in two orthogonal polarisations. Polarimetric applications enable the determination of the drop size distribution and identification of particle phase and removes uncertainties in rainfall estimates. However, such systems are about to enter operational mode in the U.S. and Europe only now. Hence it is proposed to have polarisation capability only as a retrofit in the indigenous DWR.

The other key features of the indigenous DWR include a self-supporting, truncated spherical radome of 12.8 m diameter, with curved sandwich panels. The radome can withstand steady winds of 250 kmph gusting up to 300 kmph and RF power loss through it is stated to be a low 0.2 dB. Its antenna is a prime-focus 9 m parabolic dish made of special-grade aluminium alloy. It has the continuous rotation capability in the azimuth up to 3 rpm and -2 to +90 in elevation, at selectable rates.

A host of organisation and companies were involved in the DWR developmental effort. The National Aerospace Laboratory (NAL), Bangalore, of the Council of Scientific and Industrial Research (CSIR) designed and built the radome. The antenna structure was designed by the Indian Institute of Technology, Madras, and fabricated by Sree Rama Engineering, Hyderabad. The signal feed was designed by the Electronics Research and Development Establishment (LRDE), Bangalore, of the (Defence Research and Development Organisation(DRDO). The transmitter was built by Society for Applied Microwave Electronics and Engineering Research (SAMEER) of the Department of Information Technology, Mumbai, and the receiver by Karnataka TeleElectronics Engineering Ltd (KATEEL), Bangalore. The digital signal processor and data products were developed by ISRO's radar cell and the system integration was done by BEL.

The radar's internal calibration has been completed and it is currently undergoing sun calibration (for pointing accuracy) and spherical calibration (using a hollow balloon) to enable measurement of the tangential component of velocities at SHAR. These calibrations are yet to be done for the imported systems as well so that there is a common benchmark in a network of DWRs, says Viswanathan. The first precipitating system that the system was used to measure was the rainfall in Bangalore on March 12, 2001. The PPI for reflectivity obtained recently on June 18 at SHAR shows the excellent quality of the radar's primary data products.

Given this commendable achievement, it is to be hoped that a nationwaide network of DWRs based on indigenous design will materialise soon. Besides cyclone detection and monitoring, a nationwide DWR network would be able to provide more reliable countrywide rainfall data as well as enable hydrological studies relating to flash flood warning, catchment area assessment, predicting and characterising thunder storms including hailstorms. While there is only a go/no-go option for imported systems, indigenous DWR can be improved and upgraded based on its operational experience as well as technological evolution during its design life of 20 years.

The Services too have apparently expressed interest in C-band Doppler radars for their use. A large order from the Services would give a boost to the radar-building capability of BEL, which for some reason did not build on the IMD purchase of 27 (S- and X-band) radars for the existing network and emerge as a major player in the radar market. Indeed, the IMD has had to import radars subsequently and also to abandon some radars for want of maintenance and spares. Viswanathan suggests that a special cadre of radar technicians be trained to maintain the future DWR network across the country.

A letter from the Editor

Dear reader,

The COVID-19-induced lockdown and the absolute necessity for human beings to maintain a physical distance from one another in order to contain the pandemic has changed our lives in unimaginable ways. The print medium all over the world is no exception.

As the distribution of printed copies is unlikely to resume any time soon, Frontline will come to you only through the digital platform until the return of normality. The resources needed to keep up the good work that Frontline has been doing for the past 35 years and more are immense. It is a long journey indeed. Readers who have been part of this journey are our source of strength.

Subscribing to the online edition, I am confident, will make it mutually beneficial.


R. Vijaya Sankar

Editor, Frontline

Support Quality Journalism
This article is closed for comments.
Please Email the Editor