DECEMBER 2017 M OHANAKUMAR ET AL. 2659

Technical Details of a Novel Profiler at 205MHz

a a b a c K. MOHANAKUMAR, AJIL KOTTAYIL, V. K. ANANDAN, TITU SAMSON, LINTO THOMAS, a a a a K. SATHEESAN, REJOY REBELLO, M. G. MANOJ, RAKESH VARADARAJAN, a d d K. R. SANTOSH, P. MOHANAN, AND K. VASUDEVAN a Advanced Centre for Atmospheric Radar Research, Cochin University of Science and Technology, Cochin, India b Telemetry, Tracking and Command Network (ISTRAC), Indian Space Research Organisation, Bangalore, India c Data Patterns India Pvt. Ltd., Chennai, India d Department of Electronics, Cochin University of Science and Technology, Cochin, India

(Manuscript received 18 March 2017, in final form 25 October 2017)

ABSTRACT

The Cochin University of Science and Technology (CUSAT), Cochin, India, hosts the world’s first 205-MHz stratosphere–troposphere (ST) wind profiler radar. This radar constitutes 619 three-element Yagi–Uda antennas with a power aperture product of 1.6 3 108 Wm2 and is capable of providing accurate three-dimensional wind profiles for an altitude range of 315 m–20 km. The system description and its first time validation and results from some of the radar’s potential applications are being presented. The radar wind profiles have been validated against collocated GPS– measurements during the summer monsoon of 2016. The radar and ra- diosonde profiles show very good correlation with coefficients of 0.99 and 0.93 for zonal and meridional , respectively. The standard deviation of the radar measurements with respect to radiosonde measurements is 2 2 found to be 1.85 m s 1 for zonal wind and 1.66 m s 1 for meridional wind. Moreover, the radar also detects echoes from the ionosphere. The ST radar at Cochin (10.048N, 76.338E; 40 m MSL) is an ideal observational facility, located in the tropics, for understanding the processes of the Indian summer monsoon at the region of its onset, which is expected to enhance science’s knowledge of monsoon dynamics.

1. Introduction Bragg scattered signals, which necessitate that the dominant Fourier component of the eddies be half the The ability of to profile wind was first demon- wavelength of radar (Briggs and Vincent 1973; Balsley strated in the late 1960s and since then radars have and Gage 1980; Gage and Balsley 1984). been extensively used in regional weather prediction A retrospective of radar profiler history shows that models, studying the evolution of convective events and the first profilers were developed around very high gravity waves, monitoring turbulence, and forecasting frequencies (VHF) of 40–55 MHz (Balsley and Gage rainstorms, among many other applications (Sato et al. 1980). They provided their best results above 4 km 1995; Satheesan and Krishna Murthy 2002; Satheesan and were extensively used in stratospheric–mesospheric and Krishna Murthy 2004, 2005; Hooper et al. 2005; studies (Woodman 1977; Röttger et al. 1979). Radars in Campos et al. 2007b; Kirkwood et al. 2010; Réchou et al. ultra high frequencies (UHF) were thereafter devel- 2013; Simonin et al. 2014). They remain the best means oped to resolve the lower height coverage and vertical to access three-dimensional wind profiles continuously resolution as well as for boundary layer studies. Cur- at high temporal and spatial resolutions. Small-scale rently, wind profiler radars operating the world over turbulent eddies formed as a result of mixing of air of span a frequency range of 30–300 MHz in the VHF band different densities can cause changes in the refractive and 300–1200 MHz in the UHF band. Many of these index of the atmosphere. Wind profiling radars detect the radars are operated as networks by the United States, backscattered signals resulting from refractive index Europe, and Japan. The European wind profiler net- variations, and the corresponding Doppler shift is used work [Coordinated WIND profiler network in Europe to calculate air velocities. Wind profilers detect only the (CWINDE)] provides operational support through its E-PROFILE hub, comprising several radars operating Corresponding author: K. Mohanakumar, [email protected] in frequencies around 50 MHz, 400 MHz, and 1 GHz

DOI: 10.1175/JTECH-D-17-0051.1 Ó 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses). Unauthenticated | Downloaded 09/29/21 11:02 PM UTC 2660 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 34

(Nash and Oakley 2001). The Japan Meteorological lies with the Arabian Sea on its western side and the Agency (JMA) has been operating the Wind Profiler Western Ghats on its eastern side, and this geographical Network and Data Acquisition System (WINDAS) uniqueness causes the first burst of the summer mon- since 2001 and is the first network to be operating soon to occur over this region. Cochin is therefore an 1.3-GHz radars (Tabata et al. 2011). ideal location for undertaking studies on the dynamics of In a tropical country like India, it has been felt that the the Indian summer monsoon. upper-air observations of wind are inadequate for ac- With the main rationale of studying the dynamics of curate weather modeling or for prediction of local the summer monsoon, a stratosphere–troposphere (ST) weather phenomena. Moreover, these observations are wind profiling radar at 205 MHz has been set up at Co- mostly from radiosonde networks whose spatial distri- chin (10.048N, 76.338E). This makes the radar the first of bution is far too coarse. With the aim of having high- its kind to be operating at 205 MHz in India or, for that quality, continuous wind data, it was proposed that a matter, in the world. The 205-MHz frequency has been radar wind profiler network be set up following the ex- particularly chosen over 50 or 400 MHz owing to specific ample of many other countries (Atlas 1990). The World reasons. One major motive is to study the characteristics Meteorological Organization (WMO) has also empha- of the monsoon low-level jet, an important component sized the need to adopt this technology as a step to im- driving the summer monsoon (Joseph and Sijikumar prove monitoring and for better forecasting of Earth’s 2004; Narayanan et al. 2016), the core of which lies at atmospheric processes. about 850 hPa (approximately 1.7 km). Another objective The extent of applications to which radars can cater is to study the stratosphere–troposphere exchange pro- primarily depends on the frequency band on which they cesses occurring during the monsoon season. Monsoon are operated. The mesosphere–stratosphere–troposphere low-level jet studies with 50-MHz radars are practically (MST) radar operating at 53 MHz in Gadanki since 1991 difficult on account of their limited lower height coverage, is India’s first wind profiler radar (Rao et al. 1995). and though lower heights can be covered using 400-MHz Though ideal for high-altitude studies, the large frequencies; their upper height coverage is limited to array of MST radars limits probing the lower atmo- approximately 13 km. Therefore, studying stratosphere– sphere below 3 km. One of the major reasons is that the troposphere exchange processes occurring between large aperture array of the radar operating around 17- and 20-km altitudes during the monsoon season can- 50 MHz is unable to form a well-defined beam in the not be studied using 400-MHz radars. first few kilometers above the surface. A smaller radar Galactic or cosmic noise is an essential criterion while at 1280 MHz, called the lower atmospheric wind pro- determining the accuracy of radar measurements oper- filer (LAWP), was subsequently set up at Gadanki ating in the frequency range of 50–1000 MHz (Doviak (Srinivasulu et al. 2012) for probing the troposphere and Zrnic´ 2014). The effect of cosmic noise in very from 100 m to 4–7 km. However, the effectiveness of high-frequencies scales as 22.5th power of frequency LAWP is limited, owing to its sensitivity to rain, clouds, (Campos et al. 2007b). Therefore, the 205-MHz radar is etc. A 404-MHz radar has also been operating at the less affected by cosmic noise when compared to the India Meteorological Department (IMD), Pune, since 50-MHz radar. Over the equatorial region, the galactic 2003 (Pant et al. 2005) and is capable of covering the noise of the 50-MHz radar is found to be around 6000 K troposphere from 300 m to 13 km. The higher height (Kirkwood et al. 2010); however, it has been shown in coverage above the troposphere of the radars operating Turtle and Baldwin (1962) that for frequencies nearer to at UHF range is mainly limited by the constraint im- 200 MHz, the values of cosmic noise lie within a range of posed by the inner scale of turbulence (Balsley and 140–300 K over much of the sky with a peak value of Gage 1982; Hocking 1985). 1000 K in the galactic plane. Furthermore, rainy condi- The most dominant weather phenomenon over the tions are found to saturate the radar measurements at Indian subcontinent influencing almost every sphere of UHF ranges; however, 205-MHz radars are less affected life, including the livelihood and economy of its people, by such issues. Over the region of Cochin, the 205-MHz is indisputably the Indian summer monsoon. It is prob- radar is better suited with distinct advantages over the ably the most studied weather phenomenon in the In- 50- or 400-MHz radars. The 205-MHz radar is thus the dian context. Nevertheless, the research so far has best compromise between the 50- and 400-MHz radars, unraveled far less than that needed for an in-depth un- because it is able to scan both the lower and higher derstanding of this complex process. Cochin is a coastal heights and is thus capable of addressing our major region situated in the southwestern part of the Indian scientific objectives. continent and can be best described as the gateway of Through this paper we introduce the complete system the Indian summer monsoon. The mainland of Cochin description of the 205-MHz radar and its wind validation

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC DECEMBER 2017 M OHANAKUMAR ET AL. 2661 for the altitude range of 315 m–20 km. We also present synthesizer unit produces signals at 840 and 80 MHz some of the preliminary scientific results achieved and are connected to the analog-to-digital converter through this radar. Prior to the realization of the full- (ADC) at the receiving end and the digital-to-analog fledged 205-MHz radar with 619 antennas, a prototype converter (DAC) at the transmitting end. Two DACs radar with 49 antenna elements had been tested. The are available, one for transmitting the signal and prototype was used to assess the feasibility of the the other for generating the simulated signal; the latter 205-MHz frequency for wind profiling, the details of is intended mainly for power, range, and Doppler which can be found in Kottayil et al. (2016) and Samson calibration. et al. (2016). The 205-MHz signal, having the strength of 23dBm The paper is structured as follows: Section 2 gives the generated out of the DAC, is routed through the technical details of the 205-MHz wind profiler radar, switching and synthesizer unit to a 1:13 combiner/split- section 3 presents the radar validation and some of its ter, which is then routed to a 1:7 combiner/splitter and scientific applications, and section 4 presents the sum- then to a 1:8 combiner/splitter. In this way, the radio mary and conclusions. frequency (RF) signal gets distributed to all 619 trans- mitter and receiver modules (TRM). Through different 2. Technical details of ST radar stages of amplification in the RF signal en route and within TRM, the transmitted signal reaching each an- a. Basic block diagram of radar tenna has the strength of 57 dBm. The block diagram of the 619 element wind profiler During receiving-only (Rx) mode, power from the en- radar is shown in Fig. 1. Details of the profiler are given tire 619 antennas is combined and the output of the 1:13 in Table 1 and described below. The ST radar installed at combiner/splitter is fed into the ADC. The received power CUSAT is an active phased-array radar consisting of 619 of the signal will be on the order of 2120 to 2165 dBm. three-element Yagi–Uda antennas arranged in an Out of the four ADCs, three are designated for SAM equilateral triangular grid with an interelement spacing mode and one for DBS mode. In Rx mode, the maxi- of 0.7l, where l is the operating wavelength, which is mum power input that TRM can receive is 235 dBm 1.43 m. The gain of each antenna is 7.5 dBi with a voltage and within TRM, the received signal is amplified to standing wave ratio (VSWR) of 1.2:1. The power aper- 40 dBm. The input/output control and the communica- ture product is 1.6 3 108 Wm2 with an approximate one- tion system generate signals for communicating with way half-power beamwidth of 3.28. The peak and the TRM with the purpose of monitoring the health status of mean power of the radar are 309 and 46 kW, respec- each TRM, such as its operating temperature, VSWR tively. The 619 antennas cover an area of 572 m2 with an errors, etc. effective aperture area of 536 m2. b. TRM The 619 antennas are divided into 13 clusters, 7 of which are in the center and 6 are in the periphery. The TRM forms the most important component of the seven main central clusters shown in color in Fig. 1 have wind profiler radar. It is a 500-W transmitter and re- 49 antennas each, and the peripheral clusters have 46 ceiver module operating at a central frequency of antennas each. The ST radar at Cochin is configured to 205 MHz with a bandwidth of 5 MHz. It amplifies and operate under two modes of operation, the Doppler transmits RF signals toward the target, and during re- beam swinging (DBS) mode and the spaced antenna ceiving mode the reflected echoes received at the an- method (SAM) (e.g., Cohn et al. 2001). In DBS mode, tenna are strengthened. The TRM module comprises the radar echoes received from zenith and off-zenith (i) a 500-W VHF transmitter module, (ii) a VHF re- beams are used for deriving the three-dimensional wind. ceiver module, and (iii) a transmit–receive (TR) con- In SAM mode, the central cluster will radiate energy and troller and a power supply module. the reflected energy will be received by any three clus- The block diagram of the TRM is also shown in Fig. 1. ters (shown in orange/blue in Fig. 1) surrounding the A single-pole double-throw (SPDT) switch controls central cluster. Though a provision for SAM mode is the mode of operation of TRM and depending on this, included, this is still in testing mode and therefore the the TRM works as either a transmitter or a receiver. The discussion in the paper will be oriented in DBS mode of TRM has a directional coupler and an ADC that regu- operation. late the forward and reverse power flow during The main power supply that is routed through an TRM’s transmission mode. It has an external transmit uninterruptible power supply (UPS) gets converted monitoring port, for monitoring the transmit power into dc at 112 and 124 V. A master oscillator gener- during transmission mode. The VHF transmitter mod- ates 10-MHz signals, which upon feeding to the clock ule transmits a minimum of 500-W peak power in the

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC 2662 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 34

FIG. 1. (top) Basic block diagram of the 205-MHz ST radar installed at CUSAT. (bottom) Basic block diagram of the TRM.

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC DECEMBER 2017 M OHANAKUMAR ET AL. 2663

TABLE 1. The 205-MHz ST wind profiler specifications. RS422-level signals. The power supply section generates all the supply voltages for the active devices used in the Parameters Specification TRM. It provides the control signals for TR switches, Frequency 205 MHz digital attenuators, and phase shifters. The TR switch Bandwidth 5 MHz used in TRM is an active device that works based on Type of system Active with TRM Antenna element Three-element Yagi–Uda antenna transmit–receive pulse (TRP) and can switch between Height coverage 315 m–20 km transmitter mode and receiver mode. The TRP is gen- Range gates 1024 (programmable) erated by an input/output (I/O) control and communi- Modes of operation DBS/SAM cation section, and is fed to TRM through a digital ; Height resolution 45 m–6 km, and 150 or distribution card. 300 m–20 km One-way half-power ;3.28 The field-programmable gate array (FPGA) in the beamwidth module monitors the occurrence of any abrupt change in Off-zenith angle Selectable from 08 to 308 the power supply. Excess duty and pulse width of the TR in steps of 18 cover pulse, the forward and reverse power (ensuring no 8 8 8 Azimuth angle 0 –360 with 1 resolution open path during TRM transmission), and the temper- Pulse width 0.3–76.8 ms Modulation Binary phase shift keying ature status of the TRM are also regulated by this (BPSK) coded compression module. More details on the TRM can be found in Code Complementary code/Barker code Samson et al. (2016). Baud 0.3–4.8 ms in steps of 0.3 ms The TRM can be configured into five different modes, Pulse repetition 100 Hz–16 kHz, selectable namely, neither transmission nor reception (idle) mode, frequency (PRF) TRM transmit peak power 500 W transmission-only (Tx) mode, reception-only (Rx) Duty cycle Up to 15% (max) mode, transmission and reception (Tx–Rx) modes, and Peak power ;1.6 3 108 Wm2 continuous reception (Rx all modes). The TRM of the aperture product 205-MHz ST radar is shown in Fig. 2. Radar system sensitivity 2165 dBm Dynamic range 70 dB (min) c. Radar controller and data processing Master reference oscillator Rubidium oscillator Type of receiver Direct bandwidth sampling During signal transmission, pulse compression is Type of signal processor FFT-based frequency domain used to increase the signal-to-noise ratio or the signal Gain of whole array 35 dBi processing gain. Pulse compression is implemented , Receiver noise figure 3dB through phase coding via either Barker codes or com- plementary codes. The digital receiver section consists of two 16-bit ADCs with an 80-MHz sampling rate. frequency range of 202.5–207.5 MHz. The maximum The 205-MHz signal is sampled at the rate of 80 MHz duty cycle is 15% with an average power output of 75 W. and as a result the 205-MHz signal will fall under The transmitter section has a bandpass filter (BPF) with 35 MHz as an undersampled signal. The signals then 202-MHz central frequency and 8-MHz bandwidth to pass through a cascaded integrator comb (CIC) filter broadcast the required band of frequencies. having a pass band of 62 MHz and a stop band of The input signal is provided with a gain of 40 6 2dB 63.5 MHz. Digital down conversion (DDC) is further and a noise figure within 3.5 dB by the VHF receiver performed on the undersampled signal at 35 MHz to section. This also bears a similar BPF as in the trans- extract the Doppler-shifted frequency in the received mitter section to restrict the reception of out-of-band signal. Thereafter, pulse decoding is performed on the signals. The low-noise amplifier (LNA) in this module complex time domain signal (in-phase and quadrature- strengthens the received echoes from the target with a phase signals), which is done through cross-correlating noise figure of 0.5 dB. Beam forming and steering are the incoming digital signal with a replica of the trans- achieved through the digital attenuator (6 bits) and mitted code. This has been implemented by means of a delay-line-based phase shifters. The LNA also has an correlator/transversal filter. The pulse decoding is re- external receiver injection port for functional validation peated for n number of pulses for each range gate of the receiver chain. separately and then added coherently to improve the The VHF TR controller and power supply module signal-to-noise ratio. This process is continued until generate programmable gate bias voltage for the driver the required number of fast Fourier transform (FFT) and power amplifier during the transmission-only mode points is archived. of operation. The TRM is fed with an external supply of Prior to applying FFT on the time domain signal, 48-V DC, while the digital interface is through external windowing is applied on it to minimize the edge effects

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC 2664 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 34

the mirror image formation in the orthogonal beams, namely, east and west in the zonal plane and north and south in the meridional plane. Figure 3a corresponds to observations from the zenith direction. Figures 3b,c corresponds to east and west directions, respectively, when the radar is steered 108 from zenith. A clear mirror image formation is observed, which clearly in- dicates the orthogonal beam formation. Figures 3d,e correspond to north and south beam observations, re- spectively. Here, too, the mirror image formation is clearly visible. Relatively larger standard deviations are observed in the north and south beam observations. This may be due to the influence of vertical winds as seen in Fig. 3a. Both coded and uncoded modes of operation can be performed with the ST radar. From 1 to 64 bits can be assigned for code length in steps of 2n. Baud can be se- lected within a range of 0.3–4.8 ms in steps of 0.3 ms. The radar has a provision to do coherent and incoherent FIG. 2. (top) Three-element Yagi–Uda antennas and (bottom) integration to improve the signal-to-noise ratio (SNR) TRM of the ST radar at CUSAT. of the signal. The number of coherent integrations can be set anywhere between 1 and 1024, whereas for in- coherent integration, the range is from 1 to 128. Elec- that result in spectral leakage in the FFT spectrum. tronic beam steering is enabled in the ST radar, where This will further smooth the waveform end point so as the off-zenith angle can be configured between 08 and to get better spectral resolution. The ST radar has a 308 in increments of 18 and the azimuth angle can be set provision of applying various windowing functions, within the 08–3608 range with an angular resolution of 18. such as Hanning window, Hamming window, Black- Thus, this beam steering configuration supports the man window, Blackman–Harris window, Bartlett win- three-dimensional study of atmospheric features. dow, rectangular window, etc. After windowing, FFT is A graphical user interface (GUI) in the radar control appliedonthetimedomainsignaltoextractthe computer provides the means to set up operational pa- Doppler power spectrum. The removal of clutter from rameters, such as the pulse repetition period (PRP), the the Doppler power spectrum is done by eliminating complementary code words, the number of coherent points lying around the zero Doppler in the frequency integrations (NCI), the number of FFT (NFFT) points, domain and replacing them with values interpolated the number of in-coherent integrations (NICI), the ob- from the adjacent values (Barth et al. 1994). The three servation altitude range, etc. moments of the Doppler power spectrum—the zeroth d. Radar beamwidth determination moment representing the total signal and noise power, the first moment signifying the mean Doppler, and the Virgo A is one of the brightest radio sources in the sky second moment representing the Doppler variance— and is thus one of the key targets for radio astronomical have been computed following Woodman (1985) and studies. It is a supergiant elliptical galaxy within the Barth et al. (1994). Other refined methods for moment Virgo constellation. Being a bright radio–astronomical estimation have been documented elsewhere in the source, Virgo A is widely used for calibrating remote literature (Hocking 1997; Wilfong et al. 1999). The sensing instruments. With regard to wind profiler radar, signal processing algorithm employed in the 205-MHz the Virgo transit is used for estimating radar beamwidth radar is same as that described in Anandan et al. and gain, its pointing accuracy, etc. (2001), and reader may refer to the same for more The 205-MHz radar was configured in receiver mode details. for tracking Virgo A during 1000–1100 LT 15 October Figure 3 shows the mean Doppler and standard de- 2016. The radar beam was tilted at 28 from the zenith viation estimated from the radar observation on a clear direction and was pointed toward north to detect radi- day for a period of 35 min on 22 August 2016 for five ation from Virgo A. The result of the radiation received different beam-pointing directions. One of the criteria from Virgo A is shown in Fig. 4. The received radiation for evaluating the accuracy of beam formation is to test increases as Virgo A approaches the radar beam

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC DECEMBER 2017 M OHANAKUMAR ET AL. 2665

FIG. 3. Mean Doppler velocity and standard deviation (horizontal lines) estimated for five different beam positions on 22 Aug 2016. pointing direction and decreases as it recedes. This a. Radar wind validation measurement can be used to infer the radiation pattern The radar wind profiles are validated against collo- of the radar. The estimated null–null width from the cated radiosonde profiles launched from the site of the figure is approximately 5.078 and the one-way half- radar location. Dedicated Graw radiosonde observa- power beamwidth is around 3.368. The one-way half- tions taken during the southwest monsoon period power beamwidth estimated from the Virgo A transit is (June–September) of 2016 have been used in radar wind very close to the simulated result of 3.208. validation. The radar wind measurements from two 3. Potential scientific applications different coded modes of operation at baud rates of 0.3 and 1.2 ms have been combined over the altitude range The following discussion focuses on the results of of 315 m–20 km. A height range of 315 m–5 km has been validation of the 205 MHz 619 element wind profiler taken from the 0.3-ms baud, which has a vertical reso- radar. This discussion also incorporates some of the re- lution of 45 m. Similarly, 5–20 km has been taken from sults from few of its applications like signature of rain the 1.2-ms baud, which has a vertical resolution of 180 m. from radar measurements, detection of ionospheric Scatterplots (Fig. 5) show the comparison between echoes, etc. collocated radiosonde and radar zonal and meridional

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC 2666 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 34

FIG. 4. (left) Radiation pattern of the 205-MHz radar as determined from the Virgo A transit. (right) Simulated radiation pattern of the 205-MHz ST radar. The power (dB) is uncalibrated winds. In Fig. 5, measurements from 315 m to 20 km are and 0.93, respectively. We have used a standard Pearson included. Simultaneous collocations between radar and regression, in which we have assumed that the radio- radiosonde are made, based on spatial and temporal sonde data have zero error: this is of course untrue, and constraints. This is due to radar profiles being generated it is to be noted that a better comparison could be made every 4 min. The closest height and time between the by assuming both the radiosonde and profiler winds radar and the radiosonde is chosen for making collo- have errors. However, the comparison is sufficient to cated pairs. The maximum difference in time and height demonstrate the accuracy of the radar. More details between collocated datasets is 18 min and 32 m, re- about the dual-error comparisons are provided in Belu spectively. More information on collocations can be et al. (2001) and Kottayil et al. (2016). found in Kottayil et al. (2016). The radar’s horizontal Figure 6 shows the radiosonde and radar wind speed wind components show excellent agreement with the and direction profiles observed on 22 August 2016. On 2 radiosonde with a mean bias of 0.053 6 1.85 m s 1 in that particular day, the radar provided a traceable signal 2 zonal wind and 0.005 6 1.66 m s 1 in meridional wind. from 315 m to about 22 km and the comparison shows The Pearson correlation coefficient between radar and that the radar-derived wind speed and direction are in radiosonde measurements observed for zonal wind is good agreement with the radiosonde profiles. To give an 0.99, whereas it is 0.93 for meridional wind, both of error estimate of radar-derived horizontal wind speed which are significant at the 0.01 level. The linear re- and at different altitudes, the percentage gression slopes for zonal and meridional winds are 0.97 error in horizontal wind speed and direction relative to

FIG. 5. Scatterplots of radar vs radiosonde zonal and meridional winds. The case where the abscissa and ordinate are identical [i.e., it has a slope of unity and passes through (0, 0); black line] is indicated.

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC DECEMBER 2017 M OHANAKUMAR ET AL. 2667

FIG. 6. Comparison between the collocated radiosonde and radar wind speed and wind direction on 22 Aug 2016. collocated radiosonde profiles is calculated and the re- deconvolution is performed, the RDSD and the fall sult is shown in Fig. 7. Overall, the results show that the speed can be determined using various approaches de- radar is capable of providing very accurate wind profiles scribed in the literature (Sato et al. 1990; Campos et al. for an altitude range of 315 m–20 km. The maximum 2007a; Anandan et al. 2010). error in horizontal wind speed is found to be 8% with a With the 205-MHz radar measurements, it is possible minimum of 4.5%. The upper and lower limits of error in to clearly demarcate the ambient air portion from the wind direction are found to be 5% and 11.5%, precipitation portion as shown in Fig. 8. In addition, the respectively. presence of a melting layer as a horizontal strip is clearly observed in the power spectra at about 4.5 km. This b. Signature of rain on radar signal shows the ability of this radar to discern different layers It is a known fact that VHF radars are sensitive to of rain-bearing clouds. Rayleigh scattering during precipitation (Fukao et al. c. Detection of ionospheric echoes 1985) and this can be used for retrieval of microphysical properties of precipitating clouds, such as raindrop size The site of the radar location lies in close proximity to distributions (RDSDs) and the fall speed of a pre- the geomagnetic equator (1.728N, 149.678E). The an- cipitation particle, etc. Figure 8 shows the Doppler tennas of the 205-MHz radar are oriented in the geo- power spectrum observed during a rainy day. Two dis- magnetic north–south direction, thus making it possible tinct power spectra can be clearly identified from the to have ionospheric observations. The radar does detect figure; one centered around zero Doppler represents the the ionospheric echoes and is shown in Fig. 9. The ion- 2 ambient air echoes and other lying around 8 m s 1 rep- ospheric F layer has been detected during the nighttime resents the precipitation echoes. The observed Doppler of 17 October 2016, whereas the D layer has been de- 2 power spectrum around 8 m s 1 is the convolution of tected during the daytime of 14 November 2016. To Bragg scattering from ambient air and the Rayleigh detect ionospheric echoes, the radar beam is tilted to 78 scattering from precipitation (Martin Ralph 1995). off zenith to the north direction so that the radar beam is The most important step for retrieving RDSD perpendicular to the magnetic field at the ionospheric F or raindrop fall velocity from VHF radar measure- and D regions. Coded modes of operation were em- ments is the removal of air vertical motion from the ployed with a 1.8-ms baud with 32-bit code length for the precipitation-affected Doppler power spectrum. Once detection of the F region, while for the D region a 0.3-ms

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC 2668 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 34

FIG. 7. The percentage difference between the radiosonde and radar wind speed and wind direction. The wind speed and direction are binned at 1-km-height intervals, and the per- centage differences are calculated. The bar represents the meanpffiffiffiffi difference (%), and the ver- tical line represents the standard error, which is defined as s/ N, where s refers to the standard deviation and N is the number of data points at each bin.

baud with 64-bit code length has been employed. In the features of monsoon circulation over several years the future, this radar is expected to provide valuable could yield better predictors for forecasting monsoon information for understanding the processes in the onset. Moreover, the electronic beam steering capability ionosphere. of this radar can be used to study the dynamics of the atmosphere, such as deriving horizontal wind gradients, 4. Summary and conclusions divergence, etc. As mentioned in the introductory

A novel stratosphere–troposphere wind profiler radar at 205 MHz, the first of its kind in the world, has been set up in Cochin for measuring the wind profiles in the al- titude range of 315 m–20 km. This radar consists of 619 three-element Yagi–Uda antennas with a peak power aperture product of 1.6 3 108 Wm2. Through this paper we put forth the radar’s technical description, its vali- dation, and most importantly we highlight its multifari- ous scientific applications by presenting a few of the results. The radar wind measurements show excellent agreement with the collocated radiosonde measure- ments with a correlation of 0.99 for zonal wind and 0.93 for meridional wind. The accuracy of the radar for 2 measuring zonal wind is found to be 1.85 m s 1 in zonal 2 wind and 1.66 m s 1 in meridional wind. The setting up of this radar in Cochin, the region witnessing the onset of southwest monsoon, has huge significance in the meteorological context. The main purpose of installing this radar in Cochin is to study the FIG. 8. The Doppler power spectrum of the zenith beam ob- dynamics of the Indian summer monsoon. Several sci- served during the rainy day of 18 May 2017. The power (dB) is entific outcomes are expected from this facility in the uncalibrated. The mean Doppler (blue) and the Doppler width for future. It is anticipated that continuous monitoring of rain echoes (black) are indicated.

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC DECEMBER 2017 M OHANAKUMAR ET AL. 2669

FIG. 9. The (left) F layer and (right) D layer as observed by the 205-MHz wind profiler radar on 17Oct 2016 and 5 Jun 2017, respectively. The power (dB) is uncalibrated. section, the unique location of the radar in Cochin, a Acknowledgments. We deeply acknowledge and region nestled between the Arabian Sea and the West- thank the Science and Engineering Research Board ern Ghats, helps in the study of gravity waves originat- (SERB), Department of Science and Technology (DST), ing in the mountains and tropical cyclones. Since the government of India, for providing Grant SR/S4/AS:45/ radar provides observations from lower-stratospheric 2009 and their assistance in the design and conception of regions, it is possible to study stratosphere–troposphere the ST radar facility at CUSAT. We also thank M/S Data exchange processes occurring dominantly during deep Pattern India Pvt. Ltd., Chennai, India, for developing convective processes. and installing the radar. The high-vertical-resolution measurements (45 m) from this radar aid in improving the study of turbulence REFERENCES in the sense that turbulent layers can be better resolved and more processes can be understood. One of the no- Anandan, V. K., G. Ramachandra Reddy, and P. B. Rao, 2001: table advantages of this radar is its ability to discern rain Spectral analysis of atmospheric radar signal using higher order spectral estimation technique. IEEE Trans. Geosci. echoes from background wind, which is vital for the Remote Sens., 39, 1890–1895, https://doi.org/10.1109/36.951079. study of microphysical properties of tropical rain- ——, C. J. Pan, K. Krishna Reddy, T. N. Rao, and S. V. Bhaskara Rao, bearing clouds in the coming years. A better illustra- 2010: Observation of precipitation and drop-size distribution as- tion of the evolution of tropical thunderstorms can be sociated with a typhoon using VHF radar. Open Atmos. Sci. J., 4, had using this radar. Besides these, there is also a pos- 114–125, https://doi.org/10.2174/1874282301004010114. Atlas, D., Ed., 1990: Radar in : Battan Memorial sibility of detecting the signature of cloud layers in radar and 40th Anniversary Radar Meteorology Conference. Amer. echoes, research on which is in progress. The radar is Meteor. Soc., 806 pp., https://doi.org/10.1007/978-1-935704-15-7. designed to operate for 25–30 years, and significant Balsley, B. B., and K. S. Gage, 1980: The MST radar technique: contributions can be made toward climate change Potential for middle atmospheric studies. Pure Appl. studies. In a nutshell observations from this radar can Geophys., 118, 452–493, https://doi.org/10.1007/BF01586464. catalyze studies on monsoon dynamics, stratosphere– ——, and ——, 1982: On the use of radars for operational wind profiling. Bull.Amer.Meteor.Soc., 63, 1009–1018, https://doi. troposphere exchange processes, and turbulence, and org/10.1175/1520-0477(1982)063,1009:OTUORF.2.0.CO;2. help to improve the regional weather prediction capa- Barth, M. F., R. B. Chadwick, and D. W. van de Kamp, 1994: bility in the near future. Data processing algorithms used by NOAA’s wind profiler

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC 2670 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 34

demonstration network. Ann. Geophys., 12, 518–528, https:// Narayanan, S., A. Kottayil, and K. Mohanakumar, 2016: Monsoon doi.org/10.1007/s00585-994-0518-1. low-level jet over the gateway of Indian summer monsoon: Belu, R. G., W. K. Hocking, N. Donaldson, and T. Thayaparan, A comparative study for two distinct monsoon years. 2001: Comparisons of CLOVAR windprofiler horizontal Meteor. Atmos. Phys., 128, 689–696, https://doi.org/10.1007/ winds with and CMC regional analyses. s00703-016-0459-8. Atmos.–Ocean, 39, 107–126, https://doi.org/10.1080/ Nash, J., and T. J. Oakley, 2001: Development of COST 76 wind 07055900.2001.9649669. profiler network in Europe. Phys. Chem. Earth, 26B, 193–199, Briggs, B. H., and R. A. Vincent, 1973: Some theoretical consid- https://doi.org/10.1016/S1464-1909(00)00239-2. erations on remote probing of weakly scattering irregularities. Pant, G. B., and Coauthors, 2005: Wind profiler and radio acoustic Aust. J. Phys., 26, 805, https://doi.org/10.1071/PH730805. sounding system at IMD, Pune: Some preliminary results. Campos, E. F., F. Fabry, and W. Hocking, 2007a: Precipitation Curr. Sci., 88, 761–769. measurements using VHF wind profiler radars: Measuring Rao, P. B., A. R. Jain, P. Kishore, P. Balamuralidhar, S. H. rainfall and vertical air velocities using only observations Damle, and G. Viswanathan, 1995: Indian MST radar 1. Sys- with a VHF radar. Radio Sci., 42, RS3003, https://doi.org/ tem description and sample vector wind measurements in ST 10.1029/2006rs003540. mode. Radio Sci., 30, 1125–1138, https://doi.org/10.1029/ ——, W. Hocking, and F. Fabry, 2007b: Precipitation measurement 95RS00787. using VHF wind-profiler radars: A multifaceted approach to Réchou, A., J. Arnault, P. Dalin, and S. Kirkwood, 2013: Case calibrate radar antenna and receiver chain. Radio Sci., 42, study of stratospheric gravity waves of convective origin over RS4008, https://doi.org/10.1029/2006rs003508. Arctic Scandinavia—VHF radar observations and numerical Cohn, S. A., W. O. J. Brown, C. L. Martin, M. E. Susedik, G. D. modelling. Ann. Geophys., 31, 239–250, https://doi.org/ Maclean, and D. B. Parsons, 2001: Clear air boundary layer 10.5194/angeo-31-239-2013. spaced antenna wind measurement with the Multiple Antenna Röttger, J., P. K. Rastogi, and R. F. Woodman, 1979: High- Profiler (MAPR). Ann. Geophys., 19, 845–854, https://doi.org/ resolution VHF radar observations of turbulence structures 10.5194/angeo-19-845-2001. in the mesosphere. Geophys. Res. Lett., 6, 617–620, https:// Doviak, R. J., and D. S. Zrnic´, Eds., 2014: Doppler Radar and doi.org/10.1029/GL006i007p00617. Weather Observations. 2nd ed. Academic Press, 562 pp. Samson, T. K., and Coauthors, 2016: Technical aspects of 205 MHz Fukao, S., K. Wakasugi, T. Sato, S. Morimoto, T. Tsuda, VHF mini wind profiler radar for tropospheric probing. IEEE I. Hirota, I. Kimura, and S. Kato, 1985: Direct measurement of Geosci. Remote Sens. Lett., 13, 1027–1031, https://doi.org/ air and precipitation particle motion by very high frequency 10.1109/LGRS.2016.2561965. Doppler radar. Nature, 316, 712–714, https://doi.org/10.1038/ Satheesan, K., and B. V. Krishna Murthy, 2002: Turbulence 316712a0. parameters in the tropical troposphere and lower strato- Gage, K. S., and B. B. Balsley, 1984: MST radar studies of wind and sphere. J. Geophys. Res., 107, 4002, https://doi.org/10.1029/ turbulence in the middle atmosphere. J. Atmos. Terr. Phys., 2000JD000146. 46, 739–753, https://doi.org/10.1016/0021-9169(84)90055-2. ——, and ——, 2004: Estimation of turbulence parameters in the Hocking, W. K., 1985: Measurement of turbulent energy dissipa- lower atmosphere from MST radar observations. Quart. tion rates in the middle atmosphere by radar techniques: A J. Roy. Meteor. Soc., 130, 1235–1249, https://doi.org/10.1256/ review. Radio Sci., 20, 1403–1422, https://doi.org/10.1029/ qj.03.86. RS020i006p01403. ——, and ——, 2005: A study of tropical tropopause using MST ——, 1997: System design, signal-processing procedures, and pre- radar. Ann. Geophys., 23, 2441–2448, https://doi.org/10.5194/ liminary results for the Canadian (London, Ontario) VHF angeo-23-2441-2005. atmospheric radar. Radio Sci., 32, 687–706, https://doi.org/ Sato, K., H. Hashiguchi, and S. Fukao, 1995: Gravity waves and 10.1029/96RS03316. turbulence associated with cumulus convection observed with Hooper, D. A., A. J. McDonald, E. Pavelin, T. K. Carey-Smith, and the UHF/VHF clear-air Doppler radars. J. Geophys. Res., 100, C. L. Pascoe, 2005: The signature of mid-latitude convection 7111–7119, https://doi.org/10.1029/95JD00198. observed by VHF wind-profiling radar. Geophys. Res. Lett., Sato, T., H. Doji, H. Iwai, I. Kimura, S. Fukao, M. Yamamoto, 32, L04808, https://doi.org/10.1029/2004GL020401. T. Tsuda, and S. Kato, 1990: Computer processing for deriving Joseph, P. V., and S. Sijikumar, 2004: Intraseasonal variability of the drop-size distributions and vertical air velocities from VHF low-level jet stream of the Asian summer monsoon. J. Climate, Doppler radar spectra. Radio Sci., 25, 961–973, https://doi.org/ 17, 1449–1458, https://doi.org/10.1175/1520-0442(2004)017,1449: 10.1029/RS025i005p00961. IVOTLJ.2.0.CO;2. Simonin, D., S. P. Ballard, and Z. Li, 2014: Doppler radar radial Kirkwood, S., M. Mihalikova, T. N. Rao, and K. Satheesan, 2010: wind assimilation using an hourly cycling 3D-Var with a Turbulence associated with mountain waves over Northern 1.5 km resolution version of the Met Office Unified Model for Scandinavia—A case study using the ESRAD VHF radar and nowcasting. Quart. J. Roy. Meteor. Soc., 140, 2298–2314, the WRF mesoscale model. Atmos. Chem. Phys., 10, https://doi.org/10.1002/qj.2298. 3583–3599, https://doi.org/10.5194/acp-10-3583-2010. Srinivasulu, P., P. Yasodha, P. Kamaraj, T. N. Rao, A. Jayaraman, Kottayil, A., and Coauthors, 2016: Validation of 205 MHz wind S. N. Reddy, and S. Satyanarayana, 2012: 1280-MHz active profiler radar located at Cochin, India, using radiosonde wind array radar wind profiler for lower atmosphere: System de- measurements. Radio Sci., 51, 106–117, https://doi.org/ scription and data validation. J. Atmos. Oceanic Technol., 29, 10.1002/2015RS005836. 1455–1470, https://doi.org/10.1175/JTECH-D-12-00030.1. Martin Ralph, F., 1995: Using radar-measured radial vertical ve- Tabata, Y., H. Hashiguchi, M. K. Yamamoto, M. Yamamoto, locities to distinguish precipitation scattering from clear-air M. D. Yamanaka, S. Mori, F. Syamsudin, and T. Manik, scattering. J. Atmos. Oceanic Technol., 12, 257, https://doi.org/ 2011: Observational study on diurnal precipitation cycle 10.1175/1520-0426(1995)012,0257:URMRVV.2.0.CO;2. in equatorial Indonesia using 1.3-GHz wind profiling

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC DECEMBER 2017 M OHANAKUMAR ET AL. 2671

radar network and TRMM precipitation radar. J. Atmos. of radar wind profiler spectra. J. Atmos. Oceanic Technol., Sol.-Terr. Phys., 73, 1031–1042, https://doi.org/10.1016/ 16, 723, https://doi.org/10.1175/1520-0426(1999)016,0723: j.jastp.2010.10.003. OGORWP.2.0.CO;2. Turtle, A. J., and J. E. Baldwin, 1962: A survey of galactic radiation Woodman, R. F., 1977: Mesospheric winds at equatorial latitudes: at 178 Mc/s. Mon. Not. Roy. Astron. Soc., 124, 459–476, https:// A review on observational aspects. J. Atmos. Terr. Phys., 39, doi.org/10.1093/mnras/124.6.459. 941–958, https://doi.org/10.1016/0021-9169(77)90002-2. Wilfong, T. L., D. A. Merritt, R. J. Lataitis, B. L. Weber, ——, 1985: Spectral moment estimation in MST radars. Radio D. B. Wuertz, and R. G. Strauch, 1999: Optimal generation Sci., 20, 1185–1195, https://doi.org/10.1029/RS020i006p01185.

Unauthenticated | Downloaded 09/29/21 11:02 PM UTC