Radio Science

MAARSY: The new MST radar on Andøya—System description and first results



[1] The Middle Atmosphere Alomar Radar System (MAARSY) on the North-Norwegian island Andøya is a 53.5 MHz monostatic radar with an active phased array antenna consisting of 433 Yagi antennas. The 3-element Yagi antennas are arranged in an equilateral triangle grid forming a circular aperture of approximately 6300 m2. Each individual antenna is connected to its own transceiver with independent phase control and a scalable power output up to 2 kW. This arrangement provides a very high flexibility of beam forming and beam steering with a symmetric radar beam of a minimum beam width of 3.6° allowing classical beam swinging operation as well as experiments with simultaneous multiple beams and the use of interferometric applications for improved studies of the Arctic atmosphere from the troposphere up to the lower thermosphere with high spatio-temporal resolution. The installation of the antenna array was completed in August 2009. The radar control and data acquisition hardware as well as an initial expansion stage of 196 transceiver modules was installed in spring 2010 and upgraded to 343 transceiver modules in November 2010. The final extension to 433 transceiver modules has recently been completed in May 2011. Beside standard observations of tropospheric winds and Polar Mesosphere Summer Echoes, the first multi-beam experiments using up to 97 quasi-simultaneous beams in the mesosphere have been carried out in 2010 and 2011. These results provide a first insight into the horizontal variability of polar mesosphere summer and winter echoes with time resolutions between 3 and 9 minutes. In addition, first meteor head echo observations were conducted during the Geminid meteor shower in December 2010.

1. Introduction

[2] Doppler radars operating in the VHF band between 40 MHz and 55 MHz have been used for the investigation of various dynamic processes in the middle and lower atmosphere for several decades. Their capability to detect coherent backscattered signals from altitudes starting around 2 km up to ∼100 km lead to the name MST radar, standing for mesosphere/stratosphere/troposphere. A number of these types of radar have been developed and installed around the world since the middle 1970s. Reviews on the scientific and technological development of MST radars from the early days to now have recently been given by Röttger [2010] and Hocking [2011].

[3] In 1998, the Leibniz Institute of Atmospheric Physics in Kühlungsborn, Germany (IAP) installed the ALWIN radar [Latteck et al., 1999] as a successor to the Alomar SOUSY radar [Singer et al., 1995] on the North-Norwegian island Andøya (69.30°N, 16.04°E). ALWIN was designed for unattended and continuous observations of the troposphere and lower stratosphere, and especially for observations of the mesosphere during summer time. The radar operated at a frequency of 53.5 MHz with a peak power of 36 kW and a minimum range resolution of 150 m. The antenna array consisted of 144 Yagi antennas arranged in 36 squared subsystems of four antennas. The system allowed operation in spaced antenna (SA) or in Doppler beam swinging (DBS) configuration with a 6.6° wide radar beam (full width at half power) which could be pointed to a fixed number of 9 beam directions.

[4] After 10 years of nearly continuous operation the ALWIN radar was switched off in September 2008 to be replaced by a new, more powerful and more flexible radar system. The new Middle Atmosphere Alomar Radar System (MAARSY) is designed for atmospheric studies from the troposphere to the lower thermosphere, especially the investigation of horizontal structures of Polar Mesosphere Summer Echoes (PMSE) caused by mesospheric ice clouds. An overview on the current understanding of this phenomenon has been published by Rapp and Lübken [2004]. MAARSY is capable of classical DBS operation with improved temporal and spatial resolution and free beam steering capability, multiple beam transmission and reception, and multiple receiver and multiple frequency operation for interferometric applications.

[5] The work on MAARSY started in September 2008 with the dismantling of the antenna array of the ALWIN radar. Parts of the antenna array and the equipment building housing the ALWIN transmitter and receiving units were moved approximately 100 m westward of the site to a new leveled area to be used as an interim solution called ALWIN64 during the construction period of MAARSY [Latteck et al., 2010].

[6] The construction of the new radar started in May 2009 after the ground preparation of the antenna area was completed. The installation of the antenna array, the feeding cable network as well as the infrastructure for radar control and communication was completed in August 2009. The installation of the radar control and data acquisition hardware as well as an initial expansion stage of 196 transceiver modules was completed in spring 2010 and the radar began full-time unattended operation for PMSE observation in May 2010. An upgrade to 343 transceiver modules was installed in November 2010 and the final extension for 433 transceiver modules has recently been completed in May 2011. Figure 1 shows a view of the MAARSY site with the main antenna array, the six equipment buildings at the perimeter of the array accommodating the transceivers, and the main building housing control equipment and data acquisition hardware. The antenna array located on the right hand side of the photograph is the receiving antenna of the ALWIN64 interim system.

Figure 1.

View of the MAARSY site in February 2010 showing the main antenna array, the six equipment buildings at the perimeter of the array, the main building housing control and data acquisition hardware, and the equipment building and receiving antenna array of the ALWIN64 interim system.

[7] The general idea of MAARSY has previously been presented by Latteck et al. [2010]. This paper focusses on the hardware, especially on the design of the antenna and the array, the distributed transceiver network and possible operational modes. Furthermore, first results using the two stages of extension in 2010 as well as the completely installed system in 2011 are presented to demonstrate the performance and functionality of the radar. A validation of the quality of wind measurements obtained in the troposphere and mesosphere using various operational modes and analysis techniques is presented in an accompanying paper by G. Stober et al. (manuscript in preparation, 2011).

2. Radar Design

[8] MAARSY is a monostatic radar with an active phased array antenna as used with modern MST radars and employed by the MU radar in the early 1980s for the first time [Fukao et al., 1980; Kato et al., 1984; Fukao et al., 1985a, 1985b]. The operational frequency of MAARSY is 53.5 MHz and the maximum peak power is approximately 800 kW. The nearly circular antenna array with a diameter of approximately 90 m corresponding to an aperture of ∼6300 m2 results in a symmetric antenna radiation pattern with a beam width of 3.6° (full width at half power). Pulses with various shapes and length from 0.33 μs up to 200 μs can be transmitted within an allocated bandwidth of 4 MHz. The operating frequency and the beam position can be changed after each inter pulse period (IPP). Figure 2 shows a general block diagram of MAARSY, the basic system parameters of the radar are given in Table 1.

Figure 2.

Block diagram of MAARSY. Each antenna is connected to its own transceiver. The colored areas of the antenna array framed by thick lines represent the antennas connected to transceivers accommodated in equipment buildings colored respectively. The antenna array is arranged in 55 symmetric subgroups containing 7 antennas each and 6 asymmetric areas containing 8 antennas each. The receiving signals of the transceivers of a subgroup are combined as shown in the upper left part of the block diagram for e.g. hexagon A-01. A total number of 61 combined receiving signals are lead to the main building for signal processing. A 61:16 combining and switching unit is used to either combine the receiving signals of 7 adjacent hexagonal antenna subgroups or select one of them as an input for one of the 16 channel data acquisition system. Alternatively antenna signals from e.g. a meteor interferometer can be switched via an antenna interface unit to the data acquisition.

Table 1. Basic Parameters of MAARSY
LocationAndenes, Norway 69.30°N, 16.04°E
Operating frequency53.5 MHz
Allocated bandwidth4 MHz
Frequency agility52.5 MHz–54.5 MHz
Peak power∼800 kW
Maximum duty cycle5%
Pulse repetition frequency≤30 kHz
Pulse length≥0.33 μs
Sampling resolution≥50 m
Transmitted waveformssingle pulse, complementary and Barker codes
Pulse shapessquare, Gaussian, shaped trapezoid
Antenna array433 three-element Yagi
Effective antenna area∼6300 m2
Half power beam width3.6°
Directive gainmaximum 33.5 dBi
Beam directionsarbitrary at zenith angles <30°
Receiving channels16

[9] The system is composed of an active phased antenna consisting of 433 individual antennas and an identical number of transceiver modules with independent phase control and a scalable power output up to 2 kW. This arrangement allows very high flexibility of beam forming and beam steering with a symmetric radar beam and arbitrary beam pointing directions down to 30° off-zenith. The transceivers are accommodated in six equipment buildings placed with equal spacing around the perimeter of the antenna array.

[10] The antenna array is divided into 61 subgroups. 55 subgroups consisting of 7 antennas each are arranged in symmetric hexagonal structures. 6 asymmetric subgroups consisting of 8 antennas each are located at the perimeter of the array. The output signals of the transceivers of a subgroup are combined to a “receiving signal” as shown in the top left part of Figure 2 for e.g. subgroup A-01. Since this combination is done at an intermediate frequency (IF) the received signals are labeled e.g. “A-IF-01,” where “A” denotes the equipment building and the number indicates the subgroup. A total of 61 receive signals, 10 from each equipment building A-E each and 11 from equipment building F, are fed by coaxial cables of equal length to the adjacent main building for further signal processing.

[11] The signal processing unit located in the radar main building and described in detail in section 2.2.5 is currently limited to 16 channels. Hence the 61 receiving signals are first fed to a 61:16 combining and switching unit. This unit contains nine independent and controllable modules (see section 2.2.4) and allows the further combination of receiving signals and their allocation to the 16 inputs of the signal processing unit. One signal processing channel is permanently connected to the combination of all 61 receiving signals representing the antenna signal of the whole array. This setup is used for classical Doppler Beam Swinging operation as described in section 2.4.1. The other 15 signal processing channels can be fed by receiving signals related to the 55 symmetric hexagonal antenna subgroups or by the combined receiving signals of 7 adjacent hexagonal antenna subgroups called “anemone.” The gray shaded area in Figure 4 shows the antenna area used for e.g. “anemone A.” A total of 7 combined receiving signals corresponding to anemone subgroups as depicted in Figure 11 can be formed. This functionality allows a wide range of receiving arrangements with different antenna configurations for interferometric or multi-receiver applications as described in section 2.4.2.

[12] Separately located receiving antennas used for e.g. interferometer observations of meteors (see section 2.4.4) or bistatic boundary layer observations can be connected to the data acquisition system via a 16 channel antenna interface unit or, alternatively, to receiving signals from the subgroups of the main array.

2.1. Antenna

2.1.1. Element Structure

[13] The antenna used for the MAARSY array was designed and manufactured by IAP. The main requirement of the antenna was a bandwidth of about 5 MHz in order to handle radar pulses as short as 0.33 μs without compromising other antenna characteristics such as gain, front-to-back ratio and sidelobe attenuation. Besides the electrical characteristics, a free-standing design (without any guying between the individual antennas or between active and passive antenna elements) was chosen to simplify the construction of the array and facilitate future maintenance.

[14] The antenna was designed using the kernel version 4 of the Numerical Electromagnetic Code (NEC) developed by Lawrence Livermore National Laboratory. NEC is based on the Method of Moments (MoM) code for analyzing the interaction of electromagnetic waves and is capable of lossy ground consideration using the Norton-Sommerfeld approximation.

[15] The antenna design selected was a linear polarized 3-element Yagi with a folded dipole as shown in Figure 3. The folded dipole is made of 20 mm aluminium round stock, whereas 30 mm aluminum tubes were used for the director and the reflector. The elements are mounted on a square aluminium pole of side 120 mm and 2500 mm length, and mounted on a large concrete block. The size of the concrete block was specified to provide sufficient stability to the free-standing antennas during the strong winds which frequently arise at the radar site. The antenna feed point is fed into a waterproof box mounted at the pole which houses a balun based on a half wavelength phasing line made of coiled coaxial cable. The characteristics of the antenna resulting from the modeling are a gain of 6.88 dBi and a bandwidth of 5 MHz defined by the return loss of −20 dB.

Figure 3.

Sketch of one 3-element Yagi antenna of MAARSY. The folded dipole is made of 20 mm aluminium round stock, whereas 30 mm aluminum tubes were used for the director and the reflector. The elements are fixed on a squared aluminium pole of 120 mm feed size and 2500 mm length and mounted on a massive concrete block.

2.1.2. Array Configuration

[16] The design of the MAARSY antenna array was inspired by the MU radar [Fukao et al., 1980; Kato et al., 1984; Fukao et al., 1985a] whereas the size of the array was restricted by area limitations due to the nearby ocean and a public road. Various models varying in numbers of antenna elements, spacing and orthogonal or triangular grid structure were simulated using NEC in order to find an optimum array design. The final array consists of 433 linear polarized Yagi antennas arranged in an equilateral triangular grid structure. Arranging the elements in a triangular grid generally allows antenna beam steering to larger off-zenith angles than using a rectangular grid with the same element density [Fukao et al., 1985a]. In addition, the square geometry requires approximately 16% more elements for the same amount of grating-lobe suppression [Skolnik, 2008]. The side length of the equilateral triangle or the spacing between the antennas respectively is 4 m (=0.71 λ) with one side pointing to the northwest-southeast direction. This orientation is also the alignment of one side of the square concrete footings and of the dipole mounting side of the square antenna poles. The whole arrangement forms a nearly circular aperture of ∼6300 m2 with a diameter of approximately 90 m as shown in Figure 4.

Figure 4.

Sketch of the MAARSY antenna array. The six boxes at the perimeter of the array, labeled as A–F, represent the equipment buildings. The antenna array is divided in 61 subarrays, 55 identical hexagons consisting of 7 antennas each and 6 asymmetric groups consisting of 8 antennas located at the perimeter of the array. Seven adjacent hexagonal antenna structures can form an “anemone” structure as indicated by the gray shaded area corresponding to equipment building A.

[17] The MAARSY antenna array is divided in 61 subarrays of which 55 are identical hexagons consisting of 7 antennas each and 6 are asymmetric groups consisting of 8 antennas each. These asymmetric groups are located around the perimeter of the array. Even though each antenna is connected to its own transceiver which is independently controllable in phase-offset, frequency and output power, the subgroup structure of 7 antennas (hexagon) is the minimum useful arrangement for transmission and reception due to a limitation in number of signal processing channels as described later in section 2.2.4. A further combination of 7 adjacent hexagonal structures as e.g. indicated by the gray shaded area in Figure 4 results in a so-called “anemone” antenna structure. Seven of these anemone structures can be realized in order to form beams for transmission and/or reception simultaneously allowing e.g. active multiple beam operations.

2.1.3. Array Pattern

[18] The radiation pattern of an antenna array can be approximated by the product of the array factor and radiation pattern of the individual antenna [Skolnik, 2008]. This approximation does not consider the effect of mutual electromagnetic coupling occurring in any array with closely separated antennas (or elements). The magnitude of mutual coupling depends on the distance between the elements, the pattern of the elements, and the structure in the vicinity of the elements. Furthermore the effect varies as the array forms off-zenith beams. Using the NEC Method of Moments the effect of mutual coupling is considered.

[19] Each antenna of the MAARSY array was formed by wires in the model and excited by a voltage source. The simulations were initially performed for free space environment to identify the pure mutual coupling of the elements, and finally for real ground environment using the Norton-Sommerfeld approximation. Figure 5 shows the computed radiation pattern of the MAARSY antenna array. The symmetric antenna radiation pattern of the full array pointing to zenith has a half power beam width of 3.6°, a directive gain of 33.5 dBi and an almost symmetric first sidelobe with more than 17 dB suppression with respect to the main lobe. The beam width of a hexagon structure is approximately 30°, and 11° for an anemone structure.

Figure 5.

Computed radiation pattern of the MAARSY antenna array. (left) Cartesian projection of the radiation pattern for a beam pointing to 0° (zenith). (right) Radiation pattern along the NW-SE orientation for beams pointing to 0° (blue) and to 21° (red) in NW direction (φ = 315°). The Numerical Electromagnetics Code (NEC) with the Norton-Sommerfeld approximation was used, making allowance for the imperfect ground at the antenna site.

[20] The MAARSY antenna array has also been simulated with NEC to determine the amount of mutual coupling between the individual antennas including the ground approximations. For this purpose, each simulation run included only one excited antenna, while the other antennas have been connected to a 200 Ohms load, representing the balun impedance transformator and the 50 Ohms transmit/receive-module. Applying the exact current antenna array structure of MAARSY a minimum coupling of −42 dB has been found for the closest neighboring antennas (4 m distance), while the antennas arranged along the dipole axis show even a coupling of −47 dB. The second ring of antennas (8 m distance) show an isolation of about 60 dB, while the isolation is generally increasing for any other antenna with its distance.

[21] A validation of the modeled antenna radiation pattern can be done by e.g. comparing it with a measured antenna diagram obtained from a cosmic noise source [Fukao et al., 1985a; Czechowsky et al., 1984]. A first test using the second stage of expansion of MAARSY was made in spring 2011 during the completion of the installation. A 35-beams point-to-point experiment was used to scan southward from zenith down to 34° off-zenith in steps of 1 degrees in order to locate Cassiopeia A and Cygnus A, which are the most intensive point sources detectable for MAARSY. Noise power maxima were observed at the expected time and positions of both sources, confirming the correct operation of the antenna beam steering. An active experiment using a receiving system on board of a helicopter to verify the radiation pattern on transmission is planned for 2012.

[22] The cartesian projection of the radiation pattern for a beam pointing to zenith as shown in Figure 5 (left) is symmetric in side-lobe suppression with respect to the NW-SE orientation due to the NW-SE orientation of the Yagi antennas. This characteristic leads to differences in the maximum off-zenith angle the beam can be pointed to without the appearance of grating lobes. The beam can be pointed down to a maximum off-zenith angle between 30° and 40° clear of grating lobes depending on the azimuth angle of the beam pointing direction as indicated by the blue line in Figure 6 (right).

Figure 6.

Cartesian projection of (left) the radiation pattern for a beam pointing to NW-39° and (right) azimuthal dependency of the maximum off-zenith beam-pointing angle. Due to the equilateral triangle grid structure of the array the beam can be pointed down to an off-zenith angle between 30° and 40° without the appearance of grating lobes.

2.2. Transmission and Reception

2.2.1. Antenna Feeding Network

[23] Each antenna of the array is connected to its own transceiver module accommodated in six equipment buildings and equally distributed around the perimeter of the array in a similar manner as for the MU radar [Fukao et al., 1985a]. Five of the six equipment buildings house 71 transceivers while equipment building F additionally services the 7 antennas of the center hexagon. The antennas are connected via low-loss coaxial cable (−1.8 dB/100 m) to bulk-head connectors located at cable entry plates in the building walls. Flexible patch cables inside the equipment buildings connect the bulk-heads sockets with the transceiver ports and correct differences in phase-matching of the field cables.

[24] In contrast to other phased array radar systems as e.g. the MU radar, MAARSY uses feeding cables of equal length (12λ) instead of a short connection in combination with an electronic phase and runtime compensation. The use of cables of equal length provides a feeding network with stable characteristics of attenuation, group delay and phase delay which needs no further configuration and maintenance. Figure 7 shows a sketch of the cable laying as used for the connection of the antennas fed by transceivers in equipment building F. The 433 phase matched feeding cables have a mean attenuation of 1.33 dB with a RMS deviation of 0.34 dB and 0.17° due to manufacturing tolerances and accuracy in phase-match cutting.

Figure 7.

Layout of laying of the RF cables connecting the antennas of the eleven subgroups (hexagons) with the transceivers in equipment building F. The thick gray lines represent cable trays of different widths mounted 15 cm above the ground to allow crossings. The squares indicate the antenna footings made of concrete.

2.2.2. Transceiver Module

[25] The transceiver modules are state of the art VHF solid state pulsed transmitters and down converters designed and manufactured by Genesis Software Pty Ltd.

[26] The transceiver module is implemented as a plug-in module in order to facilitate maintenance and reconfiguration of the system. The transceiver interfaces to a backplane carrying control and timing signals to all connected modules in the common enclosure. Up to four dual-channel transceiver modules are housed in a single multi-channel transceiver chassis (MCTC) and this entity is considered the basic building block of the radar hardware. Each dual-channel transceiver is under the control and supervision of a microprocessor which, in turn, is in communication with a local host computer. The microprocessor allows comprehensive control over all aspects of the two transceiver channels' behavior, such as frequency, phase, and gain control on a pulse-to-pulse basis. The microprocessor also monitors output power, phase, receiver channel phase, reflected voltage, and other parameters. The transceiver module is made up of the following submodules, each of which can be replaced as a field serviceable module; a microprocessor-based control module, two power amplifier modules, two receiver down converter modules, and a gain-phase detector module. Figure 8 shows a simplified block diagram of the transceiver module, the submodules are described in more detail in the next paragraph.

Figure 8.

Simplified block diagram of the MAARSY dual-channel transceiver module.

[27] The control module provides both an interface to the MCTC backplane as well as a physical substrate upon which the other transceiver modules are mounted. The core of the control module is a quad-channel Direct Digital Synthesizer (DDS) operating at an internal clock frequency of 336 MHz. The DDS is configured by the control module's microprocessor via a serial interface and provides the reference oscillators for the power amplifiers (carrier frequency) and the down converters (RF local oscillator frequency). Both the frequency and phase of these four reference oscillators may be adjusted by the microprocessor on a pulse-to-pulse basis, allowing both rapid steering of the transmitted and received Doppler beam formed by the full (or partial) array, as well as frequency domain interferometry operation. Pulse repetition frequencies (PRF) of up to 30 kHz are supported, although the maximum PRF available is ultimately limited by the power amplifier maximum permissible duty cycle as well as the amount of configuration information provided to the DDS on a pulse-to-pulse basis.

[28] The control module monitors various transceiver status pulse-to-pulse. In addition to those items mentioned previously, additional information such as high voltage status, power amplifier heat sink temperature, load impedance, fuse status, and DDS “lock” status are all collated and logged. In the event of a failure, or the detection of a threatening condition, the relevant transceiver channel will be prevented from further operation in order to protect it from damage. Such events are logged as transceiver “alarms” by the host computer at the end of each data acquisition sequence.

[29] The transceiver power amplifier module is implemented as a bolt-on entity mounted on the control module. The power amplifier consists of various submodules; a 2 kW FET array-based power module equipped with a heat sink, output stage matching, modulation circuitry, TR switch, output filter, and directional coupler. Modulation control and various status monitoring signals travel to or from the control module via a plug-in connector which is automatically joined when the power amplifier is physically mounted on the control module, while RF signals such as RF drive and the TR switch's receive output signal travel to other modules of the dual-channel transceiver via shielded coaxial cables.

[30] The directional coupler embedded in the power amplifier design provides samples of forward and reflected voltages to the gain phase detector module (discussed below). This allows the control module to monitor output power, output phase, and complex load impedance on a pulse-to-pulse basis. The output filter provides suppression of any spurious emissions by means of a band pass filter centered on the carrier frequency. A separate band stop filter centered on the second harmonic is also provided in the output filter. This is to mitigate any interference to local FM broadcast reception as the second harmonic falls squarely in the broadcast band.

[31] The transceiver down converter module is implemented in a similar way to the power amplifier in that control and status signals travel to or from the control module via a plug-in connector while RF signals travel via coaxial cable. The down converter is comprised of a broadcast filter, low noise amplifier, RF bandpass filter, mixer, IF bandpass filter, variable gain amplifier and output buffer. The overall gain of the down converter is user-selectable, and can range from zero to 60 dB.

[32] The transceiver down converter module is equipped with RF switches allowing the controller to select a test signal for reception rather than the usual receiver output from the power amplifier's TR switch. The test signal is provided by the DDS located on the transceiver control module, and is therefore fully configurable. This allows the controller to provide an excitation signal to the down converter to check signal paths and characterize gain, through-phase and even the passband shape of the down converter filters.

[33] The final submodule making up the transceiver module is the transceiver gain phase detector. This is a 6-input-port module which allows comparison of a reference signal with any of the 6 inputs. The reference signal is provided by an external module via the transceiver backplane and, since this reference signal is common to all transceivers in a given cabinet, allows for common measurements of, for example, output phase to be made and compared. This provides the operator with a reliable indicator of transceiver phasing and synchronization on a module-to-module basis. The reference signal is switched in quadrature on a pulse-to-pulse basis, which allows full four-quadrant characterization of selected input phase. This allows, among other things, complex impedance measurements of the load (antenna) to be made.

[34] The IF outputs of the dual-channel transceiver modules housing in a MCTC are combined using 7:1 or 8:1 IF combiners implemented as single-stage Wilkinson combiners with all ports matched to 50 Ohms.

2.2.3. Reference and Synchronization

[35] System synchronization is achieved by provision of two timing signals to all six distributed equipment buildings via 12 phase-matched pairs of coaxial cables. The timing signals originate in the main control building where a Reference and Test Signal unit in the data acquisition system supplies reference oscillators and trigger synchronization facilities. The timing signals consist of a “System Reference” clock which operates continuously at a frequency of 84 MHz, and a “System Trigger” which is produced at a rate commensurate with the pulse repetition frequency chosen for the experiment in operation. The temporal relationship between the active edge of each System Trigger and the nearest active edge of the System Reference is guaranteed by the reference and test signal unit. This relationship is invariant and is selected to prevent slight variations in phase-matched cable length from creating any instances of anomalous behavior due to out-of-specification timing downstream. In addition, the signals are distributed around the array as differential pairs in order to mitigate interference from the radio frequency pulses transmitted from the main antenna array.

[36] The System Reference signal is used by the transceiver modules as a clock reference for the quad-core DDS used to generate the signals required for the dual channel transceiver (see section 2.2.2). The System Trigger signal is used as a phase-initialization signal to the DDS and this ensures that all spatially-separated DDS's operate synchronously.

[37] Oversight on the state of system synchronization is provided by means of a third set of signals circulating around the six equipment buildings. These are Synchronization Reference signals which allow the equipment in each building to compare a locally-generated reference signal with a similar reference signal produced in the adjacent equipment building. Thus, building A compares its local reference with building F, while simultaneously providing a copy of its internally-generated reference to building B for similar comparison. This process is duplicated in building B, and so on around the array. For each and every System Trigger generated by the data acquisition system in the main building, the phase relationship between the local and remote Synchronization References in all 6 equipment buildings are recorded and examined at the end of the data acquisition sequence. Local control and monitoring software reports statistics derived from the recorded phase relationships and these are available for scrutiny in local log files. In this way, the state of building-to-building synchronization can be initially established and monitored on an on-going basis.

2.2.4. IF Combining and Switching

[38] The combined receiving signals corresponding to the 61 antenna subgroups of the antenna array as described in section 2 are led by coaxial cables laying on cable trays around the antenna array to the nearby located radar control building. Low-loss cables of type LMR-400-DB and of equal length are used for the connection between the cable-entry plates located in the equipment building walls and the wall of the radar control building, whereas flexible patch cables connect the bulk-head sockets in the cable-entry plates to the combiner ports in the equipment building and in the control building.

[39] The 61 receiving signals are managed by 61:16 port combining unit containing nine independent “IF Switch-Combiner Units” (IFSCU) controlled by a micro-controlled module under computer control known as the “IF Combiner Control Unit” (IFCCU). This idea is also inspired by the MU radar [Fukao et al., 1985b] but the realization was completely designed from scratch and adjusted to the MAARSY array configuration.

[40] Figure 9 shows a block diagram of an IFSCU. Each input signal is first split in two using a 50 Ω resistive splitter and the two copies of the input signal are fed to the combiner and a switch matrix respectively. The 7:1 (or 6:1) combiner is implemented as a single-stage Wilkinson combiner [Wilkinson, 1960] each arm of which transforms the 50 Ω IF feed impedance to 350 Ω (for the 7:1 combiner) at the common, combined point. For the 6:1 versions of the IFSCU, the combined point's target impedance for each arm of the combiner is 300 Ω. This combined signal is then fed to one of two banks of IF filters. The switch matrix allows one of the 7 (or 6) inputs to be selected under microprocessor control while ensuring that the nonselected inputs are properly terminated with 50 Ω loads. The selected signal is presented to a second, independent bank of IF filters. The two IF filter banks are each comprised of four 4th-order Butterworth filters with 3 dB bandwidths of 6 MHz, 3 MHz, 1.5 MHz and 1 MHz. The four filters are equipped with fixed attenuator pads which compensate for the varying insertion loss of each filter, thus ensuring that the IF signal level does not vary when switching from one filter to another, at least so far as in-band signals are concerned.

Figure 9.

Block diagram of an IF Switch-Combiner Unit (IFSCU). Each of the 7 IF input signals can be selected and routed to the “subgroup” output. Simultaneously the combined IF signal of all 7 (or 6) inputs is present at the “combined” output. A third output (“subgroup/combined”) can ether provide the selected subgroup signal or the combined signal.

[41] The combined and filtered signal is subsequently passed to a programmable attenuator under microprocessor control. This allows attenuation of the combined and filtered signal between 0 and 31 dB in 1 dB steps. It should be noted that the theoretical maximum increase in signal level when combining 7 identical signals is 8.5 dB. This programmable attenuator can be used to keep the combined feed within the signal processor's maximum input signal excursion range. The combined, filtered and attenuated signal is then split in two via a resistive divider. One half of the divided signal is presented to the combined output port at the rear of the IFSCU while the other half is presented to one input of a single pole, double throw absorptive RF switch. The other input of this RF switch is presented with the output from the IF input switch matrix discussed above. One of these two signals is selected under microprocessor control and presented to the switched IF output at the rear of the IFSCU. Thus, the switched IF output from each IFSCU can take the form of any one of the IF inputs or, alternatively, the “combined” IF output. In either case, the signal has undergone filtering and, potentially, combining and attenuation.

[42] The combined IF signals provided by the 9 IFSCU at the “combined” output as shown in Figure 9 are further merged in the IFCCU using a cascaded set of four 3:1 Wilkinson combiners followed by a programmable attenuator to a combined IF signal as shown in Figure 10. This IF feed containing the received signals of the full array is linked to the input of data acquisition channel 1 permanently. This setup is used for the classical Doppler Beam Swinging operation of the radar system as described in more detail in section 2.4.1.

Figure 10.

Block diagram of the array combining part of the IF-combining unit. The IF signals provided by the 9 IFSCU at the “combined” output are further merged in hardware to a combined IF signal of the full array.

[43] The “subgroup/combined” outputs of the 8 IFSCU (A, B, C, D, E, F, and M; see Figure 10) are connected via the “IF switch Unit” to channel 2 to 8 of the Signal Processing Unit (SPU). The IFSCU allows selection of either the combined IF feeds of 7 anemone groups or the signals from selected single hexagons to these seven inputs of the Signal Processing Unit. Simultaneously the selected subgroup (hexagon) signals of anemones A, B, C, D, E, F, M and S are present at the inputs 9 to 16 of the SPU. This flexibility allows the application of various analysis techniques on the basis of spaced antenna arrangements as described in section 2.4.2.

[44] Alternatively to this standard allocation of MAARSY main array subgroups to the 16 receiving channels of the SPU, RF signals from separate receiving antennas e.g. from a 5 channel meteor interferometer or from the 16 subgroups of the ALWIN64 array [Renkwitz et al., 2011] can be switched to the SPU inputs via a 16 channel Antenna Interface Unit (AIU) as shown in Figure 2. Table 2 shows the standard allocation of MAARSY antenna structures and switchable external receiving signals to the inputs of the signal processing unit.

Table 2. Allocation of MAARSY Antenna Structures and External Receiving Signals to the Inputs of the Signal Processing Unita
Signal Processing Unit InputsMAARSY Antenna StructuresExternal Signal Sources
  • a

    Single capital letters indicate anemone structures as e.g. A is “anemone A,” and “A-xx” stands for a selected hexagon of anemone A. “Met” and “But” are signals from outside the MAARSY antenna array as from meteor antennas or the Butler matrix [Renkwitz et al., 2011] respectively.

01whole arrayMet-01
02A or A-xxMet-02
03B or B-xxMet-03
04C or C-xxMet-04
05D or D-xxMet-05
06E or E-xxBut-01
07F or F-xxBut-02
08M or M-xxBut-03

2.2.5. Signal Processing

[45] A subset of the IF signals described in section 2.2.4 are routed to the inputs of the 16-channel Signal Processing Unit. Each channel of the signal processing unit comprises an analogue quadrature down converter to provide in-phase and quadrature-phase base band signals, a pair of 16-bit analogue-to-digital converters, and an FPGA-based signal averaging module. An additional module in the Signal Processing Unit controls the flow of digitized and averaged data from the 32 digital channels to the host computer via a buffer module.

[46] The analogue quadrature down converters accept the IF signals centered at 10.7 MHz and mix these with quadrature reference oscillators provided by the Reference and Test Signal unit to provide quadrature baseband signals. These are band-limited by a pair of selectable base band filters which allow the operator to match the receive signal bandwidth to the bandwidth of the transmitted signal. The filtered base band signals are amplified and provided to monitor points at the front of the unit for scrutiny by the operator, and to the digitiser/signal averager modules. Although the gain is distributed through various parts of the signal path in the analogue down converter, the overall signal gain provided by the down converter is precisely calibrated to 50 dB.

[47] The analogue to digital converter and FPGA-based averager are implemented as a single, dual-channel module. The in-phase and quadrature-phase analogue base band signals from the quadrature down converter are independently digitized to 16-bit resolution at a constant digitization rate of 6 MHz. Due to limitations in the samples control module, every second sample digitized is used which equates to a minimum sample resolution of 50 m. The twin streams of 16-bit words are sorted by the FPGA firmware into radar range gates under the direction of control signals provided by other modules in the data acquisition system. Coherent averaging is performed by the FPGA firmware for each independent range gate and the normalized averaged data points made available for transfer to a buffer module. As data become available in the data buffer module they are transferred in bursts to the host computer via a USB IO interface where they are made available for processing and storage. The data throughput rate to the host computer is around 10–15 MB/s.

[48] All sampling parameters may be varied at the discretion of the operator within wide limits. For example, the minimum sample resolution available is 50 m while there is effective no upper limit. The number of radar range gates which may be acquired may vary from 1 to 211 (2048), while the number of coherent integrations (CI) may be selected from 1 to 216 (65536). Further, data may be acquired in an unbroken stream allowing very long time series to be acquired. Refer to section 2.3 for a detailed description of the control functionality of the radar system.

2.3. System Control

2.3.1. Host Computers

[49] Configuration, control and acquisition activities are instigated by the host computer in the main radar control building under the direction of the radar system operator. Relevant configuration information is passed from the main host computer to six independent host computers located in the six transceiver buildings via an ethernet connection. Responsibility for configuration, control and monitoring of the transceiver equipment is delegated to these six computers, which remain in continuous contact with the main host computer.

[50] Whenever the operator initiates an experiment sequence, the main host computer establishes contact with the distributed hosts and issues instructions for hardware initialization to take place. Once complete, the experiment sequence is repeated until the operator halts execution, or a nonrecoverable error is encountered. Each experiment in the sequence is conducted with the aid of a set of control phases under the direction of the main host. These phases include configuration, disk space check, hardware status report, data consumer configuration, pre-data-stream configuration, trigger initiation, data acquisition, and, finally, post-stream hardware status check. All relevant information from these various control phases is written to log files and is available for immediate or later scrutiny, either by local operators or by personnel observing from remote locations.

2.3.2. Transceiver Supervisor Unit

[51] The transceiver supervisor unit provides control and monitoring functions over a set of multi-channel transceiver chassis housed in a single equipment cabinet. There are two such cabinets of equipment in each transceiver building in the MAARSY radar system. Each transceiver supervisor unit is capable of controlling up to 8 multi-channel transceiver chassis, and therefore, up to 32 individual (dual-channel) transceivers.

[52] The transceiver supervisor unit is comprised of four main subcomponents. These are, (1) a microcontroller module which provides a communications “bridge” between the host computer and the transceivers, (2) a reference generator module, (3) a quadrature down converter module, and (4) a main board for signal distribution. The microcontroller module also provides the digitally-derived analogue modulation shape for the pulse sequence in use. This may include coded pulses (Barker and Complementary), multi-pulse sequences, single pulses, etc. The reference generator module is used to provide the reference signal from which the transceivers measure such operating parameters as output phase, etc., and also provides IF reference oscillators in quadrature so that the local down converter module can be used to monitor converted IF signals in the equipment building.

2.3.3. Pulse-to-Pulse Control

[53] Pulse-to-pulse operation is implemented with the aid of a command/information protocol named “Pulse Slot Configuration,” or PSC. The PSC protocol allows for efficient programming of multiple transceivers in multiple interleaved modes of operation. For example, a large number of Doppler beam directions may be sequenced through during a single data acquisition period. In addition, the frequency of operation (52.5 MHz : 0.08 Hz : 54.5 MHz), transmitted power (40 dBm : 1 dB : 63 dBm), receiver gain (63 dB : 1 dB : 120 dB) and so on may all be simultaneously changed on a pulse-to-pulse basis. This flexibility implies that a large amount of information must be supplied to the transceivers in order to fully define their behavior for each experiment, potentially creating down time between experiments for communications to take place.

[54] The PSC protocol was designed to mitigate inter-experiment dead time by utilizing two techniques. First, PSCs may be sent to any number of individually-addressed transceivers or, alternatively, may be sent to a so-called “broadcast” address to which all transceivers respond. Second, only information which changes from one pulse to the next for a given set of transceivers is communicated. All other settings are assumed to be unchanged until the host explicitly sends updated information. Finally, information for a given pulse slot may be broken up and sent over several commands. This allows the host to broadcast common information to all transceivers and then update individual transceivers with additional information as necessary.

2.4. Operational Modes

[55] The design of MAARSY offers a multiplicity of radar operational modes. The active phased antenna in combination with transceiver modules independently controllable in phase-offset and output power provides a very high flexibility in beam forming and beam steering. The segmentation of the antenna array into 55 symmetrical subgroups for reception in combination with a 16 channel signal processing unit allows a wide range of receiving arrangements with different antenna configurations. The availability of multiple simultaneous receiving signals allows spatial domain interferometric applications [e.g., Woodman, 1997; Palmer et al., 1998; Chilson et al., 2002; Chen et al., 2008]. Furthermore the change of the operating frequency from pulse-to-pulse offers interferometric applications in the frequency domain [e.g., Kudeki and Stitt, 1987; Palmer et al., 1999; Luce et al., 2001; Chen and Zecha, 2009].

[56] The radar operation is controlled by a master computer. All experiment relevant parameters as e.g. pulse repetition frequency, pulse length and pulse shape, number of coherent integration, range limits and range resolution, number of beams and beam directions can be configured by a graphical user interface. Several experiments can be set up and run consecutively in a sequence. The TRx hardware as well as the data acquisition system define a few but essential limitations to the system operation as e.g. maximum duty cycle (5%), maximum number of range gates (2048) or PSC stack depth (50) for selected experiment parameters which need to be considered.

[57] The following sections describe the major operational modes and their characteristics with respect to the MAARSY design.

2.4.1. Doppler Beam Swinging Mode

[58] The Doppler Beam Swinging (DBS) operation is the most straight forward operation mode of an atmospheric radar. Pulses are transmitted and received through the main beam formed by phasing of all transceivers connected to the antennas of the whole array. At MAARSY the received signals of all 61 antenna subgroups are combined to one signal and led to signal processing channel 1 permanently. The beam can be steered to several directions with off-zenith angles down to 40° and all azimuth directions as shown in Figure 6.

[59] Since the PSC stack can store up to fifty parameter settings for e.g. phase offsets or frequency shifts, and the parameters can be changed with the pulse repetition frequency, fifty beam directions can be configured in one experiment allowing a three-dimensional scanning of the atmosphere. The temporal as well as the horizontal and vertical resolution of the scan depends on the altitude range to be observed and the limitation in range gates.

[60] A typical experiment configuration for e.g. mesospheric observation is using a PRF of 1250 Hz and a 8 bit complementary code and covers an altitude range between 70 and 100 km. If 50 individual beam directions and no further coherent integration are configured to be used in one experiment the resulting time series per range gate would have a minimum temporal resolution of 80 ms and a Nyquist frequency of 6.25 Hz what is for example sufficient for the analysis of PMSE. An improvement in time resolution and signal-to-noise ratio can be achieved by reducing the number of beam directions per experiment and by increasing the number of coherent integrations, respectively. Several similar experiments but using different beam directions can be set up in an experiment sequence to get a quasi-simultaneous scan with increased horizontal coverage. Results using a corresponding experiment with partial expansion stages of MAARSY are presented in section 3 of this paper and by Rapp et al. [2011].

2.4.2. Spaced Antenna Mode

[61] Spaced Antenna (SA) techniques are based on the analysis of signals simultaneously received by spatially separated antennas. Various methods using correlation analyses, spectral analyses, software beam steering or image forming on the basis of received signals from spaced antennas have been developed and applied to atmospheric radars since the end of the 1940s. A special application commonly used with MST radars is the Full Correlation Analysis (FCA [see, e.g., Briggs, 1984]). A comprehensive overview of the various techniques is given by Holdsworth [1995].

[62] Applying SA techniques to MAARSY takes advantage of the 55 hexagonal substructures of the antenna array in combination with the IFSCUs (see Figure 9) used in the IF combining and switching unit. Hexagon as well as anemone antenna structures can be selected and the corresponding receive signals can be simultaneously directed to the inputs of the signal processing channels. Figure 11 shows an example of a possible allocation of MAARSY antenna structures to the fifteen receiving channels. The receiving signals corresponding to the seven anemone structures marked in light colors are allocated to receivers 2 to 8, whereas receiving signals from eight individual hexagon antenna structures marked in dark colors are connected to receivers 9 to 16. Receiver 1 is always connected to the combined signal of the full array (see section 2.4.1).

Figure 11.

Example of the allocation of MAARSY antenna structures to receiving channels. The signals related to the seven anemone structures marked in light colors are allocated to receivers 2 to 8, whereas the signals from eight individual hexagon antenna structures marked in dark colors are connected to receivers 8 to 16. Receiver 1 is permanently connected to the total combined signal of the whole array.

[63] A typical tropospheric experiment using a PRF of 5 kHz, a shaped monopulse of 2 μs and 512 coherent integrations covering a maximum range of 30 km with a spatial resolution of 300 m and a temporal resolution of 100 ms would commonly use the receiving signals from the anemone structures for SA analysis techniques. Since seven anemone signals are simultaneously available and at least three receiving signals from separated antenna are required for FCA, various setups using two different base line lengths are possible e.g. Rx02-Rx03-Rx08 or Rx02-Rx04-Rx06. As another example, the observation of PMSE requires a different experiment configuration and shorter base lines. Since the use of hardware combiners due to the limited number of sixteen signal processing channels does not allow an overlapping of anemone structures or larger antenna structures, short baselines are given by the distances of hexagon structures only. A typical setup for PMSE observation would use e.g. Rx11-Rx12-Rx15 or Rx08-Rx09-Rx10.

[64] For getting optimal results from both the troposphere and mesosphere using spaced antenna techniques, experiments optimized to e.g. duty cycle and temporal resolution need to be configured and used in a sequence. Since receiving signals from anemone structures and hexagon structures can be used simultaneously as shown in the example of Figure 11, a suboptimal experiment can be used for probing the troposphere and the mesosphere simultaneously. Such an experiment monitoring the whole MST region takes also advantage of the dual-region feature of the MAARSY data acquisition system allowing the sampling of two selectable altitude ranges in one experiment.

2.4.3. Multiple-Beam Mode

[65] Seven combinations of seven adjacent hexagonal substructures of the MAARSY antenna array can be configured to be used independently from each other. These so-called anemone structures as labeled by different colors in Figure 11 can be realized to form seven beams simultaneously for active multiple beam operation. The phase offsets of the transceivers can be configured independently for each anemone and the IFSCU (see Figure 9) provide combined signals related to the anemone antenna structures for seven inputs of the signal processing unit. Figure 12 shows the radiation pattern of the MAARSY array for a multiple-beam experiment pointing to 7 directions simultaneously. The six 30° oblique beams realized by the anemones A, B, C, D, E and F point to n · 60° + 15° (n = 0…5) azimuths. Since the half-power beam width of an anemone antenna patch is 11° and the simultaneous forming of seven anemones yields to asymmetric beams due to mutual coupling the range of oblige angles is limited. Nevertheless the use of multiple beam experiments increases e.g. the temporal resolution and reliability of wind profiles since the radial velocities obtained from different directions to be used for the wind determination are measured at the same time and not consecutively.

Figure 12.

Cartesian projection of the computed radiation pattern of the MAARSY array for a multiple-beam experiment pointing to 7 directions simultaneously.

2.4.4. Meteor Mode

[66] MAARSY with its flexible beam forming capabilities and its high transmit power is able to measure the short duration radar reflections from the plasma which surrounds and moves with the burning up meteoroid. These weak returns are called head echoes [e.g., Close et al., 2002] and the ablating meteoroid intersects the narrow radar beam. The head echo detection is also called “down the beam” mode. Results from recent observations using various high-power large aperture radars and further reading can be found, e.g., in the works of Close et al. [2002] and Chau and Woodman [2004]. Initial results from observations of head echo during the Geminids meteor shower in December 2010 are presented in section 3.

[67] The specular meteor mode of MAARSY uses the multiple beam configuration as described in section 2.4.3 with full transmit power or a donut-like beam configuration with reduced transmit power. The latter is formed by a number of 234 antennas along the perimeter of the antenna array for transmission in order to ensure a nearly uniform azimuthal sensitivity to meteor trails. The chosen part of the array is divided in twelve equal slices containing 19 or 20 antennas each. Each group is holding a separate phase offset and produces a separate beam pointing to 30° off-zenith and n · 30° + 15° (n = 0…11) azimuthal direction. The resulting multiple beam shows a donut like radiation pattern with a maximum gain at approximately 30° off-zenith as shown in Figure 13. The meteor trail echoes are located by a 5-channel interferometer as described by Jones et al. [1998] and Hocking et al. [2004]. The five separate 2-element Yagi antennas forming the interferometer are located eastward of the ALWIN64 array shown in the very right part of Figure 1. The antennas are connected via the antenna interface unit to the combining and switching unit in the main building as depicted in the upper right part of Figure 2. The corresponding IF outputs of the AIU can be switched to data acquisition channel 1 to 5 as shown in Table 2.

Figure 13.

Cartesian projection of the computed radiation pattern of the MAARSY array to be used for transmission in interferometric meteor mode.

[68] The experiment configuration for specular meteor mode can be adapted from the operational settings of a SKiYMET system operated at Andenes and described, e.g., by Singer et al. [2004]. A 13.5 μs wide Gaussian shaped pulse in combination with a PRF of 2144 Hz and 4 coherent integrations allows a ranged aliased probing above 76 km at a duty cycle of 2.9%. However since the transmitter output power as used in the interferometric meteor mode with MAARSY is approximately 40 times higher compared to the output power of a classical meteor radar a direct probing with improved sampling resolution is possible with MAARSY.

[69] Longer lasting persistent radar reflections are observed from the expanding plasma trail behind the decaying head echo. Strong returns are observed when the radar beam and meteor trail are perpendicular in the so-called “specular” mode. These specular echoes from underdense meteors are the basis for classical meteor radars which allow e.g. the estimation of meteoroid velocities, orbits and meteor shower radiants, neutral gas winds in the altitude range 80 to 100 km, and temperatures at the mesopause level [e.g., Hocking et al., 2001; Elford, 2001].

3. First Results

[70] The construction of the MAARSY antenna array and the infrastructure for radar control and communication was completed in August 2009. The radar control hardware for synchronization, triggering and communication needed in every of the six equipment buildings, the master control system, an interim design of the combining unit and the data acquisition hardware housed in the main building was installed during winter 2009/2010. The radar operation started with an initial expansion of 196 transceiver modules in May 2010. A second stage of expansion to 343 transceiver modules was brought into service in November 2010. The final extension to 433 transceiver modules has recently been completed in May 2011. First results using the second stage of extension as well as the complete system during campaigns in 2010 and 2011 are presented here to demonstrate the performance of the new radar.

3.1. Validation of Tropospheric Wind Measurements

[71] After a short period of test operation MAARSY started full-time unattended operation in May 2010 using a partial installation containing 196 transceiver modules and sequences of experiments adapted from ALWIN operational settings for mesospheric and tropospheric observation. Especially the tropospheric experiments were used to validate the overall performance of MAARSY by comparing wind velocities obtained with different radar experiments and data analysis procedures as well as with winds measured by radiosondes. Figure 14 shows scatterplots of horizontal wind components from MAARSY DBS experiments versus radiosonde observations. The data were obtained during a campaign at the Andøya Rocket Range in July 2010 where in total 18 radiosondes were launched. The correlation coefficients of r = 0.83 (zonal) and r = 0.76 (meridional) indicate a good agreement in both wind components obtained with MAARSY and the radiosondes. A more detailed validation regarding accuracy and quality of wind measurements obtained in the troposphere and mesosphere using various operational modes and analysis techniques is presented in detail in a companion paper by G. Stober et al. (manuscript in preparation, 2011).

Figure 14.

Scatterplots of horizontal wind components from MAARSY DBS experiments versa radiosonde observations (G. Stober et al., manuscript in preparation, 2011).

3.2. PMSE Observation During Summer 2011

[72] Since one focus of IAP radar soundings is the mesopause region the majority of observation time was given to DBS and SA experiments covering a range between 50 km and 114 km. 8-bit complementary coded pulses with a subpulse length of 2 μs were transmitted using a vertical beam and a PRF of 1250 Hz. The received signals were sampled with 300 m resolution and added 16 times coherently. Since the data acquisition unit used with the 2010 partial expansion stages of MAARSY was capable to sample 8 dual channels only, a multiplexer was used to switch the 16 dual outputs of the base band down converters to the 8 dual inputs of the digitizers. This halved the resolution of the time series per range gate to a value of 25.6 ms corresponding to a Nyquist frequency of 19.53 Hz. The completion of the installation to 433 transceiver modules in spring 2011 included also an upgrade to a real 16 channel data acquisition system. Hence the number of coherent integrations used with the mesospheric experiments in 2011 were doubled to 32 in order to improve the SNR while keeping the time resolution.

[73] The diurnal and seasonal variations of the 2011 PMSE observations as shown in Figure 15 are characterized by a typical behavior comparable to results obtained with the Alomar SOUSY radar [Hoffmann et al., 1999] and the ALWIN radar [Latteck et al., 2008; Bremer et al., 2009] on Andøya during the last 15 years. The diurnal variation of PMSE in 2011 reaches a maximum between 11:00 and 16:00 LT and a minimum between 18:00 and 22:00 LT. The mean altitude occurrence of the 2011 PMSE altitude distribution peaks at 85 km and is comparable to the mean value of 84.9 km [Latteck et al., 2008]. The first PMSE of the 2011 season was observed on 7 May and the mean value of daily occurrence reached ∼37% in May 2011 which is twice the value of the previous monthly mean of May. In contrast to the mean rise of the PMSE occurrence as reported by Bremer et al. [2009] and indicated by a dashed line in Figure 16 the 2011 PMSE occurrence reaches a maximum close to 100% within a few day at the end of May. A similar behavior was observed at the end of the 2011 season when the PMSE occurrence dropped within a few days as well; much more rapidly than the mean slope of −2.5%/day as reported by Bremer et al. [2009]. During the PMSE core period in June/July the mean daily occurrence rate of the 2011 season exceeded the mean value from previous observations by ≈18% to reach about 97%. A comparison between the daily occurrence rates observed in 2011 with MAARSY and the seasonal mean occurrence of PMSE based on 9 years of observations with ALWIN at the same location is depicted in Figure 16. The comparison is based on a threshold of 10 dB above the signal detection limits of both systems defined by the power in arbitrary units of the smallest signal detected.

Figure 15.

Seasonal and diurnal occurrence as well as altitude distribution of PMSE observed with MAARSY in 2011.

Figure 16.

Seasonal variation of PMSE occurrences rates over Andøya. The 2011 rate derived from MAARSY observations (gray shaded bars) is overlayed by a mean rate (blue curve) based on 9 years of observations from ALWIN. The dashed lines indicate the rise (4%/day) and slope (−2.5%/day) of the increasing and decreasing phase respectively as reported by Bremer et al. [2009].

3.3. Three-Dimensional Resolved Structures of PMSE

[74] MAARSY was designed for middle atmosphere studies, especially the investigation of horizontal structures of mesospheric echoes such as PMSE and PMWE. As already mentioned in section 2.4.1 the PSC stack of the transceivers can store about 50 parameter settings for e.g. phase offsets for beam pointing, and the parameters can be changed with the PRF. In order to get a horizontal coverage as wide as possible in combination with a time resolution per beam direction as small as feasible a sequence of experiments each configured for different beam directions was used to perform first horizontal scans of PMSE in a quasi-simultaneous mode in 2011. Four identical experiments as described in Table 3 with 24 different oblique beam directions each and the vertical beam position were used. The overall 97 beam positions are tagged to ovals indicating the illuminated areas at 84 km and overlayed to a contour plot of SNR in Figure 17 (right). The experiments changing the beam position after every 4 coherent integrations, ran for 21 s one after another and illuminated an area of about 80 km diameter at an altitude range between 50 km and 103 km. The four scanning experiments were followed by a standard mesospheric experiment running for 27 s and covering a wider range in vertical direction with a larger number of coherent integrations. Operating MAARSY in experiment sequence mode requires additional time between the experiments for reconfiguration of the hardware. This resulted in an overall sequence runtime of 334 s corresponding to the time resolution of the scan. A single experiment operation using 50 beam directions as described in section 2.4.1 can reduce the time resolution of the scan depending on the required number of data points per beam direction, e.g. 41 s for 512 data points per beam.

Table 3. Parameters of MAARSY Experiments Used for Horizontal Scanning of PMSE in July/August 2011a
Experiments for PMSEScanningMonitoring
  • a

    Four experiments with 24 different beam directions each and a vertical beam were used in a sequence for PMSE observation. The sequence was completed by a standard PMSE experiment using the vertical beam only.

Pulse repetition frequency1250 Hz1250 Hz
Number of coherent integrations432
Wave form (pulse coding)8 bit coco8 bit coco
Subpulse length2 μs2 μs
Pulse length16 μs16 μs
→ Inter pulse period800 μs800 μs
→ Duty cycle2%2%
Sampling start range50.1 km50.1 km
Sampling end range114.3 km114.3 km
Sampling resolution300 m300 m
→ range gates215215
Number of data points p. exp.64001024
Number of beam directions251
→ Number of data points. p. beam2561024
→ Time resolution Δt80.0 ms25.6 ms
→ Nyquist frequency6.25 Hz19.53 Hz
→ Experiment runtime20.5 s26.2 s
Number of experiments41
Figure 17.

Vertical and horizontal structure of a PMSE observed with 97 beams quasi-simultaneously on 22 July 2011. (top left) The SNR determined from observations using 11 beams along the west–east direction. (bottom left) The same situation but interpolated to a grid with 100 m horizontal and vertical resolution. (right) A slice at 84 km height through the 3-dimensional interpolated data obtained from all 97 beams. The tagged ovals overlayed to the contour plot mark the areas at 84 km altitude illuminated by the beams.

[75] MAARSY was operated with the described experiment sequence from 11 July until 12 August 2011. Figure 17 (top left) shows an example of a height-distance-intensity plot of signal-to-noise ratio obtained on 22 July with 11 beams along the west–east direction. The color coded areas of SNR represent vertical slices of the radar pulse volumes illuminated by the 11 beams with 300 m radial resolution. In order to get horizontal and vertical resolved 2-dimensional and 3-dimensional maps of PMSE the range information of the profiles averaged by 15 minutes were first converted to altitude using the nominal off-zenith angle of the corresponding beams. The range corrected profiles were then interpolated to a vertical grid with a resolution Δz = 100 m and the resulting altitude values were converted back to ranges. The new polar coordinates of the interpolated data were then converted to cartesian coordinates and finally interpolated to a 2-dimensional horizontal grid with a equal resolution Δx = Δy = 100 m. Figure 17 (bottom left) shows the resulting 2-dimensional vertical slice of SNR corresponding to the example shown in Figure 17 (top left).

[76] We note that this simple gridding procedure does not take into account the following points: (1) A potentially inhomogeneous distribution of the echo power inside a given range gate. This might be particularly important at large zenith angles where the vertical coverage of a range gate will rather be dominated by the beam width than by the length of the radar pulse. (2) The variation of transmitted power over the angular spread of the radar beam (i.e., in this case 3 dB variation over 4°). (3) The potential aspect sensitivity of the observed echoes. A proper consideration of these effects will require the combination of this type of experiment with interferometry in both the spatial as well as in the frequency domain [e.g., Woodman, 1997] and will be left for future studies.

[77] The horizontal slice through the PMSE at 84 km with a resolution of 100 m depicted in Figure 17 (right) shows pattern of SNR with a variation of about 40 dB within a distance of 80 km. A stack of horizontal slices at different altitudes are presented in Figure 18. The 3-dimensional resolved PMSE shows a maximum or core located north-westward of the radar location. This indicates that PMSE or the underlying structures and processes reveal a pronounced spatial variability that asks for in-depth studies in the future.

Figure 18.

Three-dimensional structure of a PMSE observed with 97 beams quasi-simultaneously on 22 July 2011.

3.4. Meteor Head Echo Observations

[78] The second stage of expansion of the MAARSY installation increased the available power-aperture product and allowed first meteor head echo observations. MAARSY was configured for meteor head echo observation during the daily peak activity of the Geminids meteor shower in December 2010 using a sequence of three different experiments. Table 4 shows the basic parameters of the experiment configuration. Limited by the data rate of the signal processing unit and the maximum duty cycle of the radar hardware the experiments were tuned to cover a range between approximately 70 km and 150 km with a sampling rate between 2 μs and 6 μs. The vertical pointing experiments “meteor001” and “meteor003” transmitted long pulses with coded and noncoded waveforms to improve the estimation of Doppler velocities. An off-zenith pointing experiment “meteor002” using double-pulses of 2 μs length separated by 10 μs were directed towards Geminids radiant around its local maximum and the beam position was adjusted every 30 minutes.

Table 4. Configuration of MAARSY Experiments Used Meteor Head Echo Observation During ECOMA Campaigns in December 2010
PRF595 Hz320 Hz700 Hz
Number of CI111
Wave form7 bit Barkerdoublemono
Subpulse length12 μs2 μs 
Pulse length84 μs14 μs48 μs
→ IPP1.68 ms3.12 ms1.43 ms
→ Duty cycle5.00%0.13%3.36%
Sampling start range72.9 km85.3 km72.9 km
Sampling end range148.5 km134.4 km148.5 km
Sampling resolution900 m300 m900 m
→ Sampling rate6 μs2 μs6 μs
→ Range gates8516585
Number of data points819281928192
→ Time resolution Δt1.68 ms3.12 ms1.43 ms

[79] Figure 19 (top) shows an example of a meteor head echo as range-time intensity plot. Figure 19 (bottom) shows the ranges where the power is maximum for each profile. The radial velocity was derived from a fitted line though these values as described, e.g., by Chau and Woodman [2004]. An in-depth analysis of these observations is currently ongoing and will be presented in a future publication.

Figure 19.

(top) Range-time intensity plot and (bottom) range of maximum power of a meteor head echo observed on 13 December 2010 at 05:08:32 UT. The radial velocity was estimated by fitting a line though these values.

4. Summary and Outlook

[80] The present paper gives an overview about the design and describes the functionality of the basic components of the Middle Atmosphere Alomar Radar System (MAARSY) installed on the North-Norwegian island Andøya during the years 2009 till 2011. After the construction of the antenna array in summer 2009 and the installation of an initial expansion stage of 196 transceiver modules the radar operation started in May 2010 with standard observations of tropospheric winds and Polar Mesosphere Summer Echoes. A second stage of expansion to 343 transceiver modules was brought into service in November 2010 and the system was finally upgraded to 433 transceiver modules in May 2011.

[81] One of the main objectives for building MAARSY, the investigation of horizontal structures of Polar Mesosphere Echoes, could be tested using multi-beam experiments with up to 97 beams quasi-simultaneously during campaigns in 2010 and 2011. The presented first example of horizontally resolved structures of Polar Mesosphere Summer Echoes demonstrate the performance of MAARSY and gives a first insight into the three dimensional structure and variability of mesospheric echoes such as PMSE and PMWE.

[82] We also note that MAARSY has the capability for ionospheric incoherent scatter observations after a planned upgrade of the MAARSY antenna array to circular polarization. Additional 433 Yagi antennas will be installed perpendicular to the existing antennas and both antenna elements will be fed by one transceiver. The necessary power splitting and phase shifting between transmission and reception will be arranged by an passive switch mounted on the antenna pole. The peak power of about 800 kW corresponding to an average power of 40 kW at a duty cycle of 5% and the transmission of long pulses up to 200 μs will allow ionospheric incoherent scatter observations with the much lower signal-to-noise ratio compared to the backscatter from irregularities in the mesosphere, stratosphere, and troposphere. Corresponding power profile measurements should be feasible with reasonable height resolution applying pulse compression with Barker codes or complementary codes. In addition multi-pulse transmissions can be used for spectral measurements to observe drift motions. First test measurements are planned for late 2012.

[83] Finally, another plan is to upgrade the number of signal processing channels from sixteen to the total number of antenna subgroups. The sampling of the 61 IF feds could be done in the main building as well as in the transceiver building as the existing infrastructure contains a multichannel fiber network. This upgrade will allow to form e.g. larger spaced antenna structures with shorter distances.


[84] The authors would like to thank S. Fukao, T. Sato, and M. Yamamoto for their suggestions and discussions in the early stage of planing the radar. We also thank the IAP personal who worked hard with the installation of MAARSY, in particular, JörgTrautner, Thomas Barth, Jens Wedrich, Norbert Engler, Dieter Keuer, Hans-Jürgen Heckl, Torsten Köpnick, Manja Placke, and Qiang Li, as well as the students Ding Tao, Gunnar Keuer, Christian Schernus, Sophie Latteck, Danilo Hauch, and Richard Hünerjäger. We are indebted to the staff of the Andøya Rocket Range for their permanent support. The radar development was supported by grant 01 LP 0802A of Bundesministerium für Bildung und Forschung.