Abstract

The German Meteorological Service [Deutscher Wetterdienst (DWD)] operates the German weather radar network and is currently replacing all radar systems with new dual-polarization radars. One of the key components of a radar system is its antenna. The quality of the dual-polarized (dual-pol) moments is dependent on the quality of the antenna and its proper characterization. Dedicated on-site antenna pattern measurements with and without radome are performed in order to verify the antenna and radome specification of the new Hohenpeissenberg radar. Those on-site measurements are carried out from three different source sites in the far field of the antenna. The pattern measurements show that the main antenna specifications, such as the 3-dB beamwidth, beam squint, and sidelobe levels, are met. This is also true if a radome is in place. However, the data suggest that the manufacturer’s specifications are not fulfilled completely. This study finds a substantial increase of the copolar power level off the main beam. The homogeneity of the main beam with respect to differential power and phase is degraded with the radome. The measurements based on the three source sites indicate that a small (but negligible) radome-induced azimuthal variability of the radar moments can be expected. The integrated cross-polarization ratio based on the pattern measurement suggests that 1.4 dB of sensitivity is lost in the linear depolarization ratio (LDR) due to the dry radome.

1. Introduction

Weather radars with polarization diversity have become a mature technology in recent years. Meteorological services worldwide started to introduce this capability in their radar networks. Commonly, simultaneous transmit and receive (STAR)-mode systems are the choice. With the introduction of dual polarization, systems improved quantitative precipitation estimates and a reliable classification of meteorological and nonmeteorological targets are expected. This, however, requires appropriate quality of the basic polarimetric variables. Aside from calibration and monitoring aspects, there are components of the radar system that have a strong influence on the resulting dual-polarized (dual-pol) data quality. One of the key components of a radar system is the antenna (e.g., Chandrasekar and Keeler1993; Zrnić et al. 2010; Bringi et al. 2011). In particular, the quality of dual-pol variables is sensitive to the quality of the antenna and its proper characterization. Furthermore, the influence of radome on the antenna performance and the resulting radar variables should be small. Ideally, this must hold under wet and dry radome surface conditions.

The German Meteorological Service [Deutscher Wetterdienst (DWD)] is currently replacing all radar systems of the German weather radar network with new dual-polarization radars. Commonly, the compliance with the specifications of the antenna is proven through antenna patterns that are usually provided by the antenna manufacturer as cuts through the main planes of the antenna, including the strut plane. In the course of our antenna acceptance tests, it appeared that, for example, the proof of the match between the main beam between the polarizations may be limited by the existing equipment on the antenna manufacturer’s test range. In particular, the mechanical antenna’s pointing accuracy at the antenna manufacturer’s site appeared to be not sufficient to prove the compliance with our specifications. Therefore, the antenna of every new radar system is tested on the radar manufacturer’s antenna test range. In doing so, we make use of the software and hardware capabilities of the new radar system. The antenna tests are carried out during the factory acceptance tests (FATs) prior to the delivery of the radar system to DWD.

After the installation of a new radar system, we perform on-site antenna tests as part of the site acceptance tests (SATs) for a number of antennas. On-site antenna tests are a unique effort as part of DWD’s radar replacement project. They have two main goals. One goal is to prove that the on-site antenna assembly procedures guarantee the same antenna performance as shown during FATs. The second goal is related to the combined performance of the antenna and the radome, which in the end determine the data quality of the new dual-pol radar system.

The radome is an important protection component of the radar system, as it guarantees a high availability of the radar system under bad weather conditions. However, it must not affect the dual-pol data quality in a significant way. For example, previous studies have shown azimuthal dependencies of differential moments due to the radome design (e.g., Gourley et al. 2006). The importance of a high-quality hydrophobic radome surface coating is well known (Kurri and Huuskonen 2008), but it becomes more prominent for dual-pol systems. For example, differential attenuation as a function of rain intensity can cause significant bias in differential reflectivity (ZDR; Frech 2009). Based on antenna patterns of the WSR-88D radars without and with a partially assembled radome, it has been suggested that a radome may increase the sidelobe levels by about 2 dB (Doviak et al. 2000).

In an effort to verify the radome performance, dedicated antenna pattern measurements with and without radome were carried out at the Hohenpeissenberg Meteorological Observatory in spring 2011. The installed radome has a panel design that is optimized for dual-pol applications aimed at electrically seamless radio frequency (RF) performance. The layout of the radome is based on an impedance matching procedure that comprises laboratory measurements of the scattered electromagnetic field due to the flanges of the radome panels and the panels themselves.

In the following, we first give a brief overview of the radar system and discuss the antenna and radome specifications. We then describe the approach to verify parts of the specifications. In the main body of this paper, we discuss the results from the measurements before we summarize the main findings.

2. Description of the EEC radar DWSR5001C/SDP/CE

DWD is operating the national radar network. The network consists of 16 C-band radar systems, which cover most of the German territory. An additional system serves as a research radar that is operated at the Hohenpeissenberg Observatory. Here, new technologies, radar data processing algorithms, radar software, and new products are developed, tested, and evaluated before they are introduced into operational service. Within DWD’s radar system replacement project Radar-System-Ersatz (RadSys-E), all radar systems are replaced. In addition, a new site is realized in southern Germany near Memmingen to achieve a better coverage of the Alps in this region. A number of existing radar sites will be relocated to new sites in order to optimize the coverage and the operation of the weather radar network. The complete replacement of all radar systems is scheduled to be finished in 2014. In the end, the operational network will consist of 17 operational radars and one research radar. The old radar systems are replaced with Enterprise Electronics Corporation’s (EEC) Doppler weather radar DWSR5001C/SDP/CE, which has polarization diversity [simultaneous dual polarization (SDP)]. General specifications of the new system are summarized in Table 1.

Table 1.

Overview DWSR5001C/SDP/CE specifications.

Overview DWSR5001C/SDP/CE specifications.
Overview DWSR5001C/SDP/CE specifications.

a. Antenna requirements

For dual-pol application, a good match of antenna characteristics in the two polarization planes horizontal (H) and vertical (V) is a prerequisite for a good performance of the radar system. In particular, the main beams in H and V need to coincide as closely as possible. Furthermore, we require low sidelobe levels in all planes, including the struts. Low sidelobe levels are important in the presence of high spatial reflectivity gradients and low reflectivity situations, where an antenna sidelobe may degrade data quality. The feed of the antenna is supported by four struts. The struts are positioned so that they cause the same blockage to both H and V polarizations. In the strut plane, they cause an increase in sidelobe levels.

The specifications for the new dual-pol antenna are summarized in Table 2.

Table 2.

Antenna specifications.

Antenna specifications.
Antenna specifications.

Those specifications must be satisfied in the whole frequency range in which the radar systems are operated and for all planes of the antenna pattern. The antenna manufacturer is required to prove those specifications by dedicated antenna pattern measurements at its test range near Boston, Massachusetts. They are part of the FAT. In the process of the acceptance test, it was decided to carry out additional antenna tests at EEC’s test range in Alabama. The primary advantage of this approach is that we can make use of the capabilities of the new radar system. In particular, the pointing accuracy of the pedestal unit and the use of the state-of-the-art receiver and signal processor allows for a more detailed characterization of the antenna compared to those of the antenna manufacturer’s facilities. For a selected number of antennas, additional antenna tests are required after the installation of the radar system as part of the site acceptance test.

b. Radome requirements

As mentioned before, the data quality of dual-pol variables may be influenced by the radome. The influence may be due to the basic radome design such that, for example, the orientation of the radome panel flanges affects the propagation of microwaves with a given polarization more than the other polarization. This may cause a systematic azimuthal dependence of the dual-pol variables (see, e.g., Gourley et al. 2006). Another important aspect is the surface coating of the radome. It is clear that a reliable hydrophobic coating is essential to minimize any bias of the dual-pol variables in case of precipitation at the radar site. Differential attenuation can be expected if the hydrophobicity of the radome surface is of bad quality (e.g., Frech 2009).

Therefore, rather strict requirements were established in order to minimize data quality aspects due to a radome. The specifications are summarized in Table 3.

Table 3.

Radome specifications.

Radome specifications.
Radome specifications.

Some of the specifications will be difficult to prove in field. Our approach to verify some of the specifications is given in the next section. According to the manufacturer, the installed radome has a one-way (dry) attenuation transmission loss of 0.27 dB.

3. The measurement campaign

In this section, we describe the setup and the scan strategy of the antenna pattern measurement.

a. Antenna pattern measurements

The quality of antenna pattern measurements, aside from the test equipment, is very much dependent on the test range. The basic approach to obtain an antenna pattern is to place a transmitter in the far field of the receiving radar antenna. The location of the transmitter has to be chosen such that there are no obstacles in the path. Furthermore, the site has to be selected such that multipath propagation is avoided as much as possible. A favorable source site may be a tall tower or a mountain with, for example, a valley along the transmit path. Compared to typical antenna test ranges, we have the new radar system with its precise positioning system and state-of-the-art radar receiver and signal processing available. This allows for acquiring high-resolution volume data during the pattern measurements.

Some of the radome requirements refer to limits on azimuthal variations of dual-pol moments. To investigate this aspect, three source sites at different azimuthal locations relative to the radar were chosen. The working hypothesis is that high-resolution pattern measurements through different portions of the radome yield at least some impression of the azimuthal variability of dual-pol variables and of the influence of the radome on basic antenna characteristics, such as beamwidth.

In total three external transmit sites are used for the Hohenpeissenberg test: Auerberg, Bromberg, and the Hohenpeissenberg television (TV) tower. This provides pattern data from essentially three different test ranges. Some basic information about the three sites is summarized in Table 4.

Table 4.

Basic information about the transmitter sites Auerberg and Bromberg, and the TV tower. The radar antenna is at a height of 1000.6 m MSL.

Basic information about the transmitter sites Auerberg and Bromberg, and the TV tower. The radar antenna is at a height of 1000.6 m MSL.
Basic information about the transmitter sites Auerberg and Bromberg, and the TV tower. The radar antenna is at a height of 1000.6 m MSL.

The transmitter dish has a diameter of 150 cm (the 3-dB beamwidth is 2.4°, the first sidelobe is at −26 dB below the main peak). A test signal generator provides the continuous wave (CW) test signal at the nominal radar frequency of 5.64 GHz.

At the radar site, a standard gain horn provided an independent monitoring of the received power. Overall, the received signal was very stable and showed only small variations during the scan (±0.1 dBm). This indicates a good, stable signal source and favorable meteorological conditions during the measurements.

b. Scanning

Dedicated antenna scans were defined using DWD’s operational radar software MURAN 4.1 with the following settings in STAR mode:

  • “High” resolution volume with elevation angle steps of Δel = 0.1° close to the main beam and coarser elevation steps starting at 5° with Δel = 0.5 − 1° up to 30° elevation.

    • Typically, 117 sweeps are recorded, starting from el = −2° to el = 30°. Every 10 sweeps, one sweep is carried out at the elevation of maximum signal strength. This allows for monitoring the signal source during the scan.

    • Sampling rate is set using a PRF = 3000 Hz and an azimuth rate of 6° s−1.

    • Raw range bin resolution: 25 m, output range bin averaged to 1 km, and maximum range = 20 km.

    • Pulse width: 0.4 μs.

    • Dynamic angle synching (DAS) with 0.05° ray width (oversampling in azimuth).

    • The following variables are recorded: the signal-to-noise ratios SNRh and SNRv, the differential phase Φdp, and the uncorrected differential reflectivity UZDR.

  • Range–height indicator (RHI) scans: in order to sample a larger range of the 90° plane, we carried out a number of RHI sweeps.

    • Typically five RHIs are recorded starting from el = −2° to el = 90° each.

    • Sampling rate is set using a PRF = 3000 Hz, elevation rate = 6° s−1, and range = 20 km.

    • Raw range bin resolution: 25 m and range averaging of 1 km

    • Pulse width: 0.4 μs

    • DAS with 0.05° ray width (oversampling in elevation).

The radar is not radiating during the scans but tuned to the transmitting frequency. One volume scan and a set of usually four RHI scans for one polarization are acquired in about 2 ½ h.

In total three transmit modes where used: H only, V only, and V and H simultaneously. The latter is referred to as the STAR mode, where the source transmits linearly polarized signals in H and V at constant phase difference. In STAR mode we can measure antenna patterns of differential phase and power. When transmitting in single-polarized (single-pol) mode (H or V only), the transmitted cross-polar signal is minimized manually by adjusting the feed before starting the volume scan. In general, a feed dish assembly never shows a perfect cross isolation, so that the received cross-polar signal of the antenna pattern is at least in part related to the limited cross isolation of the transmitting antenna and feed.

We had very persistent weather conditions during SATs with and without radome. The measurement dates and the corresponding temperature ranges during the tests are summarized in Table 5.

Table 5.

Temperature ranges during the antenna tests. Shown are source sites, the transmit polarization (H-V-STAR), date, duration, minimum, median (med), and maximum temperature during the test.

Temperature ranges during the antenna tests. Shown are source sites, the transmit polarization (H-V-STAR), date, duration, minimum, median (med), and maximum temperature during the test.
Temperature ranges during the antenna tests. Shown are source sites, the transmit polarization (H-V-STAR), date, duration, minimum, median (med), and maximum temperature during the test.

4. Results

a. Outline of the analysis

We focus on the difference between the measurements with and without radome. To obtain a generalized view on the antenna performance and the radome, we generate composites centered around the main lobe from all measurements and show average cuts through the main planes and their variability from measurement to measurement and from site to site. The best sampling is available for the 0° azimuthal plane, where we have 20 sweeps for each polarization and particular source site. So, in total 60 sweeps (three source sites) through the main plane are available for copolar measurements both with and without a radome.

b. Antenna patterns with and without radome

We first show some examples of antenna patterns for a specific source site with and without radome in order to illustrate the main features of the radome influence on the antenna pattern.

The two examples show the copolar signal received from the TV tower site. Data are normalized relative to the peak of the main beam. Clearly visible are the pencil-shaped main beam of the antenna and the influence of the struts (Fig. 1). Qualitatively, there is an increase in copolar signal strength in some locations to levels above −43 dB, but the difference between the measurements with and without radome seems not significant. Figure 2 shows the corresponding cross-polar HV signal (transmit H, receive V). It is obvious that the cross-polar levels are increased significantly by the radome. This increase is mainly seen in between the struts up to levels of about −50 dB. There seems to be no significant increase of cross-polar power in the strut plane and in the main beam region. The typical four cross-polar peaks located around the center of the main beam are nicely visible (Fig. 2; Zrnić et al. 2010). The radome seems to equalize those cross-polar lobes compared to the pattern without radome.

Fig. 1.

Copolar H plot, and source site is the TV tower. Data are taken (top) without and (bottom) with radome.

Fig. 1.

Copolar H plot, and source site is the TV tower. Data are taken (top) without and (bottom) with radome.

Fig. 2.

As in Fig. 1, but for the cross-polar HV plot.

Fig. 2.

As in Fig. 1, but for the cross-polar HV plot.

As mentioned before, these figures provide a first qualitative view of the main features of the measured antenna patterns. The antenna patterns shown here characterize only the bottom half of the dish, and it is assumed that the top half of the dish has similar characteristics because of symmetry. In the following, we make a more quantitative analysis, focusing on cuts through selected planes of the antenna diagram.

c. Slices through the main axes and the strut plane

We now investigate slices through the main axis and later on slices through the strut planes (±45°). The main axis refers to an azimuth plane through the main beam (“0°” plane). We first show a comparison of the results from the three source sites without radome. The focus here is to investigate the reproducibility of antenna characteristics from different datasets taken during FAT and SAT. Possible differences in the patterns from different source sites may be attributed to the test range conditions or to a misalignment of the antenna after the on-site reassembly of the dish (if we compare the FAT and SAT data). In general, the results for the copolar measurements in the main beam area are comparable from site to site. For example, the first sidelobes match very well in location and level (Fig. 3). Furthermore, the location of the sidelobes from the on-site measurements matches well with that from FAT. The FAT sidelobe levels tend to differ from the on-site measurements at Hohenpeissenberg. We attribute this to less favorable test range conditions during FAT. We notice also an asymmetry in the first sidelobe level in H that might be due to lateral feed misalignment (Skolnik 2008). The increased filling of the first null in the copolar H pattern compared to the V pattern may point to different blocking effects due to the feed.

Fig. 3.

Slices through the 3D pattern. Shown are copolar results through the main axes at 0° from measurements during FAT and SAT. We compare results from FAT at EEC, and SAT results from the Bromberg (BROM), Auerberg (AUER), and the TV tower source site with a focus on the main beam area, showing (top) H polarization and (bottom) V polarization. The black envelope denotes the specification.

Fig. 3.

Slices through the 3D pattern. Shown are copolar results through the main axes at 0° from measurements during FAT and SAT. We compare results from FAT at EEC, and SAT results from the Bromberg (BROM), Auerberg (AUER), and the TV tower source site with a focus on the main beam area, showing (top) H polarization and (bottom) V polarization. The black envelope denotes the specification.

The cross-polar measurements show much larger scatter between the four test ranges (Fig. 4). We speculate that this primarily can be attributed to test range effects where part of the transmitted signal may be scattered at obstacles along the propagation path. In addition, slight misalignments of the transmit antenna feed will affect the cross-polar measurements much more than the copolar measurements. In particular, the measurements from the Bromberg site (located slightly below the radar site) show relatively large cross-polar levels.

Fig. 4.

As in Fig. 3, but for cross-polar results.

Fig. 4.

As in Fig. 3, but for cross-polar results.

We now average all slices through the main planes (0° and ±45°). This is done for the tests with and without radome separately and includes the STAR measurements. When averaging the data, we also compute the first and third quartiles of the data. This will serve as an indication of the antenna pattern variability as a function of azimuth angle. Radome effects on the antenna pattern can then be quantified. Results from the RHI scans in the 90° plane are not shown, as they do not add additional information.

The average copolar results in H and V through the main plane (0°) show no significant differences between measurements with and without radome in the main beam area, including the first sidelobe (Fig. 5, top panel). There is a significant increase in sidelobe levels beyond the first sidelobes compared to the measurements without radome. The typical roll-off of the sidelobes is not seen and the copolar signals remain, on average, on a constant level. The degradation in far sidelobe levels is approximately 13 dB compared to the antenna pattern without radome. At some azimuthal positions, copolar levels are above the specified −43-dB level (Fig. 5, bottom panel). From an operational perspective, the levels of the first sidelobes are important, as they determine the level of clutter echos at low elevations. Table 6 summarizes the average first sidelobe levels in the main plane. Without radome, the antenna fails to achieve the specified maximum sidelobe level of −30 by 0.4 dB in H. The other sidelobe levels are well within the specifications for the data with and without radome.

Fig. 5.

Average copolar power H and V, with and without radome. The average is based on 60 sweeps (single-pol and STAR mode tests). The black envelope denotes the specification.

Fig. 5.

Average copolar power H and V, with and without radome. The average is based on 60 sweeps (single-pol and STAR mode tests). The black envelope denotes the specification.

Table 6.

The average first sidelobe levels in H and V for the measurements with and without radome. Power levels are relative to the main beam.

The average first sidelobe levels in H and V for the measurements with and without radome. Power levels are relative to the main beam.
The average first sidelobe levels in H and V for the measurements with and without radome. Power levels are relative to the main beam.

Based on the specification, the first sidelobe must not be increased by >0.5 dB because of the radome. Here, in all but one case, this is achieved. In V, we find an increase of 0.9 dB (from −34.3 to −33.4 dB). This might be a measurement artifact, but the large number of measurements and the small variability from measurement to measurement suggest an effect related to the radome. The standard deviation of the average first sidelobe levels is in both cases 0.2 dB. However, the overall performance of the antenna–radome assembly with respect to the influence on the copolar antenna pattern in the main beam area is still very good. The main beams match within less than ≈1 dB down to −30 dB for a given polarization state (Fig. 6). In the main plane, the main beam shape is not affected by the radome in a significant way. This aspect will be revisited below. Larger differences outside the main beam relate to different sidelobe locations for measurements with and without radome.

Fig. 6.

Average copolar power (top) H and (bottom) V, with and without radome, together with the difference between the two sets of measurements.

Fig. 6.

Average copolar power (top) H and (bottom) V, with and without radome, together with the difference between the two sets of measurements.

The corresponding cross-polar slices are shown in Fig. 7. In the main beam, there is no significant difference between measurements with and without radome. The cross-polar signals are about 10 dB larger for azimuth angles >±10°. However, on average the cross-polar levels are within the specifications.

Fig. 7.

Average cross-polar power HV and VH, with and without radome. The average is based on 40 sweeps (single-pol tests). The black envelope denotes the specification.

Fig. 7.

Average cross-polar power HV and VH, with and without radome. The average is based on 40 sweeps (single-pol tests). The black envelope denotes the specification.

Similar to the results for the main azimuth plane, we now show the copolar sidelobe levels in the strut planes (±45°). The mean copolar level in the strut plane is based on an average over all three source sites, including both strut planes (Fig. 8). The number of samples is much smaller than for the main plane, in particular for elevation angles >5°. Overall, the difference between the results with and without radome appears small in the strut plane. The blockage and scattering effects due to the struts dominate. The first sidelobe in both cases is between −27 and −28 dB for both H and V. Up to 10°, the sidelobes remain at an almost constant level of around −33 dB. This has to be compared to the main plane results, where we see a pronounced roll-off of the sidelobe levels down to −50 dB or less for measurements without radome and −40 dB for measurements with radome (see Fig. 5). From radial angles >10° on, we see a steady decrease in sidelobe levels. The specification are satisfied for angles of about >20°. Overall, the copolar results in the strut plane are dominated by the presence of the struts. An influence of the radome cannot be seen.

Fig. 8.

Average copolar power H and V in the strut plane, with and without radome. The black envelope denotes the specification.

Fig. 8.

Average copolar power H and V in the strut plane, with and without radome. The black envelope denotes the specification.

The cross-polar results in the strut planes are shown in Fig. 9. A double peak at about ±0.8° can be seen where we cut through the typical cross-polar antenna pattern of a center-fed antenna assembly (see also Fig. 2). On average, the cross-polar levels in HV are within the specification, whereas the peaks in VH (transmit V, receive H) are above the specification (near ≈ −30 dB). The first cross-polar peaks, HV and VH, are raised because of the presence of a radome by up to 5 dB.

Fig. 9.

Average cross-polar power HV and VH in the strut plane, with and without radome. The black envelope denotes the specification.

Fig. 9.

Average cross-polar power HV and VH in the strut plane, with and without radome. The black envelope denotes the specification.

From the measured copolar and cross-polar antenna pattern, we can compute the integrated cross-polarization ratio (ICPR). It determines the lower limit of the linear depolarization ratio (LDR), which can be measured with the given antenna characteristics (Chandrasekar and Keeler 1993; Bringi and Chandrasekar 2001). Following Bringi and Chandrasekar (2001), an upper bound ICPR can be written as

 
formula

with the copolar and cross-polar power patterns fhfv, fhv, and fvh and the elemental solid angle dΩ. Using the pattern measurements based on the TV tower source site, the upper bound of ICPR amounts to −31.5 dB without radome. This includes the full volume dataset, including the struts. The upper bound of ICPR amounts to −30.1 dB with the radome. So, based on the antenna performance with radome, we lose 1.4 dB of sensitivity in LDR. In the ICPR calculation, equal gain in H and V is assumed.

d. Measurements of differential phase and differential power

In this section we focus on the match of the phase and power patterns. A good match of the phase patterns is important to obtain good quality phase-dependent radar moments, such as the differential phase Φdp and the cross-correlation coefficient ρhv. Especially across the main beam, where most of the energy is located, variations of differential phase and power should be small. A variability of the differential measurements would limit the accuracy of radar moments under real weather conditions. However, some variability might be expected because of the presence of struts and errors in feed alignment (Mudukutore et al. 1995). As illustrating examples, we show the spatial distribution of differential phase and power in the main beam areas for the measurements without radome (the figures with radome are comparable). The differential phase distribution is shown in Fig. 10. The corresponding differential power measurement is shown in Fig. 11. In the main beam area, the observed patterns for measurements with and without radome show a similar pattern. Larger differences are found closer to the border of the main beam. These differences will be quantified in the following. Large differences outside the main beam area are usually related to low SNR values.

Fig. 10.

Differential phase (top) without and (bottom) with the radome in place. Source site is the TV tower.

Fig. 10.

Differential phase (top) without and (bottom) with the radome in place. Source site is the TV tower.

Fig. 11.

Differential power (top) without and (bottom) with the radome in place. Shown are the median and the first and third quartile values. Source site is the TV tower.

Fig. 11.

Differential power (top) without and (bottom) with the radome in place. Shown are the median and the first and third quartile values. Source site is the TV tower.

For a more detailed analysis, we compute the phase difference as function of radial distance. The center or zero radial distance is defined as the location of the SNR peak power. For a given radial distance interval, we compute the statistics of the data from all angles. This is done individually for all data with and without radome. The results based on the three source sites are then averaged to obtain the overall distribution of the phase difference in the main beam area. For the differential phase, the results are shown in Fig. 12. Up to about a radial distance of 1°, the variations of differential phase are small for the measurements without radome. At about 1° we find a mean difference of −2° relative to the peak value in the main beam center. This corresponds to SNR values of about −15 dB below the main peak. Differences start to increase from about r = 0.7° on, which also coincides with increased variability as seen in the first and third quartile values (Fig. 12, bottom panel). Up to this value, the phase difference is quasi constant over the main beam. On average, the measurements with radome show a somewhat larger increase in phase difference in the main beam area. Also, the first and third quartiles indicate a larger range, suggesting an increased variability in differential phase due to the radome. The largest variability in differential phase is found around r ≈ 1.8°, where we find the first minimum in received copolar power (below −30 dB in SNRh).

Fig. 12.

The differential phase without and with the radome as a function of radial distance relative to the main SNR peak. Shown are the median and the first and third quartiles, respectively, based on the pattern data from the three source sites.

Fig. 12.

The differential phase without and with the radome as a function of radial distance relative to the main SNR peak. Shown are the median and the first and third quartiles, respectively, based on the pattern data from the three source sites.

The results for differential power are shown in Fig. 13. On average, differential power is essentially 0 dB up to a radial distance of r = 0.5°, which roughly corresponds to the 3-dB beamwidth. This is found for the measurements with and without radome. With increasing radial distance r, the power difference increases up to ≈0.7 dB at r = 1°. The increase is larger with radome, where we reach a power difference of about ≈1 dB at r = 1°. If the scattering volume is not beam filling or heterogeneous, then the observed differential power variability may affect the resulting data quality of ZDR (likely more than the differential phase measurements).

Fig. 13.

The differential power without and with the radome as a function of radial distance relative to the main SNR peak. Shown are the median and the first and third quartiles, respectively, based on the pattern data from the three source sites.

Fig. 13.

The differential power without and with the radome as a function of radial distance relative to the main SNR peak. Shown are the median and the first and third quartiles, respectively, based on the pattern data from the three source sites.

e. Beam squint and beamwidth

Beamwidth and beam squint (BSQ) are important characteristics of an antenna. In the following, we compute the beamwidth from the 3D data by extracting the location of the −3-dB isoline relative to the peak of the main beam. The beamwidths are computed from the E-plane data in both the horizontal and vertical polarization. We analyze data from the STAR and single-pol tests from three source sites. This is done separately for the tests with and without the new radome. The mean and the standard deviation of the beamwidth is summarized in Table 7. All the beamwidths are below 1°. The data indicate that the V beamwidths are on average always larger than the H beamwidths: without radome 0.90° (H) versus 0.93° (V) and with radome 0.88° (H) versus 0.91° (V).

Table 7.

BW and BSQ results with and without the radome and the differences in BW H and V.

BW and BSQ results with and without the radome and the differences in BW H and V.
BW and BSQ results with and without the radome and the differences in BW H and V.

We can relate those results to the gain of the antenna in order to demonstrate the relative importance of the results with and without radome. We use a simple formula to compute the gain as a function of beamwidth (Orfanidis 2008). Here, an antenna efficiency of 60% (a typical value based on literature) is taken, assuming an identical efficiency in H and V. The gain is g = 10 log(30 000/BW2), where the beamwidth (BW) is given in degrees. The results of this simple estimate are summarized in Table 8. The differences in beamwidth in H and V for measurements without and with radome amount to a gain difference between H and V of 0.1–0.2 dB. So, the differences in beamwidths for a given polarization relate to a gain difference of about 0.2 dB. We can conclude that the influence of the radome is rather small.

Table 8.

Gain estimates based on the analyzed BWs with and without radome; Δg is the difference between the antenna gain in H and V.

Gain estimates based on the analyzed BWs with and without radome; Δg is the difference between the antenna gain in H and V.
Gain estimates based on the analyzed BWs with and without radome; Δg is the difference between the antenna gain in H and V.

To compute the beam squint, we initially determine the position of the V peak relative to the main peak position in H. Then, we fit a 2D surface to the SNR data from which the peak positions in H and V are calculated. Then, the location of maximum SNRh and SNRv define an elevation and azimuth difference ( and ) between the location of the main beam peak in H and V. The beam squint is then computed as . The uncertainty of the beam squint is determined by the uncertainties of the surface fit. The results are shown in Table 7. It should be noted that the results actually represent the one-way beam squint. The beam squint with radome is always smaller than 0.04°, whereas two measurements without radome show a somewhat larger squint but still below the specified 0.06°.

The results for beamwidth and squint are summarized in Table 7. Overall, the specifications for beamwidth and squint are fulfilled with and without the radome.

5. Conclusions

This unique measurement campaign is the first approach to quantify on-site the performance and the effect of a radome on the RF performance of the antenna. Clearly, the specifications of the radome (in particular, in the main beam area) and their proof are difficult to achieve, since the target figures are at or beyond the measurement accuracy of the radar system. Plus, there are always some uncertainties related to the test range conditions. Nevertheless, the statistics of the measurements indicates at least the magnitude of a radome effect—in particular, if we compare the measurements with and without radome in the main plane directly. There, we have a very good database to derive quantitative results. We find that the beamwidth, shape, and squint do not degrade because of the radome. This in turn also means that the gain of the antenna should not deteriorate. The first sidelobe level is raised in one polarization plane compared to measurements without radome. The far sidelobe levels are raised significantly by the radome off the main beam (≈13 dB); however, the resulting levels are, on average, still within our specifications. The STAR mode tests indicate that the inhomogeneity of differential phase in the main beam is larger because of the radome. The larger inhomogeneity appears related to different panel combinations seen by the antenna aperture considering the three source sites. This in turn implies some azimuthal variability of the radar moments that still has to be quantified. The differential power measurements indicate a very good match up to the −3-dB level of the main beam. The power difference for the measurements with radome becomes larger with increasing radial distance (up to 1 dB). This may cause a larger ZDR bias due to the radome, especially for situations with nonbeam-filling targets.

The cross-polar isolation limits the maximum measurable linear depolarization ratio (LDR). ICPR estimates indicate that the radome causes a loss of about 1.4 dB of sensitivity in LDR. The large variability of the cross-polar data between the source sites at low elevations also implies that the achievable LDR in a stratiform rain situation may be quite variable depending on the propagation of the transmitted and received signals of the radar system.

Acknowledgments

EEC and DWD are greatly acknowledged in supporting these comprehensive antenna measurements. Kay Desler, Thomas Hohmann, Ladislav Hart, Serguei Laskovitch, and Manfred Feldman supported us significantly in achieving these measurements. We also acknowledge the comments of the reviewers, who helped us to improve the quality of this paper.

REFERENCES

REFERENCES
Bringi
,
V. N.
, and
V.
Chandrasekar
,
2001
: Polarimetric Doppler Weather Radar. Cambridge University Press, 636 pp.
Bringi
,
V. N.
,
R.
Hoferer
,
D. A.
Brunkow
,
R.
Schwerdtfeger
,
V.
Chandrasekar
,
S. A.
Rutledge
,
J.
George
, and
P. C.
Kennedy
,
2011
:
Design and performance characteristics of the new 8.5-m dual-offset Gregorian antenna for the CSU–CHILL radar
.
J. Atmos. Oceanic Technol.
,
28
,
907
920
.
Chandrasekar
,
V.
, and
R. J.
Keeler
,
1993
:
Antenna pattern analysis and measurements for multiparameter radars
.
J. Atmos. Oceanic Technol.
,
10
,
674
683
.
Doviak
,
R. J.
,
V.
Bringi
,
A.
Ryzhkov
,
A.
Zahrai
, and
D.
Zrnić
,
2000
:
Considerations for polarimetric upgrades to operational WSR-88D radars
.
J. Atmos. Oceanic Technol.
,
17
,
257
278
.
Frech
,
M.
,
2009
: The effect of a wet radome on dualpol data quality. Preprints, 34th Conf. on Radar Meteorology, Williamsburg, VA, Amer. Meteor. Soc., P13.15. [Available online at https://ams.confex.com/ams/34Radar/techprogram/paper_155405.htm.]
Gourley
,
J. J.
,
P.
Tabary
, and
J. P.
du Chatelet
,
2006
:
Data quality of the Meteo-France C-band polarimetric radar
.
J. Atmos. Oceanic Technol.
,
23
,
1340
1356
.
Kurri
,
M.
, and
A.
Huuskonen
,
2008
:
Measurements of the transmission loss of a radome at different rain intensities
.
J. Atmos. Oceanic Technol.
,
25
,
1590
1599
.
Mudukutore
,
A.
,
V.
Chandrasekar
, and
E. A.
Mueller
,
1995
:
The differential phase pattern of the CSU CHILL radar antenna
.
J. Atmos. Oceanic Technol.
,
12
,
1120
1123
.
Orfanidis
,
S. J.
, cited
2008
: Electromagnetic waves and antennas. [Available online at www.ece.rutgers.edu/~orfanidi/ewa.]
Skolnik
,
M. I.
,
2008
: Radar Handbook. 3rd ed. McGrawHill, 1328 pp.
Zrnić
,
D.
,
R.
Doviak
,
G.
Zhang
, and
A.
Ryzhkov
,
2010
:
Bias in differential reflectivity bias due to cross coupling through the radiation patterns of polarimetric weather radars
.
J. Atmos. Oceanic Technol.
,
27
,
1624
1637
.