HIAPER Cloud Radar Feasibility Study

Transcription

1 HIAPER Cloud Radar Feasibility Study Jothiram Vivekanandan, Wen-Chau Lee, Eric Loew, and Gordon Farquharson EOL/NCAR August 22, 2005

2 Executive Summary The HIAPER cloud radar (HCR) initiative provides an opportunity for expanding the envelope of airborne radar systems by delivering high spatial and temporal resolution observations with improved accuracy in comparison to existing radars. A survey of current technologies indicates that it is best to optimize the radar system to study one of the highest priority missions such as cloud remote sensing. At a future date, based on the performance of the cloud radar system, we may explore the possibility of expanding its utility to include remote sensing of the ocean and land surface. Requirements for cloud and precipitation radar remote sensing are very stringent due to the need for high sensitivity (> km) and wide dynamic range of received signal strengths (90 db or more). The intent of this document is to assess the feasibility of building an airborne millimeter wave radar to satisfy the observational needs of the atmospheric research community. The basic HCR concept was distilled from responses to a selectively distributed scientific community survey, a thorough examination of millimeter wave radar technologies, as well as from input provided in discussions with engineers and scientists at JPL, NASA Goddard, and the University of Wyoming. These discussions led to a proposed millimeter wave radar system which is capable of measuring both spectral moments and the full complement of polarimetric variables with sufficient sensitivity and accuracy to be useful in the study of cloud microphysics. Given cost constraints, and to obtain the greatest spatial coverage, we propose the vast majority of the radar system be housed in HIAPER s 20 wing pod, designated the NCAR pod. EOL staff considered four designs incorporating a pod-based polarimetric radar with scanning, dual-wavelength (K a - and W-band, matched or unmatched beams), dual-doppler (K a - or W-band). Based on the results of two community surveys, we recommend building a polarimetric Doppler dualwavelength (K a - and W-band) radar with both matched and unmatched beam configurations in a pod. The report includes a preliminary description of radar subsystems and also a cost estimate for both dual-wavelength and single-wavelength options. 2

3 Table of Contents 1. BACKGROUND SCOPE HIAPER CLOUD RADAR SCIENTIFIC GOALS DYNAMICS (KINEMATICS) Vertical Air Velocity and Mean Particle Size D Velocity Vectors MICRO-PHYSICS Detection of Super cooled Large Droplet Using S-band Polarization Radar Measurements Retrieval of Droplet Size and Liquid Water Content from Dual-wavelength Radar Measurements MILLIMETER WAVE TECHNOLOGIES ANTENNA Parabolic Reflector Antennas Lens Antennas TRANSMITTER RECEIVER RADAR SYSTEM CONCEPTS W-BAND DUAL-DOPPLER K A -BAND DUAL-DOPPLER DUAL-WAVELENGTH (W- AND K A -BAND) SENSITIVITY COMPARISONS RADAR DATA SYSTEM DATA ACQUISITION Pulse Compression Short Pulse RADAR TIMING AND CONTROL HOUSEKEEPING SIGNAL PROCESSING Polarimetric Variables Spectral Processing DISPLAY AND ARCHIVE POD INFRASTRUCTURE RADOME(S) ROTATING SPLASH PLATE(S) ENVIRONMENTAL CONTROL Pressurization Temperature Control

4 8. SYSTEM CALIBRATION INTERNAL CALIBRATION ABSOLUTE CALIBRATION TEST AND MEASUREMENT EQUIPMENT PROPOSED HCR DEVELOPMENT TEAM COST SURVEY AND RECOMMENDATION SURVEY RECOMMENDATION REFERENCES

5 List of Figures FIGURE 1 W-BAND DOPPLER SPECTRUM OF A LIGHT PRECIPITATION EVENT. THE DISTINCT MINIMUM IN POWER, AROUND 6M S -1, CORRESPONDS TO THE MIE SCATTERING OF SMALL RAINDROPS FIGURE 2 THE NASA W-BAND DOWNWARD LOOKING CLOUD RADAR SYSTEM (CRS) ONBOARD THE ER-2 CAN FLY OVER ALL CONVECTIVE SYSTEMS AND PROVIDE VERTICAL PROFILES OF REFLECTIVITY AND DOPPLER VELOCITY. THE CRS OBSERVATION OF A STRATIFORM PRECIPITATION SYSTEM (ABOVE) CLEARLY ILLUSTRATES THE BRIGHT BAND STRUCTURE AS ENHANCED REFLECTIVITY (~4 KM ALTITUDE, TOP PANEL) AND INCREASED PARTICLE FALL SPEED BELOW THE BRIGHT BAND. THE LIMITED AREA OF UPDRAFT NEAR THE CLOUD TOP CAN ALSO BE SEEN FIGURE 3: POSSIBLE DUAL-BEAM CONFIGURATIONS TO ESTIMATE 2-D WINDS IN A SPECIFIED PLANE PERPENDICULAR TO AIRCRAFT MOTION FIGURE 4: RHI FROM IMPROVE II SHOWING PID RESULTS (TOP PANEL), REFLECTIVITY, DBZ (CENTER PANEL) AND DIFFERENTIAL REFLECTIVITY, DB (BOTTOM PANEL). THE OBSERVATION WAS COLLECTED BY S-POL RADAR ON NOVEMBER 29, 2001 NEAR OREGON CASCADES FIGURE 5: OVER-THE-TOP RHI SCANS OF THE LINEAR DEPOLARIZATION RATIO (LDR) AND REFLECTIVITY (Z) FOR FOUR TYPES OF PRECIPITATION: (A) DRIZZLE, (B) DENDRITES, (C) LONG COLUMNS AND (D) BLOCKY COLUMNS (REINKING ET AL 2002) FIGURE 6: X- AND K A -BAND REFLECTIVITY, DUAL-WAVELENGTH RATIO (DWR), ALONG WITH RETRIEVED LWC AND RES ALONG A RADIAL FOR A STRATIFIED CLOUD WITH PRECIPITATION PARTICLE SMALL COMPARED TO K A -BAND WAVELENGTH. SLOPE OF THE LINE JOINING LOCAL MINIMA IN THE DWR IS USED FOR ESTIMATING ATTENUATION AND LWC. THE DATA WERE COLLECTED IN NORTHEASTERN COLORADO ON 18 APRIL 1991 BETWEEN 17:28 AND 17:30 GMT ALONG 2830 AZIMUTH AND 1.50 ELEVATION FIGURE 7: PLAN VIEW OF NCAR POD MOUNTED BELOW SURFACE OF LEFT WING FIGURE 8: SINGLE WAVELENGTH (W OR K A ), DUAL-DOPPLER, DUAL-POLARIZATION RADAR CONCEPT. NCAR POD SHOWN HANGING FROM LEFT WING FIGURE 9: DUAL-WAVELENGTH, DUAL-POLARIZATION RADAR CONCEPT. NCAR POD SHOWN HANGING FROM LEFT WING FIGURE 10: MINIMUM DETECTABLE REFLECTIVITY FOR A NON-SCANNING RADAR AT THE VARIOUS WAVELENGTHS CONSIDERED IN THE PRESENCE OF 2.5 G/M3 OF WATER VAPOR FIGURE 11: MINIMUM DETECTABLE REFLECTIVITY FOR A NON-SCANNING RADAR AT THE VARIOUS WAVELENGTHS CONSIDERED IN THE PRESENCE OF 10 MM/HR RAIN FIGURE 12: HCR DATA ACQUISITION AND STORAGE SYSTEM. DATA ACQUISITION AND INITIAL PROCESSING ARE PERFORMED IN THE POD AND DATA IS SENT FOR FURTHER PROCESSING, DISPLAY AND ARCHIVING TO THE CABIN OVER GIGABIT ETHERNET

6 List of Tables TABLE 1: COMPARISON OF EXISTING AIRBORNE MILLIMETER WAVE RADARS... 7 TABLE 2: W-BAND RADAR SYSTEM CHARACTERISTICS (FIXED POINTING ANGLE) TABLE 3: K A -BAND SYSTEM CHARACTERISTICS (FIXED POINTING ANGLE) TABLE 4: COST SUMMARY

7 1. Background The HIAPER cloud radar initiative provides an opportunity for expanding the envelope of airborne radar systems by delivering high spatial and temporal resolution observations with improved accuracy in comparison to existing radars. A number of airborne real aperture radars exist for cloud and precipitation remote sensing. For example, as shown in Table 1, the NASA CRS, Wyoming Cloud radar, Canadian Convair-580 cloud radar, UMass/JPL ACR, UMass CMR ARM UAV, and Spider (Japan; deployed on a Gulfstream III) are capable of sensing -10 to -30 dbz at 10 km range. Most of these radar systems lack scanning capability, except the Spider which has limited cross-track scanning. Also, measurement of 2-D winds, full-polarimetric observations or advanced signal processing of radar time series is uncommon. It is natural that the proposed radar for HIAPER may take advantage of many of the existing airborne radar features for an optimal configuration. Table 1: Comparison of Existing Airborne Millimeter Wave Radars 10 km (dbz) Scanning 2-D Winds Dualλ Time Series Altitude (km) NASA CRS -28 No No No No 20 Wyoming Cloud Radar Canadian Convair-580 UMass/JPL ACR CMR ARM UAV -10 No Yes No No No No Yes No No No No No No No No No 20 Spider (Japan) -20 Yes No No No 13 Proposed HCR -30 Yes Maybe Maybe Yes 15 The HIAPER platform might be expected to support a variety of science missions that may include sensing of the atmosphere, ocean and land surface either as a stand alone instrument and/or complimenting other instruments onboard the HIAPER. The challenge could be: Can we build a radar with an antenna, a transmitter, and a sensitive coherent receiver that can be reconfigured for various types of missions? A radar system characterized by low antenna sidelobes (< -30 db), multiple-wavelength feeds, flexible pulsing schemes, and extremely wide receiver bandwidth (> 100 MHz) can be reconfigured for a variety of applications. However, a survey of current technologies indicates that it is best to optimize the radar system to study one of the highest priority missions such as cloud remote sensing. At a future date, based on the performance of the cloud radar system, we may explore the possibility of expanding its utility in the remote sensing to include the ocean and the land surface. 7

8 The technical specifications of a radar primarily depend on the scientific expectations of the research community. Many researchers atmospheric remote sensing are familiar with radar system requirements for a specific scientific application. In addition, researchers tend to capitalize on multiple sensors to enhance products that are obtained using remote sensing observations. Thus, a questionnaire was created to address the following important issues: (1) scientific requirements, (2) system requirements, and (3) the usage of radar with other sensors on board the HIAPER platform. This survey was distributed to about 20 members of the cloud radar user community in December The survey results indicated a common preference for a narrow beamed W-band radar with polarimetric and Doppler capabilities. Additional capabilities included a second wavelength and/or dual-doppler winds. Requirements for cloud and precipitation radar remote sensing are very stringent due to the need for high sensitivity (> km) and wide dynamic range of received signal strengths (90 db or more). The absorption of millimeter wave signals due to water vapor, and the effects of Mie scattering (i.e. hydrometeor size comparable to radar wavelength) also greatly affect radar performance. In addition to the above-mentioned basic system requirements, it is also important to consider various configurations of a typical airborne weather radar system: Single vs. dual-doppler observations: Both microphysics and kinematics are closely related. A dual-beam configuration is capable of retrieving both radial and crossbeam winds, i.e. more accurate depictions of the up and down draft motions in clouds. Co- and cross-polarization measurements: Polarization observations are routinely used in a number of research radars for retrieving cloud microphysics such as delineating regions of liquid and ice and mixed phase vs. single phase clouds. Also polarization observations can be used for quality control of the data. Scan vs. a simple stare mode: A scan mode offers better spatial coverage but accurate Doppler observation in a scan mode requires precise beam-pointing information. As the radar will be mounted on a wing-pod, it is important to compensate for both aircraft motion and wing deflection. Is it practical to use a separate wide bandwidth receiver with the same radio frequency receiver configuration for a passive remote sensing of cloud liquid, water vapor, soil moisture, ocean winds, and sea ice characterization studies? The discussions in the following sections suggest that a wide-band receiver can be built at a later stage, but the current atmospheric community need is for a robust radar with the highest sensitivity using the current technology. Planar phased array vs. reflector type antennas: A low weight conformal phased array antenna may offer adaptive scanning for accomplishing rapid scanning and also reduction of Doppler spectral broadening due to beam motion during Doppler observation. However, a phased array system at millimeter wave band is expensive and it might be risky to build such a system. A reflector or lens type antenna may satisfy the research community s present requirements for sensitivity, Doppler velocity measurements and polarization capability. 8

9 2. Scope The intent of this document is to assess the feasibility of building an airborne millimeter wave radar to satisfy the observational needs of the atmospheric research community. One of the attractive features of a millimeter wave radar system is its ability to detect micron-sized particles that constitute clouds with lower than 0.1 g m -3 liquid or ice water content. The technical specifications of such a radar are mainly driven by climate and cloud initiation studies. The envisioned capability of a millimeter wave radar system on HIAPER is enhanced by coordination with onboard LIDAR, microwave radiometer and in situ probes. The radar measurements would be suitable for quantifying both kinematics and microphysical parameterization schemes in cloud modeling and verifying temporal and spatial scales of cloud systems. Based on a survey sent to the cloud radar user community in December 2004, the survey results indicated a common preference for a narrow beam W-band radar with polarimetric and Doppler capabilities. It is desired to have -30 dbz sensitivity at 10 km range. Additional capabilities included a second wavelength and/or dual-doppler winds. Modern radar technology offers various options: dual-beams, dual-polarization and dualwavelength. Even though a basic Doppler radar system with a sensitivity poorer than -25 dbz at 10 km. is capable of satisfying most of the common scientific needs, the abovementioned options significantly extend the measurement capabilities to further reduce any uncertainty in radar-based retrievals of cloud properties. An optimal radar configuration that is capable of maximizing the accuracy of both qualitative and quantitative estimated cloud properties is the most attractive option to the research community. However, a pod-based system on a high altitude platform imposes engineering design challenges in terms of the radar s size, weight, and ability to handle pressure and temperature extremes. This document describes various radar configurations based taking into account these limitations and current millimeter wave technologies. 9

10 3. HIAPER Cloud Radar Scientific Goals The HIAPER cloud radar (HCR) will serve in both cloud microphysics and cloud dynamics studies. The following sections describe the research that HCR may support Dynamics (kinematics) Scanning or vertically-pointing ground-based millimeter wavelength radars have been used to study stratocumulus (e.g., Vali et al. 1998; Wilczak et al. 1996; Kollias and Albrecht 2000), fair-weather cumulus (e.g., Kollias et al. 2001) and fog properties (e.g., Hamazu et al. 2003). Airborne millimeter wavelength radar systems such as the University of Wyoming King Air Cloud Radar (WCR) and the NASA ER-2 Cloud Profiling System (CRS) have added mobility to observe clouds in remote regions and over the oceans Vertical Air Velocity and Mean Particle Size Vertical pointing cloud radars have been used to measure vertical velocities in convective clouds. At shorter wavelengths (e.g., W-band), small raindrops (> 0.5 mm) are in the Mie scattering region rather than in the Rayleigh region. Oscillations in Mie scattering cause distinctive minima in radar cross section for certain droplet sizes. The first minimum occurs at a raindrop diameter of around 1.7 mm. Hence, backscattered power in the Mie region can be clearly detected by the Doppler spectrum as indicated in Figure 1, where the velocity is essentially the combination of the terminal fall speed of the hydrometeors and the vertical air motion in a resolution volume. Since the theoretical spectrum of a mean particle size can be calculated for a given wavelength, velocity values of a known power minimum can be compared with the theoretical spectrum, and the mean vertical air motion can be estimated (e.g., Lhermitte 2002; Kollias 1999). A vertical velocity profile with altitude can then be obtained by analyzing power spectra at multiple levels. Figure 1 W-band Doppler spectrum of a light precipitation event. The distinct minimum in power, around 6m s -1, corresponds to the Mie scattering of small raindrops. 10

11 Figure 2 The NASA W-band downward looking Cloud Radar System (CRS) onboard the ER-2 can fly over all convective systems and provide vertical profiles of reflectivity and Doppler velocity. The CRS observation of a stratiform precipitation system (above) clearly illustrates the bright band structure as enhanced reflectivity (~4 km altitude, top panel) and increased particle fall speed below the bright band. The limited area of updraft near the cloud top can also be seen D Velocity Vectors The 2-D wind field can be obtained by installing two antennas pointing at different angles in the same plane. This approach has been successfully demonstrated by the WCR (Geerts et al. 2005) on either the horizontal or vertical plane as illustrated in Figure 3. The angle separation of the WCR is 30. With the sensitivity at W-band, the WCR can resolve detailed two dimensional flow structures in a developing cumulus cloud. The high spatial resolution of the WCR has revealed the detailed internal structure of developing clouds and thermals in the boundary layer allowing the turbulence characteristics and cloud edge entrainment to be studied. 11

12 Figure 3: Possible dual-beam configurations to estimate 2-D winds in a specified plane perpendicular to aircraft motion Micro-Physics Several studies have shown that polarimetric observables (both linear and circular) can be used to identify hydrometeor types (Doviak and Zrnic 1993; Hall et al. 1984; Hendry and Antar 1984; Straka et al 2000). Hendry and Antar (1984) used circular polarization measurements for delineating the major precipitation types such as drizzle, rain, melting layer, snowflakes, ice crystals, and ice pellets. However, rain or anisotropy precipitation in the propagation path between the antenna and the radar resolution volume introduces more bias in circular polarization radar measurements than in the case of linear polarization (Doviak and Zrnic 1993). Linear polarimetric radars, transmit and receive both horizontally and vertically polarized radiation, providing more information about the scattering media than conventional radar. Polarimetric radar observables depend on the microphysical characteristics of hydrometeors; namely, (a) particle size, (b) particle shape, (c) particle orientation relative to the local vertical direction, (d) phase (liquid or ice), and (e) bulk density (wet, dry, aggregate, or rimed). In addition to traditional reflectivity (Z HH ) and Doppler measurements, linear polarimetric observables include differential reflectivity (Z DR ), linear depolarization ratio (LDR), specific differential propagation phase (K DP ) and correlation coefficient (ρ HV ). Reflectivity is related to the power of a horizontally polarized backscattered electric field from a radar resolution volume for a horizontally polarized transmitted wave (co-polar). Reflectivity is the sixth moment of the particle size distribution when particle size is small compared to the wavelength. Differential reflectivity is the ratio of the horizontal co-polar return to the vertical co-polar return and can be interpreted as the reflectivity weighted mean-axis ratio of the precipitation particle in the radar resolution volume (Jameson 1983). Linear depolarization ratio is the ratio between vertically polarized power backscattered for a horizontally polarized transmitted wave and co-polar backscattered power. Tumbling, wet non-spherical particles such as hail, melting 12

13 aggregates, wet graupel, and bright band due to melting (Vivekanandan et al. 1990; Zrnic et al. 1993) are identified with large LDR values whereas light rain, cloud droplets, and dry ice particles are associated with low LDR values. Specific differential propagation phase (K DP ) is the difference in phase per kilometer of the received horizontal and vertical polarized waves. It is almost linearly proportional to rain rate and ice water content (Sachidananda and Zrnic 1986; Vivekanandan et al. 1994). Here, K DP can be used to identify non-spherical particles, such as ice crystals and raindrops. Vivekanandan et al. (1994) describe a method for delineating regions of pristine ice crystal and snow using Z HH and Z DR observations. The correlation coefficient between co-polar returns is denoted as ρ HV. Values of ρ HV are close to unity for rain and pure ice crystals. In the case of melting and mixed phase (rain and hail or graupel) conditions, ρ HV is smaller than unity because of the variability in scattering characteristics of precipitation particles for a given size. Low values of ρ HV may be used for detecting hail and mixed phase. Additionally, polarization techniques (as in Reinking et al., 1997; Vivekanandan et al., 1994, 1999b) may be able to identify the presence of ice and quantify mass and/or size. The development of remote sensing methods to detect, measure, and map cloud liquid water content (LWC) is in its infancy, but it has enormous potential benefits for cloud seeding, and basic cloud physics research. One of the methods uses dual-wavelength radar and relies on the fact that the shorter of the two wavelengths is more strongly attenuated by liquid water than the longer wavelength. For Rayleigh scattering conditions (hydrometeors much smaller than the radar wavelengths), the range derivative of the observed wavelength reflectivity difference is linearly related to the LWC of the cloud in that location. Gosset and Sauvageot (1992) describe a theoretical approach to estimating liquid and ice mass contents using dual-wavelength attenuation and reflectivity, respectively. The range-differentiated difference between the returned signals is proportional to the amount of liquid present. A preliminary analysis of dual-wavelength radar (X- and K a -band) and microwave radiometer observations is discussed by Martner et al. (1993). A least-squares fit of the rangedifferentiated reflectivity difference over a distance of 4 km was used for estimating attenuation and LWC. Factors such as non-rayleigh scattering, a short liquid water path (< 4 km) and sensitivity of the X-band radar system limited the LWC retrieval. By combining a detailed model study with simultaneous radar and radiometer observations, Vivekanandan et al. (1999a) demonstrated the feasibility of a dualwavelength system for the derivation of range-gated LWC along the beam path. The LWC retrieval is confounded by the presence of ice crystals and hydrometeors, either ice or liquid, in the Mie-scattering size range (where the particle diameter is comparable to the radar wavelength). Ice crystals will not affect the attenuation measurement for this wavelength pair, but they will have an effect on the reflectivity difference. They are generally larger than liquid droplets in mixed-phase clouds. Even though their dielectric properties give them smaller scattering cross sections, their larger sizes will tend to dominate the reflectivity. Vivekanandan et al. (1999a) discussed the methods by which Mie scatterers may be identified. 13

14 Detection of Super cooled Large Droplet Using S-band Polarization Radar Measurements At millimeter wave frequencies, polarimetric observations not only depend on the mean shape and orientation of hydrometeors but also the mean size of hydrometeors, Hence, interpretation of polarization observation at millimeter wavelength has to take into account of Mie and Rayleigh scattering phenomena. As discussed in the previous section, polarimetric observations either at S-band or at millimeter wave band have the potential to delineate regions of liquid and ice in both stratiform and convective clouds. This section shows an example of cloud phase detection using S-band observations. This technique is applicable to millimerer wave band observations. The SLW (super-cooled liquid water) category based on S-band polarization observation in the past consisted only of the smallest drops and the larger drops were classified as drizzle or rain, which was confusing when it occurred above the freezing level. Previous improvements to the Particle IDentification (PID) focused on the larger particles such as graupel and heavy rain. Optimization efforts have shifted to the smaller, more difficult to distinguish particles. Recent upgrades to the PID algorithm, which utilize dual-polarization weather radar data, include improving the distinction of ice crystals from SLW and expanding the SLW category to include a full range of sizes from cloud droplets to large drizzle drops The improvement in SLW detection follows Vivekanandan et al. (1999). Preliminary tests on IMPROVE II data show encouraging results. On November 11, 2001 the University of Washington Convair, a research aircraft experienced icing while penetrating a cloud consisting mostly of SLW. Simultaneous observations with the NCAR S-Pol radar were performed allowing comparison to the aircraft. Figure 4 shows a RHI scan taken during the time when SLW was observed. It can be seen that the PID identifies the cloud as mostly SLW, in qualitative agreement with the aircraft in situ observations. Figure 5 shows over-the-top RHI scans of reflectivity (Z) and the linear depolarization ratio (LDR) in various types of precipitation. The data were collected by the NOAA ETL K a -band radar during the Mount Washington Icing Project (Reinking et al. 2002). The two panels of radar observations in each of the precipitation type correspond to LDR and Z. Reflectivity varies as a function of cloud intensity since it is a function of both the concentration and the mean particle size. However, LDR is primarily sensitive to particle shapes and it also depends on the elevation angle of the radar beam i.e. aspect ratio of particle with respect to the radar beam. In the case of drizzle i.e. spherical particle, LDR is independent of the elevation angle of the beam whereas LDR is a strong function of elevation angle for dendrite and column. Thus, it is straightforward to delineate regions of cloud and drizzle drops from non-spherical or irregular ice particles. Within the ice particle habits, it is also possible to differentiate between columns vs. dendrites using the following: (i) the value of LDR and (ii) the dependence of LDR on elevation angle of radar beam. 14

15 Figure 4: RHI from IMPROVE II showing PID results (top panel), reflectivity, dbz (center panel) and differential reflectivity, db (bottom panel). The observation was collected by S-Pol radar on November 29, 2001 near Oregon Cascades. 15

16 Figure 5: Over-the-top RHI scans of the linear depolarization ratio (LDR) and reflectivity (Z) for four types of precipitation: (a) drizzle, (b) dendrites, (c) long columns and (d) blocky columns (Reinking et al 2002) Retrieval of Droplet Size and Liquid Water Content from Dualwavelength Radar Measurements In the case of spherical hydrometeors such as cloud droplets and drizzle, the utility of polarimetric measurements to estimate LWC and droplet size are limited since backscatter measurement is independent of polarization. A dual-wavelength method that makes use of absorption and Mie scattering is more attractive for estimating LWC and droplet size. An example of the application of the dualwavelength technique for estimating both droplet size and liquid water content using dual-wavelength (K a and X-band) radar measurements is shown in Vivekanandan et al The first of these is the radar-estimated size (RES), defined as the cube root of the 6 th moment (Z) and 3 rd moment (A) size distribution ratio. The RES is shown to be useful in characterizing icing environments through simulations using modified Gamma droplet size distributions with realistic bounds based on several sets of in-situ measurements. The RES is also relatively easy to retrieve using dual-wavelength radar measurements. RES can be written using radar measurements as 16

17 RES = Z Where, RES is in mm, Z is in mm 6 m -3 and A is in db km -1. In a size distribution with both small and large particles (i.e. broad spectrum) the RES value is primarily biased towards large particle size. Thus in a mixed-phase cloud with a few large ice particles, RES might not characterize the size distribution well. In general, for a liquid cloud with maximum particle size smaller than the radar wavelength, RES is greater than or equal to MVD (median volume diameter). RES has several advantages. It can be estimated directly from dual-wavelength radar measurements without having to know or infer information about the shape of the drop size distribution. The second parameter is the liquid water content (LWC), the total mass of liquid droplets of all sizes. The LWC may be estimated by taking advantage of the difference in attenuation due to the presence of liquid water between the X- and K a -band radars. The attenuation at K a -band is composed of both absorption and scattering losses. For small droplets, the scattering loss is negligible, and hence attenuation is essentially the absorption loss. The absorption cross-section is proportional to the volume (or LWC) of the droplets and depends on the imaginary part of the refractive index of water. At 10 o C, the one-way attenuation of the K a -band radar signal by cloud droplets, A, is related to LWC as LWC = A where LWC is in g m -3 and A is in db km -1. As ambient temperature decreases (increases), the corresponding attenuation increases (decreases). For example, as the temperature decreases from 0 to 30 o C, attenuation increases by 200%, and thus temperature measurement is a desired feature of a system designed to accurately retrieve LWC. The K a -band attenuation is estimated by comparing reflectivity gate-by-gate with that from the co-located X-band radar. Figure 6 shows the X- and K a -band reflectivity along a radial for a stratified cloud with precipitation particles that were small compared to K a -band wavelength. The attenuation-corrected dual-wavelength ratio (DWR) suggests that the scatterers are small compared to the K a -band wavelength. K a -band attenuation is used to estimate LWC. Although there is no in situ verification of LWC, possible presence of ice or particle size, the results demonstrate an automatic retrieval of both size and LWC using dual-wavelength observations. RES was derived from (5). Between 14 and 16 km, larger LWC amounts correspond to smaller RES, indicating higher amounts of liquid in smaller, cloud-sized (<~50 μm diameter) droplets. The maximum RES and LWC along the radial were 300 μm and 0.35 g m -3, respectively. RES of 300 microns suggests maximum drop size in the spectrum 300 μm. Note: for a narrow DSD, RES is the same as MVD. The results illustrate that LWC and RES are not necessarily correlated. There is no physical reason to expect such a correlation in general. A 17

18 RES, mm LWC, g m -3 DWR, db Ka-band, X-band, dbz X band, dbz Ka band, dbz DWR, db Figure 6: X- and K a -band reflectivity, dual-wavelength ratio (DWR), along with retrieved LWC and RES along a radial for a stratified cloud with precipitation particle small compared to K a -band wavelength. Slope of the line joining local minima in the DWR is used for estimating attenuation and LWC. The data were collected in northeastern Colorado on 18 April 1991 between 17:28 and 17:30 GMT along 2830 azimuth and 1.50 elevation. 18

19 4. Millimeter Wave Technologies Before any viable millimeter wave system concept could be defined it was necessary to explore the prevalent technologies that exist for the various major radar subsystems, as these differ considerably from their centimeter wave counterparts. These subsystems consist of the antenna, transmitter and receiver. This research was done by having technical discussions with groups who deploy airborne millimeter wave radars, namely JPL, University of Wyoming and NASA Goddard. It was also conducted using the internet, reading relevant publications, talking with vendors and through textbooks. Below is a summary of the relevant technologies for millimeter wave radar systems Antenna A wide range of antenna options exist at millimeter wave frequencies. These fall into two main classes: lens antennas and parabolic reflector antennas. Corrugated feed horns are used to illuminate either lens or reflector in order to optimize antenna sidelobe performance. Each class has its advantages and disadvantages; these will be discussed in the following sections Parabolic Reflector Antennas Parabolic reflector antennas use a parabolic reflector of some type to focus electromagnetic waves. For millimeter waves these include: prime focus, Cassegrain and offset Gregorian antennas. The prime focus antenna places the feed at the focal point of a parabolic reflector. This approach is undesirable for two reasons. First, beam blockage by the feed and its support structure lead to increased sidelobe levels and decreased cross-pol isolation. Second, the additional waveguide length required to reach the feed results in a significant loss of sensitivity at millimeter wave frequencies. The typical one-way loss for W band waveguide is 0.8 db/ft. The Cassegrain antenna has its origins in optical telescope design. It is a variant of the prime focus concept, but instead places a sub-reflector at the focal point, while the feed is located at the center of the parabolic reflector. While this design overcomes the loss in sensitivity experienced by prime focus antennas due to additional waveguide length, it still suffers from beam blockage due to the subreflector and its supports. Typical cross-pol isolation of 30 db can be achieved with sidelobes on the order of -23 to -25 db. These antennas are available commercially from several sources. The Gregorian antenna uses the basic concept of the Cassegrain, but offsets the feed and sub-reflector in such a way that no beam blockage occurs. This type of antenna offers the best performance of any parabolic reflector antenna, achieving cross-pol isolation of 30 db and sidelobe levels of -25 db or better. The tradeoff in this case is physical size. Gregorian designs typically have a focal length to diameter ratio (f/d) of about 1.0 as compared to 0.5 for a typical Cassegrain or prime focus design. In addition, they require increased space to accommodate the offset sub-reflector a potential issue for the HCR. These antennas also require 19

20 very precise design and fabrication tolerances and consequently are a custom or semi-custom commodity Lens Antennas While parabolic reflector antennas rely on reflection to focus electro-magnetic waves onto the feed, lens antennas use refraction to accomplish the same goal. The millimeter wave lens works much the same as an optical lens. The lens is made of a material whose dielectric constant and shape will focus the waves appropriately. Typical dielectric materials are: rexolite, polystyrene and polyethylene. The thickness of the lens is directly proportional to the transmit wavelength. This generally makes dielectric lenses too heavy and impractical to build for microwave applications, but they are ideally suited for the smaller antennas generally associated with millimeter wave systems. As there is no beam blockage by the feed or a sub-reflector, superior sidelobe performance and crosspol isolation can be achieved over a parabolic reflector antenna having the same aperture. The tradeoffs are size and weight. Whereas parabolic reflector antennas typically have f/d ratios between 0.5 and 1 depending on the type, lens antennas require f/d of 1.3 or greater to achieve cross-pol isolation greater that 30 db Transmitter Millimeter wave transmitter technology consists of slow wave devices such as magnetrons, traveling wave tubes (TWT), and extended interaction klystrons (EIK), and fast wave devices such as gyrotrons. A magnetron is a cross-field device that generates an RF signal at a frequency proportional to the size of the cavity. Since cavity size changes are mostly affected by temperature changes during operation, the frequency of the RF signal will drift. Thus, receivers in radars that use magnetron transmitters must be designed to track the changes in the magnetron frequency, and thus be necessarily wideband which leads to additional complexity in the receiver. At microwave frequencies, magnetrons are often used in radars to provide high transmit powers for improved detection. However, conventional magnetrons at frequencies of 95 GHz are impractical to manufacture due to small cavity vane sizes, and are limited to relatively low peak and average powers due to voltage breakdown, heat dissipation, and high current densities which limit the life of magnetrons to hundreds of hours. Non-conventional magnetron designs have been explored such as spatial-harmonic magnetrons with a cold secondary emission cathode, but the maximum average power that can be achieved by these magnetrons is around 20 W, which is comparable to what can be obtained from a fixed frequency transmitter, such as a traveling wave tube amplifier (TWTA) or an extended interaction klystron amplifier (EIKA) with pulse coding. Finally, high peak powers produced by some magnetrons would also not be suitable for a HIAPER cloud radar since the waveguide would need to be filled with SF-6, which would make in field maintenance difficult. Most millimeter wave systems and almost all W-band systems built today use linear beam amplifiers such as TWTAs and EIKAs. These are devices that amplify input RF signals. Conduction cooled EIKAs are available at both K a - and W-band from Communications & Power Industries (CPI, Canada). These units are based on the 20

21 design of the tube in the CloudSat radar and have a maximum power of 1.3 kw at W- band and 1.5 kw at K a -band with a 5% duty cycle. CPI makes a modulator that is able to drive these tubes. This option is appealing in that a dual-wavelength system with two modulators naturally provides a spare for one of the wavelengths if one were to fail in the field. This modulator is large and heavy, but currently is the only proven option for driving these tubes. However, another company, Pulse Systems, has recently introduced modulators for these millimeter wave tubes. Several groups such as Prosensing, the National Research Council Canada, and the University of Massachusetts are currently evaluating these modulators, and if proven to be reliable, may provide us with a low cost and physically smaller alternative. However, the options and performance outlined in this document are based on the size and weight of the CPI modulator. A coupled-cavity TWTA at K a -band is also available with conduction cooling. This unit from CPI (Millitron) has a peak power of 850 W with a 15% duty cycle. It has lower power than the EIK, but has a higher duty cycle and also may provide a more linear phase response which may be important for good pulse compression. Conduction cooled systems are desirable for altitudes above about 30,000 feet since the decreased air density at these altitudes makes convective cooling far less effective. See Section for a more detailed explanation. Other devices such as gyrotrons exist. Gyrotrons are high powered electron tubes which emit a millimeter wave beam by bunching electrons with cyclotron motion in a strong magnetic field. Typical output powers range from tens of kilowatts to 1-2 megawatts. Output frequencies range from about 20 to 200 GHz. These devices are large and often require cryogenically cooled magnets, and therefore are unsuitable for an airborne pod-based instrument. A further consideration for the transmitter is the power handling capacity of waveguide at W-band. The maximum average power range for a TE 10 mode copper waveguide at K a -band is W (26 GHz) to W (40 GHz) and W (75 GHz) to W (110 GHz). Thus at K a -band we are not limited by the amount of transmitted power by the waveguide. However, at W-band, the goal is to transmit around 32 W of average power which is higher than the rated maximum for W-band waveguide. This limitation is predominantly due to heating in the walls of the waveguide (King 1961). The average power rating can be increased by using forced air cooling and radiating fins. However, care must be taken in designing such a system, since the maximum power rating decreases with increasing VSWR, due to high current points caused by standing waves Receiver Most receiver components are available from a variety of commercial vendors. In the past, millimeter wave low noise amplifiers had high noise figures, and a lower noise front-end could be constructed using a mixer followed by a low noise amplifier (Pazmany 1994). Today, commercial low noise amplifiers at K a -band and W-band have noise figures of around 2.5 and 4.0 db (Farran Technology) respectively and thus it has become possible to place the low noise amplifier directly after the polarization switching network. Typical front-end noise figures are 7-8 db at W- band, and 5-6 db at K a -band are thus possible. 21

22 Latching circulators are typically used to construct the front-end polarization switching network. These devices are available from EMS Technologies. The maximum power that these devices can handle is limited by the heat they are able to dissipate. At K a -band, 100 W of average power is supported which is above what we plan to transmit. However, at W-band, only 20 W is supported. Thus, to achieve maximum transmitted power and therefore sensitivity, a polarization switch should be avoided. With a single antenna and no polarization switch we are limited to only measuring half of the scattering matrix by either transmitting a single polarization and receiving both polarization returns, or transmitting both polarizations simultaneously ( slant 45 ) and receiving both. Many manufacturers will also construct an integrated receiver based on a customer s design, or on customer requirements. These can often be densely integrated to minimize waveguide losses and provide thermal stability to the receiver. The cost estimate provided in this report is based on integration of the receiver front-end by EOL staff. However, it may be cost effective to subcontract the receiver front-end to a commercial vendor. 22

23 5. Radar System Concepts The basic HCR concept was distilled from responses to a selectively distributed scientific community survey, a thorough examination of millimeter wave radar technologies, as well as from input provided in discussions with engineers and scientists at JPL, NASA Goddard, and the University of Wyoming. These discussions led to a proposed millimeter wave radar system which is capable of measuring both spectral moments (e.g., reflectivity, velocity, and spectrum width) and the full complement of polarimetric variables with sufficient sensitivity (>-25 10km) and accuracy to be useful in the study of cloud microphysics. In addition, dual-doppler winds and a second wavelength (K u or K a band) are highly desirable capabilities. Given cost constraints, and to obtain the greatest spatial coverage, we propose the vast majority of the radar system be housed in HIAPER s 20 wing pod, designated the NCAR pod. The HIAPER instrumentation philosophy dictates that only power and a high-speed network connection will be available to equipment located in the wing pods. This necessitates that nearly the entire radar system be located within the pod; the exception being the radar control and data display/archive computer(s). The pod will be attached to the mid-wing hardpoint as depicted in Figure 7, which shows a plan view scale drawing of the pod in relation to the aircraft s left wing. This drawing is based on the most recent information available from EOL s Design and Fabrication Services (DFS). The dimensions of the NCAR pod limit the performance and capabilities of the HCR in two ways. First, the antenna aperture of the HCR is limited to less than 15. This limits both sensitivity and spatial resolution as the gain of an antenna is directly proportional to its aperture, while its beamwidth is inversely proportional to its aperture. This aperture constraint is necessary to avoid beam blockage by the pod s internal support structure as well as to enable zenith viewing through the use of a rotating reflector (splash plate). It also makes a Ku band (2.2 cm wavelength) impractical, owing to its 3.9 degree beamwidth and its significantly reduced sensitivity ( km). Second, dual wavelength and dual Doppler are mutually exclusive properties of the HCR as space is lacking to accommodate the equipment necessary for both. This conclusion is based on a volume analysis which calculated the ratio of estimated equipment volume to the available volume for radar instrumentation. Ratio s of greater than 0.85 were considered unacceptable. Equipment volume was estimated based on manufacturer s specifications, when available. Pod volume was estimated based on current NCAR pod dimensions. Pod volume constraints led to the following radar system designs W-band dual-doppler, dual-polarization radar K a -band dual-doppler, dual-polarization radar Dual-wavelength (K a and W band), dual-polarization radar These conceptual radar systems are discussed in detail in the following sections. For each concept, radar system performance was calculated based on manufacturer s specifications and reasonable engineering assumptions. This was done for both scanning and nonscanning systems. A millimeter wave radar with a continuously scanning beam (360 coverage) was analyzed and rejected due to a loss of sensitivity (~4 db) and unacceptable along track beam spacing (~3 km). A rotation rate of 30 degrees/sec is the fastest which can be achieved while still keeping velocity errors under 50 cm/s and reflectivity 23

24 accuracy under 1 db. This limits the dwell time to about 13 milliseconds, which in turn reduces the number of samples to incoherently average as well as the number of independent samples. Hence, the loss is sensitivity and greater velocity errors. Figure 7: Plan view of NCAR pod mounted below surface of left wing W-Band Dual-Doppler The W-band dual-doppler radar system concept is illustrated in Figure 8. It is comprised of two antennas fed by a single transmitter via a waveguide power divider. Both fore and aft antenna beams are capable of steering in elevation through the use of two rotating reflectors (splash plates). The aft antenna and splash plate produce a beam that is tilted approximately 40 degrees forward of nadir in azimuth and covers the complete lower hemisphere without blockage from the fuselage and up to 60 degrees off zenith. The fore antenna and splash plate produce a beam normal to the fuselage (0 degrees azimuth) that can scan from zenith over 270 degrees without blockage. Both fore and aft radars transmit and receive horizontal polarization and can estimate reflectivity, velocity and spectrum width. In addition, the aft radar can also transmit and receive vertical polarization and consequently has full polarimetric capability, measuring differential reflectivity (Z DR ), linear depolarization ratio (LDR), differential phase, and correlation coefficient at zero lag. 24

25 Figure 8: Single wavelength (W or K a ), dual-doppler, dual-polarization radar concept. NCAR pod shown hanging from left wing. With both splash plates set at the same rotation angle, a dual-doppler wind-field can be calculated for that plane. A dedicated receiver is required for each antenna to down-convert the radar returns to intermediate frequencies (IFs). The IFs are then digitized by commercial off the shelf (COTS) hardware whose original application was Software Defined Radio (SDR). Fully demodulated quadrature data is then sent via high speed Ethernet to the in-cabin data system for further processing, display and archiving. A more detailed explanation of radar system hardware can be found in Sections 4 and 6. The characteristics for the W band system, fixed pointing angle, is given in Table 2. The first column describes the characteristics for the forward-looking horizontally polarized beam, while the second column describes the characteristics for the aftlooking dual-polarized beam. 25

26 Table 2: W-band radar system characteristics (fixed pointing angle) PARAMETER W BAND DUAL- DOPPLER SYSTEM (H ONLY) W BAND DUAL- DOPPLER SYSTEM (H-V) Wavelength 0.32 cm 0.32 cm Antenna Gain 48 db 48 db Antenna Beamwidth Antenna Sidelobe Level < -25 db < -25 db Antenna Cross-Pol Isolation > 30 db > 30 db Peak Power (Circulator Limted) 640 W 400 W Transmit Duty Cycle 5 % 5 % PRF 10 KHz 10 KHz Pulse width 5.0 μsec 5.0 μsec Range Resolution (minimum) 30 m 30 m Unambiguous Range 15 km 15 km Unambiguous Velocity ±8 m/s ±8 m/s Along Track Resolution 30 m 30 m Reflectivity Error 0.4 db 0.4 db Velocity Error (1 m/s Spectrum Width, 10 db SNR) 0.2 m/s 0.2 m/s Receiver Noise Figure 10 db 10.5 db Receiver Bandwidth 5 MHz 5 MHz System Noise Floor dbm dbm 10 km (0 db SNR, 94 millisecond dwell, 0.1 db/km atmospheric attenuation) dbz dbz 5.2. K a -Band Dual-Doppler The K a -band dual-doppler radar system concept is identical to the W-band with the exception of wavelength. The characteristics for the K a -band system, fixed pointing angle, is given in Table 3. The first column describes the characteristics for the 26

27 forward-looking horizontally polarized beam, while the second column describes the characteristics for the aft-looking dual-polarized beam. Table 3: K a -band system characteristics (fixed pointing angle) PARAMETER K a -BAND DUAL- DOPPLER SYSTEM (H ONLY) K a -BAND DUAL- DOPPLER SYSTEM (H-V) Wavelength 0.86 cm 0.86 cm Antenna Gain 39 db 39 db Antenna Beamwidth Antenna Sidelobe Level < -25 db < -25 db Antenna Cross-Pol Isolation > 30 db > 30 db Peak Power 900 W 900 W Transmit Duty Cycle 10 % 10 % PRF 5 KHz 5 KHz Pulse width 20.0 μsec 20.0 μsec Range Resolution (minimum) 30 m 30 m Unambiguous Range 30 km 30 km Unambiguous Velocity ±11 m/s ±11 m/s Along Track Resolution 30 m 30 m Reflectivity Error 0.5 db 0.5 db Velocity Error (1 m/s Spectrum Width, 10 db SNR) 0.2 m/s 0.2 m/s Receiver Noise Figure 5.3 db 5.6 db Receiver Bandwidth 5 MHz 5 MHz System Noise Floor dbm dbm 10 km (0 db SNR, 94 millisecond dwell, 0.04 db/km atmospheric attenuation) dbz dbz 5.3. Dual-Wavelength (W- and K a -Band) The dual W- and K a -band radar system concept is illustrated in Figure 9. It is comprised of a single antenna with a dual-wavelength feed. Polarimetry is achieved 27

28 on both wavelengths through the use of ortho-mode transducers (OMT s). As in the case of the single wavelength system, a rotating reflector is used to steer the beam in elevation. The antenna and splash plate produce a beam normal to the fuselage (0 degrees azimuth) that can scan from zenith over 270 degrees without blockage. The antenna is fed by two separate transmitters, one for each wavelength. Two receivers are required to down-convert the radar returns to intermediate frequencies (IFs). The IFs are then digitized by commercial off the shelf hardware for which the original application was software defined radio (SDR). Figure 9: Dual-wavelength, dual-polarization radar concept. NCAR pod shown hanging from left wing. Fully demodulated quadrature data is then sent via high speed Ethernet to the in-cabin data system for further processing, display and archiving. A more detailed explanation of radar system hardware is discussed in Sections 4 and 6. A key issue for this system is whether to match the W and K a antenna beamwidths. In general, matched beamwidths are desirable because they allow direct data retrieval with the fewest external assumptions. The sensitivity of radar is proportional to the gain (G) and sampling volume (V). A narrower beam has a larger gain (G) because all of the radiated energy is focused within it. Similarly a wider beam has a smaller gain. For example, a typical car radio antenna is a wide beam and low gain. In order to match beams, the W-band beam is widened by a factor of three ( 0.5 deg to 1.5 deg) i.e. gain is lowered. A factor of three enlargement in W-band beam width reduces antenna gain by factor of 9. Interestingly enough, even though the wider beam illuminates a larger volume, the increased sensitivity due to a larger volume is off-set by reduction in antenna gain much more strongly. In summary, the sensitivity of the radar is reduced by a factor of 9 when the beams are matched. 28

29 It should be noted that for the given pod size, the minimum beam width at Ka-band is 1.5 deg. Thus in a matched configuration both radars will have 1.5 deg beam width. As in the case of the HCR, matched beams result in a significant degradation of W-band sensitivity (~9 db) which may limit the scientific usefulness of the W- band radar altogether. The characteristics for the W- and K a -band system (unmatched non-scanning beams) are identical to column 2 in Table 2 and Table Sensitivity Comparisons Figure 10 shows radar sensitivity (non-scanning) as a function of range for the three wavelengths originally considered: K u, K a, and W. For W- and K a -band two configurations are shown; single polarization only: W(H) and K a (H) and fully polarimetric: W(HV) and K a (HV). In all cases, atmospheric attenuation due to water vapor is included; the assumed density is 2.5 g/m3. 29

30 Min Detectable Reflectivity (2.5 g/m 3 ) dbz Range (km) Ku Ka(HV) W(HV) W(H) Ka(H) Figure 10: Minimum detectable reflectivity for a non-scanning radar at the various wavelengths considered in the presence of 2.5 g/m3 of water vapor. Figure 11 shows non-scanning radar sensitivities in the presence of precipitation (10 mm/hr). It is clear that precipitation has a huge impact on W-band sensitivity due to attenuation. 30

31 Min Detectable Reflectivity 10 mm/hr rain dbz Range (km) Ku Ka(HV) W(HV) W(H) Ka(H) Figure 11: Minimum detectable reflectivity for a non-scanning radar at the various wavelengths considered in the presence of 10 mm/hr rain. 31

32 6. Radar Data System Although the proposed radar data system is not particular to millimeter wave radars, the HCR places some unique requirements upon it; among these are size, weight and environmental factors. For the purpose of this document, the radar data system is defined to encompass the following: radar control and timing, data display and archiving, data acquisition, signal processing and housekeeping functions. Housekeeping consists of time, antenna angle, and aircraft attitude, etc. Functionality will be divided as depicted in Figure 12. Wing Pod Cabin GPS Antenna Arinc 429 GPS Receiver Master Clock Analog IF (from radar) Digital Receiver Gigabit Network Radar Products Generator Display Archiver To Antenna Positioner Antenna Controller CMIGITS INS Embedded PC 10/100 Network Radar User Interface Figure 12: HCR data acquisition and storage system. Data acquisition and initial processing are performed in the pod and data is sent for further processing, display and archiving to the cabin over gigabit Ethernet Radar control, preliminary signal processing and data display and archiving will be performed on 19 rack-mount PC(s) located in the cabin, while radar timing, real-time data acquisition and housekeeping will take place in a satellite data system located in the NCAR pod. The cabin and pod data systems will be linked by a high speed network connection. This connection will relay radar control commands from the cabin as well as digital quadrature data, housekeeping and status from the pod. Due to the limited antenna aperture and low peak transmit powers realistically attainable within the confines of the NCAR pod, pulse compression techniques must be employed to attain the required sensitivity. Pulse compression entails transmitting a coded pulse whose bandwidth equals that for the range resolution desired, but whose length is significantly longer than the desired range resolution. In this way, more average power can be transmitted, which increases sensitivity. The radar return is decoded and the desired range resolution is obtained. Some disadvantages of pulse compression include 32

33 increased minimum range due to the longer transmit pulse and smearing of strong signals into adjacent range bins (range-time sidelobes). Plans to mitigate both these issues are discussed below. Pulse compression also introduces complexity into the areas of data acquisition and signal processing. The additional complexity required to both encode the transmit IF signal and decode the IF return signal can be completely handled by the profusion of COTS hardware developed for military SDR applications. This technology implements reconfigurable hardware which has the flexibility to code and decode IF signals in real-time. A current development exists for EOL wind profilers, and we plan to use the expertise gained from this development in an HCR development Data Acquisition Pulse Compression Pulse compression allows radars to increase the average transmitted power by transmitting a longer pulse but without reducing the range resolution of the radar. This is achieved by frequency or phase modulating the transmitted pulse carrier. Received signals are processed using a pulse compression filter which compresses the reflections of the long transmitted pulse to obtain a range resolution that is finer than that of a monochromatic frequency long pulse. In HCR, pulse compression is required to achieve the desired system sensitivity of around -30 dbz at 10 km since the peak power of the millimeter-wave EIKAs and TWTAs is not sufficient. Different coding schemes include linear frequency modulation (LFM), non-linear FM (NLFM) (Cook 1964), phase coding such as a Barker bi-phase code. Frequency modulated signals are often characterized by their time bandwidth (TB) product which is defined as the product of the length of the chirp with the difference between the start and end of the frequency modulation. For example, a transmitter with a duty cycle of 5% will produce a pulse of 5 µs at a PRF of 10 khz. For 30 meter range resolution, we require a LFM modulation of at least 5 MHz and therefore the time bandwidth product of the waveform will be 25. This is typical of the time bandwidth product waveforms we expect to use for HCR. All pulse compression radars suffer from range sidelobes which cause energy from strong reflections to leak into adjacent range cells. Weather phenomena can have significant reflectivity gradients, and reflections from the ground and sea surface can be between db larger than 0.5 mm/h rain depending on the ocean cross section and radar range resolution (Okamtot et al. 1988; Hanado and Ihara 1992). Therefore, range sidelobes must be suppressed by at least 60 db to prevent contamination in adjacent range cells. The simplest pulse compression filter is a matched filter in which the time domain response of the filter is matched to the transmit pulse (Cook and Bernfeld 1967). This is an optimum filter in terms of achieving maximum signal to noise ratio, but suffers from high range sidelobes (approximately -13 db for a linear FM rectangular pulse). Range sidelobes can be suppressed using various techniques at the expense of range resolution or signal to noise ratio. These include weighting the matched filter with an amplitude taper, or using an optimum mismatched filter (Ackroyd and Ghani 1973). Matched filter weighting broadens the main beam of 33

34 the compressed pulse which decreases the range resolution of the system, and also results in a mismatch loss. Inverse filtering will by definition have suboptimal signal to noise ratio and requires a filter length that is several times longer than the pulse which results in large memory requirements in the signal processor. Furthermore, an inverse filter can be designed only for a sampled continuous waveform, not continuous waveforms. However, an infinite set of different sequences can result from sampling reflections that have arbitrary time delays or Doppler shifts which increases sidelobe response. More tolerant inverse filters that optimize the filter response over arbitrary sampling phases have been investigated (Hwang and Keeler 1995). Range sidelobes can also be reduced by predisorting the transmitted signal to reduce the Fresnel ripple in the spectrum (Cook 1964; Kowatsch and Stocker 1985). One technique involves amplitude tapering the transmitted signal; however, this comes at the expense of transmitted power. This technique was used in the ARMAR and APR-2 systems (Durden 1994; Sadowy et al. 2003) to achieve -60 db sidelobes for time bandwidth products of 80 and 40 respectively. Another technique involves pre-distorting the phase of the transmitted signal. These techniques have been studied in low time bandwidth product waveforms (TB < 50) using Hamming weighted compression filters (Kowatsch and Stocker 1985) and sidelobes of around -40 db were achieved in this study. The Doppler tolerance of the waveform must also be considered for dual-doppler systems. In general, range sidelobe suppression is reduced for scatters with significant radial velocity with respect to the platform. Thus, a dual-doppler HCR will require a good Doppler tolerant waveform. For an antenna pointing forward at 40 degrees from nadir, the radial component of the aircraft velocity will be m/s. At W-band (94.5 GHz), this corresponds to a Doppler shift of 81 khz. To achieve range sidelobes of 40 db below the main lobe response, it is likely that a taper will have to be used on the transmit pulse which will decrease the sensitivity of the system. Finally, it is likely that the transmitter will introduce some amplitude and phase distortion into the transmitted signal and to achieve the theoretically predicted sidelobes of a chosen scheme, we may need to dynamically account for this additional distortion during operation. Thus, we will need to choose a pulse compression waveform that achieves the required sidelobe suppression level and Doppler tolerance Short Pulse Pulse compression will be used for HCR to achieve desired sensitivity. Since the system is transmitting a longer pulse, and since it is not able to receive while transmitting, the system will be blind for the duration of the transmitted pulse. For example, a 5 µs pulse, reflections from ranges 0 to 750 meters from the aircraft will not be measured. To retrieve these, we proposed to transmit a short pulse of 200 ns that follows the long pulse. The transmitted pulses will be separated in frequency which will allow us to detect both simultaneously. The ummodulated short pulse power will be sufficient to achieve good sensitivity close to the radar. 34

35 The near field range for a 15 inch antenna at 94.5 GHz is 91 m, and for 35 GHz is 33 m. Since the radar will not be able to receive reflections within the first 30 m (the length of the short pulse), all reflections for a K a -band system will occur outside the near field pattern. However, to recover W-band reflections from 30 to 90 m, a near field correction will need to be applied (Sekelsky 2002) Radar Timing and Control Radar timing and control functionality is split between the cabin and pod data systems. A radar control graphical user interface (GUI) will reside on a rackmount PC located in the cabin. This GUI can be used to configure the operational parameters of the radar, such as PRF, pulse width, antenna pointing angle and rotation rate. These parameters are relayed via a high speed network connection to the satellite data system located in the pod. This information is interpreted by the embedded host processor and is then communicated to the appropriate peripheral (e.g. antenna controller or radar timing generator) Housekeeping Housekeeping is a term generally used to describe information that is essential in the analysis of radar data. For the purpose of this discussion it is reduced to the following: time, antenna rotation angle, and aircraft attitude. All of this functionality can be obtained with COTS hardware. If accurate particle velocities are to be measured, then it is essential to know the position of the radar beam in space very precisely. Given that the radar system is located in a pod hanging from the wing, it is unlikely to experience the same motion as the inertial reference unit (IRU) mounted in the cabin. To obtain more accurate inertial data, it is proposed to mount a separate IRU in the pod. Relatively inexpensive (< $30K) commercial IRUs exist. One such unit is the C-MIGITS III from BEI technologies. It incorporates a 28-state Kalman filter to integrate GPS, accelerometer and gyroscope data into a complete navigation solution. It is also robust enough to handle the environmental extremes of the pod Signal Processing Processing of the time-series data stream will be done in one or more rack-mount PC s located in HIAPER s cabin. Spectral processing, calculating the power spectra for each range gate using the FFT and determining its spectral moments, is highly desirable. It allows maximum flexibility for data analysis. Polarimetric capability is also a highly desirable feature Polarimetric Variables In addition to traditional reflectivity (Z HH ) and Doppler measurements, linear polarimetric observables include differential reflectivity (Z DR ), linear depolarization ratio (LDR), specific differential propagation phase (K DP ), and correlation coefficient (ρ HV ) can also be produced as described by Zahrai and Zrnic (1993). This entails transmitting alternating sequences of horizontally and vertically polarized pulses and receiving both polarizations. Although this 35

36 complicates the design of the receiver and signal processor, the benefits are substantial as was presented in Section Spectral Processing As was seen in Section 3.1.1, spectral processing of millimeter radar returns can yield additional information about cloud properties over simple pulse-pair or autocovariance processing. It can yield an estimate of the mean vertical air motion. For this reason, spectral processing and the storage of raw time series data will be performed Display and Archive Data display and archiving are performed by the in-cabin data system which consists of one or more high performance rack-mount PC s. As previously stated, time-series data will be processed to yield spectral moments and polarimetric quantities. These parameters are displayed locally on a rack-mount LCD flat panel monitor and can be made available to other users via HIAPERS internal data network, if desired. Raw quadrature time-series data will be recorded to redundant hard drives so that no information will be lost in real-time signal processing. This provides the most research flexibility. 36

37 7. Pod Infrastructure The pod infrastructure refers to the various mechanical and electro-mechanical assemblies which must by added to the basic pod structure in order to support a fully functional radar system. The major assemblies consist of the radome(s), rotating splash plate(s) and environmental controls (both pressure and temperature) Radome(s) Depending on the final radar system configuration, either one or two radomes will be required. For the dual-wavelength configuration, a nose radome is required while for a single wavelength configuration, both a nose and tail radome are required. Design and fabrication of these radomes present unique challenges. Not only must they conform to the aerodynamic shape defined by the NCAR pod, but they must also withstand the temperature, pressure, shock, vibration and flutter generated by the pod s environment. In addition, they must pass millimeter wave frequencies relatively undistorted in phase or amplitude. These requirements are very likely conflicting. An ideal radome would have a radius of curvature which matches the phase fronts of the antenna and a constant wall thickness (equal to λ ε /2, where λ ε is the wavelength in the radome dielectric). Both these properties ensure that the transmitted signal incident normal to the radome will pass through the radome undistorted in phase, amplitude and direction. This is especially difficult to achieve in practice on the NCAR pod as the dimension and shape of the radome are governed solely by aerodynamic considerations. The curvature of the radome will cause the beam to bend, unless the thickness of the radome is tapered to compensate for the change in beam angle. In addition, "The preferred separation between the radome and the antenna is greater than 10 wavelengths. This minimal separation ensures that the radome is in the radiating near field and not in the reactive near field where other forms of electromagnetic coupling between antenna and radome wall are possible"(currie and Brown 1987). This criteria must be examined once the final nose cone and tail cone shapes of the NCAR pod are known. "Maximum transmission at any desired angle, theta, is achieved, for a monolithic wall, by choosing the thickness of the wall, d, as follows: where, λ d = N, 2 1/ 2 ( ε sinθ ) for N > 0 N Ζ θ = incident angle ε = the relative dielectric constant of the wall material λ = the free space wavelength. This equation shows that the required wall thickness for maximum transmission increases with increasing incident angle and decreasing dielectric constant." (Currie and Brown 1987) Based on the present NCAR pod shape, the incident angle will be more shallow than the desired 90 degrees to the radome s surface. The added radome thickness to compensate for this may or may not be a factor, given that the structural requirements of the radome will likely require substantial wall thickness anyway. The 37

38 specifications of the radome are critical to obtaining the desired radar performance. For example, a hydrophobic coating should be specified to allow water to be shed during operation in order to minimize radome attenuation in the presence of precipitation Rotating Splash Plate(s) As the radar concept figures show, one or more rotating splash plates are required to position the beam in the desired orientation. The plates themselves will likely be constructed of aircraft grade aluminum, machined to a flatness of at least a tenth of a wavelength. Precise positioning of the splash plate relative to the antenna is required to avoid beam distortion, and spillover. A direct-drive DC stepper motor coupled with an optical shaft encoder would offer the simplest solution to positioning the splash plate precisely. A brushless DC motor would also be a good choice if more torque and acceleration are required. Several feasible options exist for positioning the splash plate and the choice should probably be deferred until the detailed design phase of the project Environmental Control The flight profile for HIAPER extends from sea level to approximately 51,000 feet The pressure and temperature extremes that exist over this altitude range present unique challenges to radar system electronics, especially the high-voltage sections of the transmitter. The following sections discuss in some detail overcoming these challenges. It should be recognized that a far more detailed thermal analysis is required before a suitable environmental control solution can be found. This is beyond the scope of this document Pressurization As altitude increases, both air pressure and air density decrease. For example, at sea level the air pressure is 760 mm Hg and the density is 1.12 kg/m3, but as 51K ft. the air pressure and density have fallen to about 83 mm Hg and 0.2 kg/m3. This decrease in air density actually has critical implications for both cooling and avoiding arcing in electronics. In the case of electronic arcing, air acts as a dielectric to insulate the voltage potential between two electrical contacts. The minimum separation between two contacts to avoid arcing is called the spark gap. As the density of the air decreases with altitude, the dielectric constant of the air also decreases and consequently its insulating properties. In order to avoid arcing, the spark gap must either be increased, the dielectric constant of the insulator increased or the potential decreased. Decreasing the potential and increasing the spark gap is usually not practical when buying commercial or semi-custom electronics. The most practical solution is to increase the dielectric constant of the insulator. This is accomplished by either immersing the critical components in a liquid, such as Diala oil or constructing a pressure tight vessel around the components. For an airborne application, where weight is critical, a pressure vessel is the better choice. The pressure vessel will likely house the high voltage electronics used in the 38

39 transmitter high voltage power supply and modulator. In addition, it is likely that the waveguide will need to be pressurized with 2-3 psi. of Nitrogen Temperature Control As stated in the previous section, the decrease in air density with altitude also affects the ability to convectively cool electronic components. The ability to convectively transfer heat depends on a variety of factors, such as surface geometry, velocity of the fluid, fluid density and fluid temperature. In this case ambient air is used as the fluid and higher air velocity, lower air temperatures and the higher air density, all aide in convective heat transfer. In general decreasing ambient temperature offsets the effects of decreasing air density up to an altitude of about 9 km (~30Kft.). Above this altitude, the decrease in air density outpaces the decrease in ambient temperature and heavily limits convective cooling capability. This effect can be somewhat overcome by using higher cfm (cubic feet per minute) fans, larger heat sinks, or even slip fans. Conductive cooling of critical components is also attractive, since a large thermal mass at fairly low temperatures (-55 C at 51 Kft.) is available. Unfortunately, cooling components is not the only problem. Some of the electronic components, such as the ferrite switches, are only specified to operate down to 0 degrees Celsius. Since temperatures will be around -55 degrees Celsius at ft, these components will likely require heating during parts of a typical mission. Locating such components in the pressure vessel with the high voltage portions of the transmitter may eliminate the need for much additional heating. The NASA CRS radar which flies on the ER-2 requires no extra heating or cooling. A detailed thermal analysis will need to be performed to properly gauge the cooling and heating requirements for the pod mounted radar system electronics. 39

40 8. System Calibration Obtaining an accurate system calibration is essential if the HCR is going to fulfill its scientific mission. Both an absolute calibration, which includes reflection measurements from a target of known radar cross-section, and an internal calibration, which tracks changes in receiver gain with temperature, are required Internal Calibration For a polarimetric radar system, internal calibration is important to track changes in the receiver gains. Common schemes include well calibrated test signal, noise source, or injecting a small part of the transmitted signal into both receivers. All schemes are feasible, but using the transmitted signal to track differential changes in the receiver gains also allows one to monitor the transmitted pulse. This is important when using frequency modulation since one can then compensate for phase nonlinearities due to the transmitter in the compression filter. In this calibration scheme, it is important to achieve high isolation between the transmitter and the receiver. Coherent leakage can cause fluctuations in the calibration signal. To achieve fluctuations less of ±0.05 db, the unwanted leakage of the transmitted pulse through other parts of the circuit must be 45 db below the calibration signal. In addition to monitoring differential changes in the receiver gain, we propose to monitor the transmitted pulse power during operation. This will allow us to calibrate to the system as it is likely that the transmitted power will change with temperature system change during flight Absolute Calibration Absolute calibration will require the use of a trihedral corner reflector positioned in a clutter and multipath free environment. It is likely to be a ground-based measurement, but could also be done in the air if the corner reflector was positioned high enough above the ground such that ground contamination in range sidelobes could be negligible. Propagation affects such as attenuation will need to be known to obtain a good calibration. This may require in-situ measurements. Ocean surface measurements have also been used to calibrate W-band airborne and spaceborne radars (Durden et al. 1994). This technique has been proposed as a calibration reference for Cloudsat (Stephens et al. 2002). Measurements of the ocean surface at 94 GHz have shown good agreement with quasi-specular ocean surface scattering models for low incidence angles (Li et al. 2003). However, these measurements have shown that attenuation due to water vapor and oxygen absorption affect calibration targets such as the ocean surface and thus vertical in situ profiles are required to fully calibrate the radar. Furthermore, the normalized radar cross section of the ocean surface at these incidence angles is dependent on wind speed, thus requiring in situ measurements of near surface winds by dropsondes (Li et al. 2004). 40

41 9. Test and Measurement Equipment Millimeter-wave system development at EOL requires an investment in test and measurement equipment. This investment is also necessary for supporting deployments of HCR by EOL staff. Such deployment tasks include calibration, troubleshooting, and maintenance. Required equipment will include a spectrum analyzer, a signal generator, a power meter, and low loss test and measurement cables. 41

42 10. Proposed HCR Development Team EOL has had a long and proud tradition of developing state of the art centimeter wavelength radars covering X, C and S bands (3cm, 5 cm and 10 cm wavelengths). These include the ground based CP-3, CP-4, CP-2, and S-Pol. radars and the airborne ELDORA. More recently, EOL ventured into development of the first millimeter-wave radar built at NCAR. The K a -band radar (8mm wavelength) extends the scientific capability of S-Pol through a second wavelength. The matched beams of S-Pol and the K a -band radar allow researchers to study dual-wavelength scattering. This sets S-Pol apart from other transportable radar systems available to NSF researchers and reestablishes S-Pol as a unique remote sensing tool. Through the K a -band radar development, RTF engineers have gained valuable experience designing, fabricating and fielding a millimeter-wave radar. Furthermore, some millimeter-wave test and measurement equipment was purchased for this endeavor, and thus a partial infrastructure for developing millimeter-wave radar systems at NCAR already exists. We propose that EOL leverages its extensive radar engineering experience to develop a HIAPER Cloud Radar. Since EOL staff will be responsible for deployment and maintenance of the system, we feel that the experience gained in developing the radar will allow our engineers and technicians to more effectively support the system during field campaigns. However, we do realize that our experience with millimeter-wave systems is limited, and thus we have discussed collaborations with various groups to contribute and guide us through the development. A number of millimeter wave radars (CPRS, ACR, CMR) have been built at the University of Massachusetts Microwave Remote Sensing Laboratory (MIRSL). Alumni of MIRSL have gone on to build other millimeter wave radars at companies such as Prosensing, and research institutes such as NASA Goddard and the Jet Propulsion Laboratory (JPL). It is likely that a private company would be interested in building a radar for HCR; however, we feel that contracting the development of the whole system out to such a company would not be beneficial to the development of radars at NCAR. Furthermore, maintenance and deployment of the HCR facility may be complicated due to future schedule conflict between a private company and research establishments such as the EOL. Engineers and scientists at NASA Goddard, JPL, CSU, University of Wyoming, and NOAA have expressed an interest in providing guidance in a millimeter wave radar development. In addition, a number of radar engineers from private companies who have been involved in millimeter wave radar development for many years have expressed an interest in consulting with EOL on the development of HCR. Such a team would be responsible for reviewing designs and making suggestions and would provide the required technical oversight of a HCR development. It should also be noted collaboration with JPL or other foreign research establishment such as DLR, Germany can be carried out only if we carefully resolve issues that are related to International Traffic in Arms Regulations (ITAR). 42

43 11. Cost A cost summary appears in Table 4. The cost for three different radar configurations is considered: W-band dual-doppler, K a -band dual-doppler and dual-wavelength (W- and K a -bands). Cost was broken down into the following broad categories: hardware, pod and infrastructure and labor. Hardware Costs (incl. O/H) Table 4: Cost Summary W-band Dual- Doppler K a -band Dual- Doppler Dual Wavelength (W/K a -band) Pod & Infrastructure (incl. Radome(s)) 443, , ,252 Antenna(s) 60,000 50, ,000 Dual-wavelength second feed for matched beams ,000 Transmitter(s) 215, , ,000 Receiver 450, , ,124 Data System (incl. archive & display) 158, , ,943 Test Equipment 167, , ,242 Misc. 95,782 95,782 95,782 Ground Platform 45,000 45,000 45,000 Total Hardware Cost 1,635,332 1,505,303 2,408,343 Labor (incl. O/H & Benefits) 1,801,706 1,801,706 1,901,246 Total Cost 3,437,038 3,307,009 4,309,589 Hardware costs are fairly self-explanatory and include overhead. The cost of manufacturing a second feed for the dual-wavelength, matched beams option can be reduced from $150k to $125k, if this feed was designed and manufactured at the same time as the first. The miscellaneous category includes consulting fees, travel, environmental testing and an IRU. The ground platform will allow the system to be used as a ground-based upward looking radar for testing and science studies. This cost includes the necessary mounting structure and repackaging for ground-based operation. Labor costs are based on the assumption that the HCR transmitter and antenna subsystems will be outsourced, but the data system and receiver subsystems will be designed and fabricated by NCAR. In addition, NCAR would assume responsibility for radar system design and system integration. These assumptions were made for cost purposes only. As UCAR salaries, benefits and overhead was applied, the labor cost 43

44 likely represents the high-end of the spectrum and would likely be reduced if more university collaboration were desired (assuming graduate students are involved). The pod and infrastructure category includes radomes, splash plates, motors, encoders, a pressure vessel and equipment mounting infrastructure. 44

45 12. Survey and Recommendation This section describes a two-step procedure that was used to narrow the specifications for the HIAPER cloud radar system. The first survey primarily focused on basic scientific requirements and radar sensitivity specifications. The results of the survey preferred both detailed microphysics and 2-dimensional wind retrievals. A polarization Doppler radar with dual-wavelength and dual-beams is capable of retrieving microphysics and 2- dimensional winds. However, the size of the wing-pod limits the implementation to only one of the following choices: dual-wavelength or dual-beam, but not both. Hence, a second survey was designed to narrow down the choice between the dual-beam or dualwavelength option. The majority of the second survey results favored the dualwavelength option rather than the single-wavelength dual-beam configuration Survey Technical specifications of a radar primarily depend on the scientific expectations of the research community. Many of the researchers in the atmospheric remote sensing area are also familiar with radar system requirements to accomplish a specified scientific requirement, and also researchers tend to capitalize on multi-sensors to enhance products that are obtained using remote sensing observations. Thus, a questionnaire addressed the following important issues: (1) scientific requirements, (2) system requirements, and (3) the usage of radar with other sensors on board the HIAPER platform. The survey questionnaire was sent to a wide spectrum of researchers in atmospheric remote sensing. The following is the list of the survey questions: 1. What are your scientific objectives in using the proposed HCR (e.g., cloud and precipitation, ocean wind sensing, radiation, energy balance, vapor and cloud liquid, surface moisture and cloud liquid)? 2. For each of your scientific objectives, what are the system requirements of the HCR in order to fulfill your scientific objectives? 2.1. Reflectivity sensitivity at 10 km, i.e., detection limit for liquid water content or ice water content such as 0.1 g/m What is the required operation mode (e.g., scanning or staring mode)? What are the Range gate spacing, beam width, temporal resolution, and along track resolution? 2.3. What is preferred wavelength (K u, K a, W)? 2.4. Is dual-wavelength required? If yes, what is the preferred combination of wavelengths? 2.5. Is Doppler capability required? If yes, what is the preferred resolution, i.e., cm/s? 3. How will cloud radars be used in conjunction with the other proposed instruments on HIAPER, such as a temperature sounder, LIDAR, and various in situ probes? This survey was distributed to about 20 members of the cloud radar user community in December The survey results indicated a common preference for a narrow beamed W-band radar with polarimetric and Doppler capabilities. Additional 45

46 capabilities included a second wavelength and/or dual-doppler winds. A summary of results obtained as a response to the survey questionnaire is summarized in the following slide. Research Community Survey Scientific Goals Microphysics phase (liquid/ice) discrimination, particle sizing, LWC and IWC Kinematics vertical velocity Cloud radiation studies energy balance System Requirements Sensitivity: -25 to -35 dbz at 10 km Resolution: 20 to 100 m Preferred wavelength: W-band (3 mm) Both co- and cross-polarization measurements Doppler capability Additional Capability Options 2-D winds (dual-beam) or Enhanced microphysics - second wavelength Ka-band (8 mm) Slide 1 Based on the survey results, EOL staff considered a design incorporating a pod-based radar with scanning, dual-wavelength, dual-doppler, and polarization capabilities. Owing to the size limitation of the HIAPER wing pod, it is not possible to physically fit all of the required hardware components in the pod for such a state-of-the-art radar system. In addition, the tradeoff among sensitivity, Doppler velocity accuracy, and spatial resolution suggested that a staring mode, with the option of scanning, would be the optimal way to operate HCR. Therefore, a total of four possible designs were explored, as follows: Option 1: W-band (3 mm) radar system with full polarimetric (on the AFT antenna) and 2-D dual-doppler winds capability. This system contains one transmitter and two antennas. This system has a beamwidth of ~0.5 degree. Option 2: K a -band (8 mm) radar system with full polarimetric (on the AFT antenna) and 2-D dual-doppler winds capability. This system contains one transmitter and two antennas. This system has a beamwidth of ~1.5 degree. Option 3: Dual-wavelength (W and K a ) radar system with full polarimetric and single Doppler capabilities. This system contains two transmitters and one antenna. In order to achieve the highest system performance (i.e., best sensitivity on both W and K a ), the beams are unmatched between W- (0.5 degree) and K a - (1.5 degree) bands. Option 4: Same as Option 3 but beams are matched between the W- and K a -bands (~1.5 degree). In this configuration, the W-band sensitivity will be ~9 db less than that offered in Option 3. Even though the wider beam illuminates a larger volume, the increased sensitivity due to a larger volume is off-set more strongly by a reduction in 46

47 antenna gain. As discussed in section 5.3, the sensitivity of the radar is reduced by a factor of 9 (i.e., 9 db) when the beams are matched. In all four designs, the radar is housed in a pod hanging beneath the HIAPER left wing as illustrated in Slide 2. System Configurations, Side View (Looking into the fuselage from left of the aircraft, the HCR pod is hanging beneath the left wing) Left Wing Pod Pylon Antennas and electronics W- and Ka-band, Single-Doppler (Options 3 and 4) Vertical Beam Second Beam (Polarimetric) Left Wing Pod Pylon Antennas and electronics W- or Ka-band, Dual-Doppler (Options 1 and 2) Slide 2 Vertical Beam (Polarimetric) Side view of the wing and pod. The green and gray beams extending from the pod show the fore and aft beams (options 1 and 2) and the fore beam (options 3 and 4). All four options include dual-polarization, a single-doppler, and a time series data recording capability to allow Doppler spectrum computation in the post-analysis phase. The proposed physical layout of a radar system in a wing-pod limits mixing the above described options, i.e., only one of the options can be implemented. All four options meet the common requirements of the survey and provide additional capabilities either focusing more on kinematics (options 1 and 2) or microphysics (options 3 and 4). A brief summary of these options is listed in Slide 4. 47

48 Slide 3 A perspective view of the pod from top of HIAPER above the left wing. The axis show the scanning limits of HCR for the fore and aft beams as viewed from the front of the aircraft. System Options* Option 1: W-band, Dual-Doppler Option 2: Ka-band, Dual-Doppler Option 3: Dual- Wavelength (W + Ka), Single-Doppler Option 4: Option 3 with matched beams Radar Parameters -27/-31 # 10 km 30 meter resolution Vnyq=±8 m/s 0.5 degree beamwidth (30 x km) W: km Ka: km Vnyq=±8 m/s and ±11 m/s 0.5/1.5 degree beamwidth Applications 2-D winds Cloud particle phase (liquid/ice) Cloud and drizzle -26/-27 # 10 km 30 meter resolution Vnyq=±11 m/s 1.5 degree beamwidth (30 x km) 2-D winds Cloud particle phase (liquid/ice) Cloud, drizzle and light rain Cloud particle phase (liquid/ice) LWC profile Particle size distribution W: km Ka: km Vnyq=±8 m/s and ±11 m/s 1.5/1.5 degree beamwidth Cloud particle phase (liquid/ice) LWC profile Particle size distribution Issues to Consider Higher attenuation in presence of liquid Mie scattering for water/ice particles > 0.5/1 mm More uncertainty in microphysics Slightly Lower sensitivity than W- band 30 x km More uncertainty in microphysics No dual-doppler Cost Unmatched beams will add uncertainties in microphysics retrieval No dual-doppler Cost Less uncertainties in microphysics retrieval but W- band sensitivity is km Slide 4 *All options offer at least single-doppler, full polarimetric, and time series recording capability # The first reflectivity number represents the sensitivity for the dual-pol antenna while the second number represents the single-pol antenna. System options presented to the HIAPER Advisory Committee (HAC) on 11 May