DESIGN AND OPERATIONAL ASPECTS OF A
REDUCED-APERTURE 449 MHZ WIND PROFILING RADAR
Herb Winston
John Neuschaefer
Vaisala Incorporated
Boulder, Colorado, 80301, USA
Scott McLaughlin
Daniel Wolfe
National Oceanic and Atmospheric Administration
Environmental Technology Laboratory
Boulder, Colorado, 80305, USA
Introduction
The United States Air Force operates a network of tethered “lighter-than-air” aircraft (aerostats) along the
southern U.S. border. Having volumes of up to 625,000 cubic feet and supporting flight altitudes up to 15,000 feet, each
Tethered Aerostat Radar System (TARS) carries aloft a surveillance radar that is used to observe aircraft entering U.S.
airspace (Figure 1). Like any aircraft, aerostats can only operate safely within certain operational constraints. Wind is
often the most important factor determining when it is unsafe to fly.
Over ten years ago the National Oceanic and
Atmospheric
Administration’s
(NOAA)
Environmental
Technology Laboratory (ETL) began working with the TARS
operators to study whether the use of a radar wind profiler
(RWP) could improve both total time aloft and overall safety to
the Aerostat and ground crew. A 404 MHz RWP (Figure 2)
using a Yagi antenna array was temporarily installed at a
TARS site in southern Arizona (Moran et. al., 1989). The RWP
collected wind profiles every half hour and sent them directly to
the TARS site where they were used to assess atmospheric
flight conditions. The project was considered a success.
Funding to install a permanent RWP became available in late
1999. The Air Force asked NOAA ETL to act as system
integrator; to design, specify, purchase, and assemble a
complete RWP with advanced signal processing and remote
data display and control capability.
Figure 1. Aerostat system deployed by the US Air Force
System Description
Figure 2. NOAA ETL 404 MHz RWP with Yagi antenna
array installed for testing in support of TARS in 1989
Several options were evaluated to determine the
best overall hardware, software, and location for a RWP
demonstration program. For cost considerations, the Air
Force wanted to use as many commercial-off-the-shelf
(COTS) parts as possible. The requirement for high data
quality and a 449 MHz frequency meant that new parts
and software would be required. There was also interest
in placing the RWP as close to the aerostat as
possible. This led NOAA to a “hybrid” RWP design
consisting of commercially available parts augmented
with NOAA-designed monitoring hardware and ETL’s
advanced Signal Processing Software (SPS). It was also
determined that the best location for the RWP
demonstration was approximately 3 km from the TARS
site at Ft. Huachuca, Arizona, USA.
Unlike the 404 MHz tropospheric wind profilers
operated in NOAA’s Wind Profiler Network and similar international instruments, the 8 km altitude requirement
established by the Air Force allowed for a smaller antenna array aperture and lower transmit power. This “reducedaperture” system operates at 449 MHz and utilizes two orthogonal arrays of twelve, 18-element coaxial-collinear (Co-Co)
antennas. A 2 kW solid-state amplifier drives a 36 square meter antenna array which has a row-to-row phasing of 60
degrees implemented using delay cables and RF switches to change
array axis and phasing. An air-conditioned 2.4 by 3 meter shelter is located directly adjacent to the antenna, which
houses the computers, receiver, amplifier and the beam steering unit. COTS components integrated into the system are
from the Vaisala Inc. LAP® wind profiler product line. These include a Radar Processor Unit consisting of a PC with
control and signal processing boards, Receiver/Modulator Unit, Interface Unit, Co-Co antennas, and a Beam Steering
Unit.
The radar processor runs Vaisala LAP ®-XM control and signal processing software. Most of the items were
selected because they were commercially available and are compatible with NOAA’s SPS software. The deployed
antenna and system electronics are shown in Figures 3
and 4, respectively.
A specialized Hardware Monitor system was
developed by NOAA ETL to support the maintenance
and operation of the RWP. The monitoring is mostly
performed by the use of an A/D board located in the
radar processor PC. One unique item developed to
support the radar was a 6-way power divider that utilizes
an RF divide/combine technique allowing the use of
individual RF power reflect loads. Each load's individual
temperature is directly monitored by the Hardware
Monitor. This monitoring technique allows noninvasive
detection of failed antenna components.
Figure 3. 449 MHz Reduced-Aperture RWP antenna, TARS
site in Ft. Huachuca, AZ, USA.
NOAA Advanced Signal Processing
The NOAA/ETL Signal Processing Software (SPS) is responsible for
generating meteorological products from RWP averaged-Doppler spectra. It
differs substantially from the traditional signal processing system, in which
signal processing is linear and sequential throughout and where one signal per
Doppler spectrum is detected and reported. Fundamental to the NOAA signal
processing is a recognition that, even with attempts to suppress possible
contamination from ground clutter, RFI, spurious signals, noise, etc., RWP
averaged-Doppler spectra may contain multiple spectral peaks. The SPS
signal processing removes the constraint of being restricted to a single data
channel while addressing the possibility of multiple signals in a single
spectrum. Multiple data channels (at different ranges, at different times, and on
different antenna beams) are analyzed from the spectra level up to the
meteorological parameter calculations to determine which signals are windinduced, even in the presence of different kinds of contamination.
The SPS software runs on a separate PC from the radar processor
and is collocated at the radar site. The SPS consists of four individual signal
processing modules, each processing a different part of the radar data stream
beginning with the averaged spectra level and ending with meteorological data
products. All of the data for each level is managed by complementary modules
providing specialized high-speed database. An additional routine handles
spectral data ingest into the database from the radar processor computer
(Wolfe, et. al., 2001).
Figure 4. 449 MHz RWP system
electronics, Ft. Huachuca, AZ, USA.
The SPS uses time-height continuity, opposing beam continuity, and other parameters to aid in objectively
determining which peak (if there is more than one) is most likely to have originated from clear-air radar back scatter. After
the winds have been calculated, additional quality control software is run to check the resulting wind profiles again for
time height continuity. A “confidence” parameter is calculated and carried along with the data at each level. This
parameter can be set to limit the presentation of the data at differing confidence levels. The wind data is sent directly to
the TARS site for real-time presentation on a PC. Data is updated every five minutes utilizing a 15-minute sliding
window. The data is also ingested by a specialized software suite (provided by a third party) that analyzes and displays
stresses placed on the aerostat by the measured wind profiles.
Wind Data
Figure 5. shows an example of data processed using the standard LAP ®-XM consensus method (top panel) and
NOAA signal processing system (bottom panel). Differences in the data density are due to a fifteen-minute averaging with
a five-minute sliding window used in producing the SPS figure and a thirty-minute block consensus averaging for the
standard method. Though it would be possible to produce a fifteen-minute consensus average, it would not provide a
fair comparison; A fifteen-minute consensus would limit the amount of data used, while the SPS method provides for a
larger quality control window and spatial cross beam checks in quality controlling the data. Agreement between the two
processing methods is quite good. Validation (not shown) with balloon and 915 MHz profiler data both located ~7 miles
NNW of the TARS site shows general agreement despite the spatial separation, which is compounded by the complexity
of the topography including the Huachuca Mountains rising to more than 2000m directly west of the profilers.
Figure 5. Sample data from a Reduced-Aperture 449 MHz Radar Wind Profiler collected from
0600 May 20, 2001 to 0600 hours, May 21, 2001, Ft. Huachuca, AZ, USA. Top panel shows
thirty-minute consensus averages produced from LAP-XM software. Lower panel shows
fifteen-minute time averages with a five-minute update cycle produced from NOAA/ETL’s Signal
Processing System (SPS).
Future Plans
Installation of the Ft. Huachuca system represents the Air Force’s first step in deployment of a 449
MHz RWP network in support of its eleven operational Aerostat sites. The next phase calls for installation of a
second RWP system at the Air Force’s Aerostat site in Cudjoe Key, Florida, USA in early 2003. Unlike the
recently deployed system, NOAA plans to incorporate Vaisala’s new Digital IF receiver that is scheduled for
product launch in September, 2002. This new platform provides several system enhancements over the
current LAP® architecture, originally developed at NOAA and licensed to Vaisala as part of a Cooperative
Research and Development Agreement. The new receiver incorporates a 14-bit A-D converter operating at
the 60 MHz IF frequency, digital demodulation and matched filtering, and completely redesigned PCI-based
radar processor that enables several improved signal processing and wind derivation techniques into the
operational LAP®-XM software architecture. These new techniques include a multiple-peak selection
algorithm (Griesser, 1998), wavelet filtering of time-series data (Jordan, 1997), and a “running consensus”
algorithm.
In conjunction with the deployment of the Cudjoe Key system, an inter-comparison program of the
various signal processing techniques is planned under the auspices of NOAA. The objective of this study is to
establish an optional signal processing and wind determination suite that is best-suited to satisfy the
operational requirements for the Air Force’s Aerostat program. Wind profilers operating at 449 MHz and
located at the Boulder Atmospheric Observatory (BAO) tower in Erie, CO, USA and at Ft. Huachuca, AZ,
USA will be used for this inter-comparison study.
References
Griesser, T., H. Richner (1998): Multiple Peak Processing Algorithm for Identification of Atmospheric Signals
in Doppler Radar Wind Profiler Spectra, Meteorologische Zeitschrift, 7, pp. 292 – 302.
Jordan , J. R., R.J. Lataitis, and D.A. Carter (1997): Removing Ground and Intermittent Clutter Contamination
from Wind Profiler Signals Using Wavelet Transforms, J. of Atmospheric and Oceanic Tech., 14, p. 12801297.
Moran, K.P., R.G. Strauch, K.B. Earnshaw, D.A. Merritt, 1989, Lower Tropospheric Wind Profiler, Amer. Met.
Soc., 24th Conference on Radar Meteorology, March 27-31, Tallahassee, FL pp 728-731.
Wolfe, D. E., B. L. Weber, T. L. Wilfong, D. C. Welsh, D. B. Wuertz, D. A. Merritt, NOAA Advanced Signal
Processing System for Radar Wind Profilers, Amer. Met. Soc., 11th Symposium on Meteorological
Observations and Instrumentation Jan 2001, Albuquerque, NM, pp 339-344.