School of Electrical

1
School of Electrical Computer Engineering
Weather Radar Research at the University of
Oklahoma
An interdisciplinary group between the
School of Meteorology
and
at the University of Oklahoma
http//arrl.ou.edu
Presented at LMCO, January 2006
2
Overview of Research Activities
http//arrl.ou.edu
Winter Clouds/Icing
Tracking
Dual-Polarization Quantitative Precipitation
Estimation Microphysics Severe Storms
Clear-Air Turbulence
Advanced Profiling Radar
Tornado Detection
Multi-Function Radar
Refractivity Retrieval
Phased Array Techniques Hardware/Digital
IF Signal/Array Processing
Inter-Disciplinary Education Research
In-Situ Validation
Clutter Mitigation
Our group works in close collaboration with the
National Severe Storms Laboratory
3
Adaptive Scanning Strategy Based on Beam
Multiplexing (BMX) Using the Phased Array Radar
Beam Multiplexing
Step Scan
BMX optimizes the use of radar resources and data
quality
For the same scan time, BMX provides data with
lower statistical error and uncertainty compared
to those obtained by the scanning strategy
similar to the one used with NEXRAD.
4
Possible configurations of Spaced-Antenna (SA)
interferometry to measure wind, shear, and
turbulence
Transverse wind, shear and turbulence measurement
with interferometry

• SPY-1 three channels
• Sum
• Azimuth difference
• Elevation difference

Azimuth SA
Elevation SA
Dual-beams to separate shear and turbulence
Auto- and cross-correlations
Cross-correlation peak shifts due to signal
delay passing over antennas from R1 to R2
5
Measurement of Moisture/Temperature Fields Using
Ground Clutter Targets

• Electromagnetic waves delayed by factor related
  to atmospheric refractivity and this delay can be
  measured by radar
• Moisture/temperature fields related to
  refractivity
• Currently implemented on the Phased Array Radar
  and KOUN
• (research WSR-88D) radar located in Norman

KOUN
Ground Clutter
Phased Array Radar
Ground Clutter
PAR Moisture Field Compared to Mesonet
6
Target Tracking Emerging Non-Linear Stochastic
State Estimation Techniques
Recently presented at NASAs Goddard Space
Flight Center.
Networked tracking algorithms based on particle
filtering will provide improvements to these
areas that require the analysis of dynamic
movements, particularly when the underlying
dynamic models are non-linear and when the
measurement noise is non-Gaussian. As discussed
in the most recent literature, particle filtering
is defined as an emerging Bayesian method based
on sequential Monte-Carlo non-linear state
estimation techniques.
Left PAR detections made in a 90 degree sector
looking towards the OKC airport, Sept 15, 2005.
Right Specific tracks are formed for data
collected Sept 19, 2005.

• Why is particle filtering important for PAR?
• Tracking in the presence of strong maneuvers. As
  seen above, when the object denoted by the green
  curve makes a sharp change in course,
  conventional techniques (such as the Extended
  Kalman Filter) cannot maintain track.
• One step ahead prediction algorithms for optimum
  beam-steering

7
Real-Time Signal Processing
Inputs
Data from various radar platforms such as (1)
CASA or (2) phased array
By combining existing strengths in hardware and
signal processing, technology for the complete
system is now available for the first time.
1.
Cooling Fans
Heat Sinks
2.
Design Sequence for Circuit Boards at OU
Power Supply
DSP Control Board Signal Processor
1. Device Characterization and Preparation for
Layout.
RF Amplifier / Receiver
Outputs
2. Circuit Board Design Layout
Advanced tracking algorithms for optimum beam
steering for maneuvering targets enabling
closed loop control.
The unique advantage of this endeavor is
real-time computational intelligence.
Innovative WX processing algorithms are combined
with customized hardware to create
computationally intelligent systems that are
readily deployable.
3. Finished Circuit Board with DSP
By leveraging the teams previous experience, as
depicted by the prototype above and the circuits
on the right , prototypes for remote sensing
systems can be explored at OU.
8
Next Generation Radar Simulator for Advanced
Signal/Array Processing Studies
Example simulator output with mixed weather and
point target

• Would like to develop/study advanced signal/array
  processing algorithms using realistic, Level-I,
  time-series radar data
• Real experimental data are not controlled or
  flexible
• Solution is an advanced radar simulator using
  high-resolution numerical weather simulation data
  as input fields

Point Target
Weather
Standard Beamforming
Simulated multi-function phase array antenna 155
elements 17 degree coverage
Adaptive Beamforming
Possible Applications

• Multi-function (weather surveillance and target
  tracking)
• Neural network training under varying conditions
• Simulation of advanced radar designs (phased
  arrays, frequency agility, etc.)
• Filter design for clutter mitigation (e.g.,
  matrix clutter filter)
• Resolution enhancement using interferometry/deconv
  olution
• Optimization of beam scanning strategies using
  phased array radars

Conceptual diagram of model-based radar simulator
capable of simulation of realistic, complex
time-series data
9
Advanced Algorithms for Accurate and Early
Tornado Detection
Statistical analysis of tornado signatures
The innovative neuro-fuzzy system exploits
available tornado signatures and provides
self-learning and robust capability
More accurate tornado detection is obtained using
the neuro-fuzzy algorithm
10
Clutter Mitigation
Study of non-stationary clutter from wind turbine
farms
Adaptive array processing exploited for
mitigation of biological and ground clutter
Wind Farm Clutter
Spectral/wavelet processing may prove important
for analysis and mitigation of wind turbine
clutter
conventional array processing
Matrix Clutter Filters for Phased Array Radars
Following the work of Urkowitz and Owen (1998),
we have experimented with clutter mitigation
techniques for agile-beam radars
adaptive array processing
11
Polarimetric Radar for QPE and Validation Using
In-Situ Instrumentation
Electromagnetic wave scattering
from non-spherical hydrometeors is different for
horizontal and vertical polarization, which can
be used to accurately characterize
cloud/precipitation microphysics
KOUN
A 2D video disdrometer measures the size, shape,
density, orientation, and fall speed of
precipitating particles.

• differential reflectivity (ZDR)
• cross-correlation coefficient(?hv)
• specific differential phase (KDP)

12
Imaging Studies of the Atmospheric Boundary Layer
Range IMaging (RIM) is a multi-frequency
technique that offers a novel means of improving
the range resolution of radar measurements
Using multiple, independent receivers, digital
beamforming has been used to image the atmosphere
Radar developed by the Univ of Massachusetts.
Comparison of Radar Backscatter from a wind
profiler (upper panels) and an FMCW radar (lower
panel)
A 2.5-min sequence of three-dimensional conical
images of isolated precipitation (blue) and
clear-air turbulent (yellow) echoes
13
Summary of Expertise
Dr. Guifu Zhang (405-325-3507, guzhang1_at_ou.edu) Mo
deling of wave propagation and scattering in
geophysical media, remote sensing techniques for
environmental monitoring, radar polarimetry and
interferometry for weather quantification and
forecasting, detection and imaging of military
and civilian targets in the presence of clutter
and noise, measurement of transverse wind, shear
and turbulence, cloud/precipitation microphysics
retrieval and parameterization, resolution
refinement and sensitivity improvement with
oversampling Dr. Tian-You Yu (405-325-3344,
tyu_at_ou.edu) Phased array weather radar, novel
tornado detection algorithms using spectral
processing, technique developments for improving
radar measurements, profiler radars to study
atmospheric dynamics from boundary layer to the
mesosphere Dr. Mark Yeary (405-325-4748,
yeary_at_ou.edu) Radar signal processing, customized
embedded DSP systems, including prototype
development, digital/image signal processing,
adaptive filter design, including neural networks
and pattern recognition, Kalman filters and state
estimation for radar based tracking (point and
distributed targets), real-time systems (hardware
implementation of radar algorithms), radar system
modeling and studies (e.g., all-digital systems
on-a-chip), multi-mode radar operations for
atmospheric, aviation, civilian, homeland, and
military use Dr. Robert Palmer (405-325-6319,
rpalmer_at_ou.edu) Signal processing applied to
weather radar and remote sensing problems, phased
array Doppler radars, adaptive beamforming/imaging
methods applied to the study of the spatial
structure of the atmosphere, clutter mitigation
using array processing, spatial and frequency
diversity methods, resolution enhancement,
refractivity retrieval Dr. Jerry Crain
(405-325-0456, crain_at_ou.edu) Phased array
antennas and radars, radomes, microwave antennas,
and electro-magnetic systems Solid State Radar
architecture Multi-beam, multi-polarization and
distributed aperture techniques Radome design
and interactions Microwave feed networks and
lenses Flat Plate array synthesis and design.
Broadband and multi-frequency arrays.
Mission-based cost vs complexity analysis and
trade studies for civil and military systems
Phenomenology Near-field, far-field and system
simulation and testing. NWRT/TEP
development Dr. Phillip Chilson (405-325-5095,
chilson_at_ou.edu) Atmospheric layered phenomena and
their interactions with the background wind field
using profiling atmospheric radars, ablation of
meteors and the resulting ionization, dynamics in
the summer polar mesosphere region, dynamic
instabilities in the lower atmosphere,
development, implementation, and utilization of
spatial and frequency interferometry, studies of
spatial variability in precipitation using ground
based sensors in conjunction with weather radars,
precipitation studies using combined horizontally
scanning and vertically pointing radars
http//arrl.ou.edu
14
Backup Slides
15
Mission Statement
Operating under the auspices of the University of
Oklahomas strategic radar initiative and in
close collaboration with the National Severe
Storms Laboratory, experts from the Schools of
Meteorology and Electrical/Computer Engineering
have united to form an interdisciplinary team of
scientists and engineers to solve challenging
research problems and prepare the next generation
of students.
16
The University of Oklahoma has a rich heritage of
experimental radar design and field experiments.
The synthesis of a wide array of radar
instrumentation in Norman provides unique
research opportunities.
Microphysics Studies Using Profiling Radars
Snow echoes
Transition region
Rain echoes
Examples of precipitation data collected using
a A UHF boundary layer radar
Radar data used to study precipitation
microphysics
Clear-air echo (vertical air motion)
Rain echo (microphysics)
Boundary layer profiling radars can be used to
study both clear-air signals (which provide wind
data) and precipitation
17

• Polarimetric radar measurements lead
• accurate QPE (rainfall rate) and detailed
  information of microphysics such as
• number concentration
• median volume diameter
• shape of drop size distribution
• evaporation rate and
• accretion rate

Quantitative Precipitation Estimation (QPE)
Electromagnetic wave scattering
from non-spherical hydrometeors is different for
horizontal and vertical polarization, which can
be used to accurately characterize
cloud/precipitation microphysics
KOUN

• In addition to radar reflectivity (Z),
• polarimetric weather radars also
• measure
• differential reflectivity (ZDR)
• cross-correlation coefficient(?hv)
• specific differential phase (KDP)

18
In-Situ Verification of Radar QPE
The PicoNet rain gauge network is being developed
on the Kessler Farm Field Laboratory (KFFL) as
means of validating weather radar measurements.
KFFL is located 30 km South of the OU campus.
A 2D video disdrometer (2DVD) measures the size,
shape, density, orientation, and fall speed of
precipitating particles.
2DVD derived radar variables agree well with
conventional and polarimetric radar measurements.
Locations of the precipitation monitoring nodes
at KFFL
19
Digital Beamforming Studies of Precipitation
Time history of echo power showing mixture of
precipitation and clear-air turbulence
Radar developed by the Univ of Massachusetts.
Array configuration designed to maximize angular
resolution using a sparse array
A 2.5-min sequence of three-dimensional conical
images of isolated precipitation (blue) and
clear-air turbulent (yellow) echoes. The
isosurfaces where obtained at power levels of
52.5 dB (precipitation) and 45.8 dB (clear-air).
20
Applications of Range Imaging for Detailed
Studies of the Boundary Layer
Range Imaging is a multi-frequency technique that
offers a novel means of improving the range
resolution of radar measurements
Comparison of Radar Backscatter from a wind
profiler (upper panels) and an FMCW radar (lower
panel). The same data were used for both the
conventional and range image processing.
Schematic showing the principles of range imaging
21
Example of radar image contaminated by ground and
biological clutter
Advanced Array Processing for Clutter Mitigation
Optimized array design for clutter rejection
Same data as above but using adaptive clutter
rejection scheme
Implementation of subarray design
Experimental results confirm theoretical
performance gains yielding significant clutter
attenuation for both stationary and moving
clutter targets
Radar developed by the Univ of Massachusetts.
22
Mitigation of Non-Stationary Clutter From Wind
Turbine Farms

• Renewable energy production is becoming
  increasing important due in part to economic,
  political, and environmental concerns
• Wind farms cause non-stationary clutter signals
  and wake turbulence-induced radar echoes,
  adversely affecting the operation of military,
  air traffic, and weather surveillance radars
• Conventional ground clutter filters are
  ineffective for mitigating clutter signals from
  rotating blades

Clutter Signal
WSR-88D KDDC
Spectral/wavelet processing may prove important
for analysis and mitigation of wind turbine
clutter
Wind Farm
Future Project Using CASA Radars
CASA network located close to Blue Canyon Wind
Farm 45 turbines, 100 m tall
Current Study Using WSR-88D Radar
Gray County Wind Farm - 170 turbines (near Dodge
City, Kansas)
23
Cross-Disciplinary Curriculum in Weather Radar
and Instrumentation

• GOALS OF CURRICLUM
• Provide a world-class education in radar
  meteorology and instrumentation at both the
  undergraduate and graduate levels
• Combine talents of faculty in the School of
  Meteorology, School of Electrical Computer
  Engineering, and Norman scientists
• Extensive hands-on experience for students

24
Resources on Campus
National Weather Center
New Electrical/Computer Engineering Building
Students OU's College of Engineering ranks in
the top five among colleges of engineering at
public universities in its number of National
Merit Scholars enrolled. Almost 10 percent of
OU's freshman class ranks in the top one-half of
1 percent in the nation in SAT and ACT scores.
Also, the University of Oklahoma ranks in the top
five public universities in the United States in
graduation of Rhodes Scholars since the
scholarship was established.  The graduate
program at OU is strong as well. The School of
Meteorology has over 320 undergraduate and 80
graduate students, making it the largest
meteorology program in the nation. Is ranked 1
in ,the nation in mesoscale and severe storms
research and is among the top 7 programs overall.
OUs president, David Boren envisions Norman as
being home to worlds leading Radar Meteorology
program. Facilities The weather and remote
sensing enterprise in Oklahoma began formally in
1960 with funding by the National Science
Foundation of a proposal to create a meteorology
curriculum in the OU Department of Physics. Four
years later, the National Severe Storms Project
(NSSP), then located in Missouri, established its
Weather Radar Laboratory on OU's north campus, at
which time it was renamed the National Severe
Storms Laboratory (NSSL). This initial
partnership between OU and the Federal
meteorology community foreshadowed the
development of the largest meteorological
education-operations-research enterprise of its
kind in the world and the nation's third largest
concentration of meteorological talent. The
meteorology community in Oklahoma today consists
of nearly 700 professionals across a dozen
organizations. Collectively known as the
Oklahoma Weather Center and soon to be known as
the National Weather Center, this confederation
of Federal and State institutions expends more
than 60M annually and has an economic impact of
approximately 100M. The Oklahoma Weather Center
operates more than 500M in infrastructure
including the nation's premiere state-wide
environmental monitoring sensor network, the
Oklahoma Mesonet.
25
About the Team

• The group has expertise in radar development,
  which includes the physical design, architecture
  of new instruments, signal processing, etc. Our
  strengths and expertise are in radar
  design/architecture, especially in regards to
  the proposed Multi-Function Phased Array Radar
  (MPAR), which has the goal of simultaneous
  aircraft tracking and weather surveillance.
  Several of us have worked with phased array
  processing algorithms related to
  spatial/frequency interferometry, both angular
  and range resolution refinement, adaptive
  beamforming/imaging, beam multiplexing, clutter
  rejection using array processing, and transverse
  winds using spaced antenna. We have recently
  developed a time-series radar simulator based on
  numerical weather models, which can be used in
  the design/testing of conventional, as well as
  phased array radar. General radar signal
  processing is also a major thrust at OU,
  including topics such as detection and imaging of
  military and civilian targets in the presence of
  clutter and noise, spectral processing, adaptive
  filter design, whitening filters, Kalman
  filtering target tracking, real-time systems,
  customized embedded DSP systems, including
  prototype development.
• The group is also keen on radar analysis, which
  encompasses the development and technical support
  of analytical processing of radar data. As
  mentioned, one of our major strengths is in the
  area of weather radar signal processing.
  Currently, we are working on several projects,
  including some related to radar polarimetry and
  interferometry for weather quantification and
  forecasting, novel tornado detection algorithms,
  resolution enhancement, non-stationary clutter
  mitigation, neural networks and pattern
  recognition, and quantitative precipitation
  estimation. Analytical and experimental research
  is also being conducted in modeling of wave
  propagation and scattering in geophysical media,
  development of remote sensing techniques for
  environmental monitoring, and precipitation
  studies using combined horizontally scanning and
  vertically pointing radars.

http//arrl.ou.edu/
the University of Oklahoma
26
Phased Array Radar Research and Development in
Norman
For the first time, moisture measurements have
been made with the dual-mode, military-grade
phased array radar in Norman.
First tornado observations using the PAR in Norman
Preliminary results to demonstrate that beam
multiplexing can improve the data quality.