Wind Profiler Signal & Data

pancakesbootAI and Robotics

Nov 24, 2013 (3 years and 9 months ago)

149 views

Wind Profiler Signal & Data
Processing

-
Anil Anant Kulkarni

SAMEER, IIT Campus,Powai Mumbai 400076

anilakulkarni@hotmail.com


Wind Profiler Signal & Data Processing


Background


Signal Processing Steps


Data Analysis Step


Data QA/QC


Wind Profiler : Basics…..


Clear Air Doppler Radar


Detects Reflection from Turbulence and
eddies


Typical frequencies used in wind profiling


45
-
65 MHz



404
-
482 MHz



915
-
924 MHz



1280
-
1357.5 MHz


Wind Profiler Basics ….


Electromagnetic pulse is sent into the Atmosphere


Detection of the signal backscattered from
refractive index in
-
homogeneities in the
atmosphere


In clear air the scattering targets are the
temperature and humidity fluctuations produced
by turbulent eddies



Scale is about half of the wavelength for the
transmitted radiation (the Bragg Condition)


Wind Profiler : Back Scatter Signals

Wind Profiler : Scattering Mechanism


Scattering from atmospheric targets:


irregularities in the refractive index of the air


hydrometeors, particularly wet ones (rain, melting snow,
water coated ice)


Scattering from Non
-
atmospheric targets:


birds and insects (frequency dependant)



smoke plumes


Interfering signals:



Ground and sea clutter



Aircraft and migrating birds



RFI (depends on frequency band)


Wind Profiler : Scattering Mechanism

When a pulse encounters a target...

It is scattered in all directions.


Of interest is the signal component

received back at the radar.

This signal is typically much weaker

than the original sent from the

transmitter and is called the "return

signal".


The larger the target, the stronger

the scattered signal.


Wind Profiler : Scattering Mechanism


Refractive index fluctuations are carried out
by the wind; are used as tracers


Irregularities exist in a size range of a few
centimeters to many meters


Different methods of wind measurement used
with numerous variations:



SA (Spaced Antenna)



DBS (Doppler Beam Swinging)


Doppler shift in the backscattered signal is
used to derive the wind speed and direction
as function of height


Doppler Beam Swinging (DBS)



DBS method for wind vector
calculations (u,v,w)


Radial velocities measured with
one vertical and 2 off
-
zenith
beams


Beam
-
pointing sequence is
repeated every 1
-
5 minutes


Electronic beam pointing with
phase shifters using one
antenna


Local horizontal uniformity


of the wind field is assumed

Doppler Shift


Doppler Formula:


f
d

=
-

2 *V
r

/



Doppler

Measurement of wind speed
based on the Doppler shift in the received
signal:



where
V
r

is the radial velocity of the scatterers



Examples of Wind Profiler Doppler shift
(radial velocity 10m/s)


50MHz, wavelength 6m, Doppler shift 3.34Hz


449MHz, wavelength 0.66815m, Doppler shift
29.9Hz

– 1290MHz, wavelength 0.23m, Doppler shift
86Hz

Time
Domain
Processing
(1.0)

Spectral
Domain
Processing
(2.0)

Doppler
Profile
Analysis (3.0)

Wind
Profiles

Rx I/Ps

WP Signal Processing Steps

DSP System : Data Flow Diagram

Power Spectra
+ Moments

Power Spectra
+ Moments

Power Spectra

Radar Control PC

Post Processor PC

Front End PCI
DSP Card(1)

PCI DSP
Card(2)

I & Q I/P


Time Domain Signal Processing…….


ADC Sampling


Coherent Integration


Affects data rate, Nyquist frequency, SNR


8 bit Decoding


Improving the Range Resolution


Fourier Transform



Broadens spectral features


Power Spectral Computation.



Moments of the Average Doppler




Spectral Averaging


Reduces data rate,improves detectability


Estimation of Noise Level


Identification of Doppler Signals


Maximum Peak


Construction of Doppler Profile


Computation of Moments and SNR


Spectral Domain Processing……

Basic
Signal
Processing
Steps

Doppler Profile Analysis:



The

Doppler

profiles

from

three

beam

directions

from

lower

heights

and

higher

heights

are

available

as

inputs



To

analyse

input

data

to

generate

the

6

minute

and

hourly

wind

profiles
.



In

this

process

the

input

Doppler

profiles

are

subjected

extensive

quality

assurance

checks

before

generating

the

6

minute

and

hourly

wind

profiles
.






Separation

of

Precipitation

echoes





Mode

Merging





Calculation

of

Radial

velocity

and

height

(
6

min)





Computation

of

Absolute

Wind

Velocity

Vectors

(UVW)





Quality

Assurance

of

sub
-
hourly

velocity

profiles





Computation

of

Horizontal

Wind

Speed

&

direction

(
6

min)





Computation

of

Hourly

Averages




Basic Issues in Signal Processing….

Signal Detection



Discrimination between signal and noise. (Hildebrand/Sekhon)


Are one or more non
-
noise signals present in spectrum?


Signal Identification Signal Identification


If more than one signal is detected, which one is due to the (clear
(clear
-
air) atmospheric return? air) atmospheric return?


What kind of What kind of a
-
priori information priori information can be
used to select it?


Can unwanted contamination be effectively filtered out without affecting
(biasing) the desired

Identification of Doppler Peaks…



Basic Assumptions….



There exist temporal and spatial continuities in a
time series of spectral profiles which can which
can be be employed.



Echoes back
-
scattered from the atmosphere
exhibit continuity in time and height that can
restrict the search of restrict the search of signal
peaks to a certain part of the spectrum
.



Identification of Doppler Peaks…



Multiple Peak Identifications….


Identify maximum 5 Spectral Peaks in each
range bin


Mark spectral peaks which are below the
noise level threshold


Compute three Moments for remaining
spectral peaks.


Build the spectral chain across different
range bins using wind shear criteria


Doppler Peak Identification
continued..


Challenges …


Identification of Atmospheric Targets but
not the Clear Air echoes


Precipitation echoes


Identification Interference Signal


Identification of Clutter


Identification of Non
-
Atmospheric Targets


Birds, Planes, non
-
stationary objects from near by
buildings , roads (from Radar Side lobes)

Interferences….


Interference from migrating birds:


Birds act as large radar targets so that signals from birds overwhelm the
weaker atmospheric signals
This can produce biases in the wind speed
and direction


Precipitation interference:


During precipitation, the profiler measures the fall speed of rain
drops


Ground clutter:


Ground clutter occurs when a transmitted signal is reflected off of
objects such as trees, power lines, or buildings instead of the
atmosphere.
Data contaminated by ground clutter can be
detected as a wind shift or a decrease in wind speed at affected
altitudes.


RF Interference:


The RF Interference signals looks similar to the CAT echoes and some
times are inseparable

Power Spectra : Vertical Beam with Precipitation echoes

Power Spectra : North Beam with Precipitation echoes

During precipitation, the
profiler measures the
fall speed of rain drops

Power Spectra : East Beam with Precipitation echoes

Power Spectra Higher Heights

Power Spectra: Lower Heights

QA/ QC of Data


Definition:


The process of identifying and if possible
eliminating inconsistent observations
(outliers)


Outliers:


Data that are spatially, temporally, or
physically inconsistent.

Recent development in QA/QC


Coherent Integration


Wavelet pre
-
processing / No coherent integration / Low
-
pass filter


Windowed FFT :


No windowing for long time series.


Spectral Averaging


Statistical Averaging Method (SAM
-
ICRA)


Signal Identification


Multi
-
Peak Picking (MPP) / ETL Signal Processing System (SPS)
/NCAR Improved Moments Algorithm (NIMA)


Wind finding


NCAR Winds and Confidence Algorithm (NWCA)


ETL Signal Processing System (SPS)


Weber/Wuertz (QC)