1. CorsodiDottorato
in IngegneriadeiSistemiElettronici
XXIII ciclo
Advancements on interferometric radar:
studies and applications
Coordinatore: Prof. Carlo Atzeni
Tutor: Prof. Massimiliano Pieraccini
Candidato: Andrea Barucci
4. CWSF Ground BasedSynthetic Aperture Interferometric radar
CentralFrequency 16.75 GHz
Banwidth 400 MHz
Tx Power 26 dBm EIRP
SAR scan length 2m
SAR acquisition time 4 min (minimum)
Power consumption < 40 W
6. Measurementcampaign
Start 16 feb 2009
End 25 mar 2009
Days of measurement 39
N. Radar acquisitions about 600 (1 every 2 hours)
7. RCS SAR image
Range 0.5 m
resolution
Cross 5mrad
range
resolution
8. Interferograms in scarp area
(m)
(m)
(m)
(m) (m) (m)
(m)
(m)
(m) (m) (m) (m)
Eachinterferogramrelatesto a timelapseoftwohours
9. Interferometricanalysisin the wallarea
1. PSsselection
DispersionIndex
About 50.000 PSsselected in
all the area
2.
Atmosphericmeanphasescreen
removal
3. Calculationofdisplacements
14. Interferometric DEM of the quarry
Laser
Conclusions
• A ground based Synthetic Aperture Interferometer was
successfully tested for the first time in a quarry.
• The system showed an accuracy of few tenths of a millimeter
in displacement’s measurement, over a long period and at
800 meters of distance.
• In short term measurements a simulation showed an
accuracy in measurement of displacements of the order of a
tenth of a millimeter.
• DEM reconstruction showed some limitations especially in
high slopes.
Radar
15. Universal fluctuations in tropospheric radar measurements
Introduction
• Long term atmospheric artifacts on GB-SAR are known
• Radar signal propagation through tropospheric turbulence is a new
research topic
Experiment
Data
Analysis
S. T. Bramwell, P. C.W. Holdsworth, and J.-F. Pinton,
Classic: Modern:
Universality of rare fluctuations in turbulence and critical
phenomena, Nature 396, 1998
Fluctuations of signal
Correlated systems, Physical Review Letters [2000, 2005, 2008]
as a function of
Gumbel statistics
turbulence
17. Scenario and raw-data
CR Ultrasonic Anemometer / Radar sampling freq Range Observational time
distance Weather sensor Resolution
65 m 4/1Hz About 50 Hz 1m 6h
18. Detrending – High-pass filter
• Turbulent phenomena occur
at the micro scale,
corresponding to distances
shorter than 1 km, which are
the usual working distances for
a GB-SAR.
• We use a moving average to
remove the fluctuations on
temporal scales greater than 30
minutes
19. Time series data analysis
We use overlapping
temporal windows, with
length of about 30
minutes:
- About 10^5 points ->
rare fluctuations & good
statistics
- Constant Reynolds
number
Turbulent Kinetic
Energy (TKE) is a
measure of the
turbulence
21. Radar signal statistics & Correlated Systems
Radar signal through a Turbulence
turbulent atmosphere Correlated
Ising& 2D XY
Systems
River water
level
Forest fires,
etc.
Generalized
Gumbel PDF
Global quantity
Radar Data
statistics?
22. Generalized Gumbel
• Generalized Gumbel describes the fluctuations in Correlated Systems, its
shape depends on a real parameter a
• As a varies, the shape of the distribution varies from a completely
asymmetric, negatively skewed, distribution to a symmetric one quite similar
to a Gaussian function
• The parameter a has been proposed to be inversely related to the correlation
length of the system
23. Data distribution
• Fit of the data with the GG(a)
• Amplitude statistic appears to fit with the GG
• Phase statistic doesn’t show the same behaviour
Amplitude
• The radar data distributions calculated over the
analysis windows show different means and
standard deviations new method from Nature
Phase
24. • a is determined by the turbulence strength
• In hard windy conditions, a is nearly equal to pi/2 which is the value
for which GGa is approximately the Bramwell-Holdsworth-Pinton
distribution observed in turbulence and critical systems
Andrea Barucci
25. Conclusions
• The radar signal statistic:
– It is influenced by the turbulence
– It is a global quantity able to measure the “correlation of the
atmosphere” at the micro-scale
• We confirm the adequacy of the Gumbel statistic to describe
highly correlated complex systems
• SAR images - turbulence: work in progress
• Paper published by Europhysics Letters: A. Barucci, G. Macaluso, D. Mecatti, L. Noferini, D. Fanelli, A.
Facchini, M. Materassi, M. Pieraccini and C. Atzeni, Universal fluctuations in tropospheric radar
measurements, EPL, 89 (2010) 20006, DOI 10.1209/0295-5075/89/20006
Hinweis der Redaktion
Effettoatmosfericosuimmagini SAR noto, peròèpossibilemodellizzare solo la parte media.Rimozione trend tramiteconoscenzameteoe fit con i LSNelleimmaginicisonoevidenzedieffettidovutialledisomogeneitàspazialidell’atmo. Questieffettiadesso non sonomodellizzati.Esempio: Formigal in casodi molto vento.L’indicedirifrazionerisentedei trend lunghiedellaturbolenza.Parlando con DuccioFanelli, abbiamoscoperto un modo in cui sipuòmettere in evidenzacertifenomeni -> distribuzioneuniversale
In order to investigate more deeply this connection, an outdoor experiment was arranged in a controlled environment. - A Ku band radar system was mounted on a plateau of an disused quarry looking at an artificial target. - The temporal behaviour of the radar signal was analysed in comparison with the wind measured by an anemometer installed nearby.
Amplitude and Phase time series: 1 million points long
Eseguire un detrending con unafinestra a media mobile equivale a fare un filtropassa-alto. Ampiezzaefasedetrended con finestra 30 minuti.
Analisiclassica: stdvstke. Prendofinestrasegnalelunga per avere max numerodipunti, ma devotenereilnumerodi Reynolds costante. Le finestresisovrappongono. Devo fare confronto con energia del vento, chestimo come TKEAnalisistatistica: pdf, GG
AmpiezzaeFaseanalizzatesufinestretemporali diverse. Ampiezzafino ad 1h, mentre la fasedeverimaneresull’ordinedeiminuti.GeneralmenteAmpiezza 30-60 minuti, Fase 3-10 minuti
The atmosphere crossed by an e.m. wave can be ideally segmented in neighboring volume cellsThe signal impinging on the junction between two adjacent cells is partly reflected and partly transmittedThe data collected at the receiver is sensitive to the sequence of scattering events occurred along the propagation pathIn this sense the radar echo is a global quantity indirectly representing the specific state of the crossed medium
The Gamma function is an extension of the factorial function shifted down by 1, to real and complex numbers.
By varying a, the GG distribution can fit with the radar data distributions found over different temporal windows La GG descrivebene solo l’Ampiezza, ilvalore del parametro “a” per la faseoscillacontinuamentedavaloribassi a valorialtissimi, non èquindipossibiletrarreconclusioni. Forse la Faseèpiùsensibiledell’Ampiezza???
Immaginequalitativa del comportamentodi “a” con ilvento
Immaginequalitativa del comportamentodi “a” con ilvento- Il valoreditKEditransizione, circa 6 (m/s)^2 non varia al variaredellalarghezzadellafinestrausata per l’analisi. Potrebbeperòvariare con la distanza del bersagliodal radar. Ulterioriesperimentisarebberonecessari per chiarire tale aspetto.
By varying a, the GG distribution can fit with the radar data distributions found over different temporal windows La GG descrivebene solo l’Ampiezza, ilvalore del parametro “a” per la faseoscillacontinuamentedavaloribassi a valorialtissimi, non èquindipossibiletrarreconclusioni. Forse la Faseèpiùsensibiledell’Ampiezza???
By varying a, the GG distribution can fit with the radar data distributions found over different temporal windows La GG descrivebene solo l’Ampiezza, ilvalore del parametro “a” per la faseoscillacontinuamentedavaloribassi a valorialtissimi, non èquindipossibiletrarreconclusioni. Forse la Faseèpiùsensibiledell’Ampiezza???