Sheet 1 of 18
Transcription
Sheet 1 of 18
Sheet 1 of 18 SAR Synthetic Aperture Radar Why use microwave frequencies? - atmospheric attenuation is low at microwave frequencies,particularly 1-10GHz,apart from resonances due to oxygen and water molecules. - microwave signals will penetrate cloud and fog,in contrast to visible radiation. - microwaves can continuously monitor ploar regions and tropical forests that are often in darkness or covered in clouds. - SAR is to be used to monitor natural disasters -floods,forest fires,earthquakes etc that are often accompanied by poor optical visibility. - fairly narrow beams can be obtained with resonably sized antennas to give acceptable resolution. Features of SAR - a large effective antenna aperture is obtained by mounting the antenna on a moving platform-an aircraft or a satellite. - resolutions of down to 10m can be obtained,so that images of the earth’s surface can be generated. - complex signal processing is required to extract images,so that real-time operation makes heavy demands on computer processing power. - very large amounts of data are generated - as can be seen if say 10 items of information are generated for each 10m2 of the earth’s surface. [Exercise :Calculate the number if items of data,given that the earth’s radius is 6000km]. - particularly using satellites,large areas of the earth’s surface can be observed in a short time so that changes can be observed. Conventional radar (fixed antenna) - resolution - the radar detects objects that lie in the antenna beam by the echoes they produce. - the range to the target is found from the time delay between the transmitted and returned signal R = 1 c. TR 2 - the range resolution is determined by the pulse duration τ .A 1uS pulse gives a range resolution of 150m. - Pulse Compression is used to improve the range resolution ,without the requirement for a very short duration,high power pulses.a short pulse is generated to give a good range resolution,but this is frequency modulated to give a longer,lower power pulse prior to amplification and transmission.This is achieved using a dispersive filter.The received pulse is compressed by passing it through a complementary dispersive filter so that it is effectively shortened ,thereby achieving good range resolution.Range resolutions of about 10m can be achieved,corresponding to an effective pulse duration of 70nS, 1 Sheet 2 of 18 - the angular resolution (or azimuthal resolution) of a conventional radar is determined by the angular width of the main lobe of the antenna radiation pattern.This ,in turn,is determined by the λ/D ratio for the antenna aperture. - the linear resolution in the cross-range direction depends on the angular width of the antenna beam and the range to the target area. - for a typical satellite-borne SAR the slant range is about 900km. - for SEASAT SAR frequency = 1.3GHz hence λ = 23cm. The antenna length is 10.7m , giving a beamwidth of 23/1070 = 0.022 radians. Thus, the ground resolution = 0.022 x 900km = 19km The conclusion is that the linear resolution in the range direction (about 10m using the pulse compression) is acceptable,but the linear resolution in the cross-range (azimuthal) direction is inadequate (19km). The cross range resolution can be dramatically improved to about 10 using SAR which comprises - antenna mounted on an aircraft or satellite moving parallel to the earth’s surface - sideways looking radar - regular pulses emitted,data collected for each pulse - data from all echoes from a given target area are combined to form an image of the area. Platform movement velocity v Antenna La Lc Angular beamwidth in acrosstrack(range) direction ‘Look’ angle (angle of incidence) ψ Ground track vertically below satellite, along-track direction (azimuth) Slant range to target Satellite altitude swathe SAR Geometry Radar footprint determined by antenna dimensions 2 Across track direction (range) Sheet 3 of 18 Key parameters for Free-flying SAR satellite systems Satellite Seasat ALMAZ Agency/country NASA/USA USSR Launch date 1978 1991 Altitude(km) 800 280 Frequency Band (GHz) L(1.3) S(3.0) Polarisation HH HH Incidence angle 23 30-60 (degrees) Antenna Size (m x m) 10.7 x 2.2 15 x 1.5 (two) Noise Equiv (dB) -18 Swathe width (km) 100 25-50 Az resolution (m)/Looks 23/4 15/2 Range Bandwidth 19 Uncoded (MHz) Quantisation (bps) Analog 3 3 E-ERS-1 ESA 1991 785 C(5.3) VV 23 J-ERS-1 NASDA/Japan 1992 565 L(1.2) HH 35 Radarsat Canada 1995 792 C(5.3) HH 20-50 10 x 1.0 -18 100 25/3 13.5 12 x 2.2 -20 75 30/4 15 15 x 1.6 -21 50-500 28/4 11.5,17.3,30 5 3 4 Sheet 4 of 18 SEASAT (1987) altitude ‘look’ (incidence) angle Antenna size - along track - across track frequency swathe width 800km 23 degrees 10.7m 2.2m 1.3GHz(λ = 23cm) 100km 10.7m 2.2m Satellite movement θ2 θ1 Along track 870km 23° 800km Across track Across-track resolution Ground resolution (conventional radar) Swathe width from this data: slant range beamwidth along track beamwidth across track 870km θ1 = 23/1070 = 0.022 radians θ2 = 23/220 = 0.1 radians from simple geometry swathe width = 870km x 0.1/cos23 = 95km along track resolution on ground = along-track width of beam x slant range = 0.022 x 870km = 19km This is the resolution of a conventional radar which can only determine whether or not a target lies within the beam. 4 Sheet 5 of 18 SAR Principle - the across-track position of a target is determined by a conventional delay-time measurement. - the along-track position is determined by the Doppler shift of the echo signal. Pulse emitted Along track speed v θ1 φ Vr = radial velocity of radar w.r.t to target = v.sinφ Slant range R targets A B x Along track ∆x 2v.sinφ ⎞ Change in radar frequency due to Doppler effect ∆f = ⎛⎜ ⎟ fo = f D ⎝ c ⎠ - the angular positions - or along-range positions - of two targets within the beam can be determined by the Doppler frequencies of their echo signals. - the angular resolution or along-track resolution of the target is determined by the frequency resolution that can be obtained for the Doppler frequencies. 2v.sinϕ ⎞ f D = ⎛⎜ ⎟ fo ⎝ c ⎠ Now ϑ 1 is small (0.02 radians) & φ ≤ ∴ sinφ ≈ φ ≈ x R 1 ϑ1 2 2v x and f D = ⎛⎜ ⎞⎟ ⎛⎜ ⎞⎟ fo ⎝ c ⎠⎝ R⎠ 2v fo ∴ ∆f D = ⎛⎜ ⎞⎟ ⎛⎜ ⎞⎟ ∆x ... difference in Doppler frequency for two targets with separation ∆x ⎝ c ⎠⎝ R ⎠ 5 Sheet 6 of 18 Resolution The ground resolution in the along-track direction ∆x (min) corresponds to the minimum change in Doppler frequency that can be measured , ∆fD (min). c ⎛ R⎞ ∆x (min) = ⎛⎜ ⎞⎟ ⎜ ⎟ ∆f D ⎝ 2v ⎠ ⎝ fo ⎠ ↓ resolution The minimum change ig f D that can be measured = Tobs = 1 Tobs Rϑ 1 R λ antenna beamwidth measured on ground = . = v La satellite speed v ∴ ∆f D (min) = L a = length of antenna 1 v La = . Tobs R λ c ⎛ R⎞ v ⎛L ⎞ ∴ ∆x(min) = ⎛⎜ ⎞⎟ ⎜ ⎟ ⎛⎜ ⎞⎟ ⎜ a ⎟ = ⎝ 2v ⎠ ⎝ fo ⎠ ⎝ R ⎠ ⎝ λ ⎠ θ1 1 La ie from Seasat ∆x ≈ 5.4m 2 Tobs = (R. θ1)/v R. θ1 Speed v 6 Sheet 7 of 18 Limits on SAR PRF (Pulse Repetition Frequency) Limits on the PRF (equal to the reciprocal of the time interval between successive pulses) arise because of the requirements of avoiding cross-track range ambiguity and along track phase ambiguity. S a t e llit e t r a v e ls p e r p e n d ic u la r t o t h is p la n e A & B a r e p o in t t a r g e t s a t the sw athe ed g es C ro ss-T r a c k r a n g e a m b ig u ity R2 R1 C r o s s - t r a c k d ir e c t io n A B S w a t h e w id t h 1 A 0 No pulses here 1 3 2 τ (PRF) A B 2R2 c 2 R1 c time 3 2 τ‘ (PRF) Higher PRF A A 0 B time 2R2 c To avoid cross-track range ambiguity,signal from B must arrive earlier than that for A.ie 7 B Sheet 8 of 18 2 R2 ⎛ 2R ⎞ < ⎜ 1 + τ ( PRF )⎟ ⎝ ⎠ c c ( R2 − R1 ) < 1 c c. τ ( PRF ) or ( R2 − R1 ) < 2 2( PRF) Limits on PRF - along track phase ambiguity v Satellite path θ/2 θ/2 Along track direction fD(min) fD(max) fo As beam moves over target Doppler frequency changes from fD(max) to fD(min). f D(max) = f D(min) = ϑ = λ La 2v ϑ . . fo c 2 (ϑ is small) L a = antenna length (along track) Doppler frequency bandwidth = 2v λ v . . fo = c 2 La La Echo signal is sampled during pulses - sample rate = PRF Sampling theorum: Sampling rate ≥ 2 f D(max) = 2v ∴ PRF ≥ La ie minimum PRF ≡ 1 pulse each time antenna moves through 8 1 its length L a 2 Sheet 9 of 18 Limits on PRF Along track resolution ∆x = ∴ La 2 PRF * ∆x ≥ 2v ie PRF determines along - track resolution ∴Two conditions on the PRF are : PRF ≤ c to avoid range ambiguity 2(R2 - R1) PRF ≥ 2v La sampling theorem - to avoid phase ambiguity From these results a suitable compromise must be reached between - along track resolution - swathe width - antenna area 9 Sheet 10 of 18 Focussed SAR To attain highest along-track resolution of La/2 the observation time Tobs = time taken for beam to sweep across the target.But in this time the Doppler frequency changes from +fD(max) to - fD(max). The solution is to process the SAR data so that it focusses on each along-track target position in turn. Pulse emission positions x xo Satellite path ∆R Closest-approach Ro range Ro Slant-plane Geometry Target Constant range line within swathe Two-way phase delay to antenna position x,relative to xo 2π ∆φ = - 2⎛⎜ ⎞⎟ ∆R ⎝ λ ⎠ Ro + ∆R = ∴ ∆R ≈ [ R 2o + (x − xo ) (x − xo )2 2 Ro 1 2 2 ] since x - x o <<R To focus on target at xo apply phase corrections ∆φ to echos collected at each antenna position & then add coherently all returns to find the energy reflected from xo. 10 Sheet 11 of 18 The change in phase with satellite position x for a target at closest approach xo is given by ⎧⎪ 4π x − x 2 ⎫⎪ o f ( x ) = exp ⎨− j. . Differentiate phase term ⎬ ( ⎪⎩ λ ) 2 Ro ⎪⎭ Phase term ∴ Instantaneous Doppler frequency fD as a function of the position of the satellite = 2(x - x o ) 1 ∂ . .(phase) = λRo 2π ∂x f = rate ofchange of phase θ = 2πft + φ ; ∂θ = 2πf ∂t ∴f = 1 ∂θ . 2π ∂t To determine the scattering from the target position associated with xo we find the correlation between this reference function for fD;over the range of positions for which data is collected for xo,with the actual returns to the satellite at these satellite positions. This must be done for all ranges Ro across the swathe and for each resolution cell along the swathe. 11 Sheet 12 of 18 Applications of SAR Studies of - Ploar ice. - Ocean waves. - Surface/subsurface mapping. - soil moisture. - Forest ecology. - Ship’s wakes Each type of study dictates different aspects of the (1) Experimental requirements width, eg frequency,polarisation ,angle of incidence,swathe resolution. (2) System performance resolution eg EIRP,dynamic range,data rate,quantisation resolution. (3) Platform design eg altitude,launch date/time,mission duration,orbit Surface interactions of electromagnetic waves Characteristics of reflected electromagnetic waves Surface parameters - ampltitude relative permittivity - phase depends mainly on roughness polarisation local slope Surface scattering at interface between two different media Volume scattering by particle distribution in a non-homogenoeous medium. Surface/subsurface scattering If surface roughness << λ (radar wavelength) Example Libyan desert - specular reflection i = r normal refraction - no vegetation,very smooth surface 1 -2 m of sand on bedrock. Main dielectric interface is sand - bedrock optical image shows sand surface. Microwave image shows detailed map of ancient natural drainage channels in the rock,beneath the sand.From these the geological history of the region can be understood. Images may help to locate deep water sources in the region. 12 Sheet 13 of 18 Bragg scattering - slightly rough surfaces rms height variation < λ/8 - main application is in the study of ocean waves Bragg Equation = nλ = 2dsinθ nλ/2 θ θ Ln - for large angle of incidence the total scattering is a combination of Bragg scattering and specular scattering. - natural surface are represented as a series of facets upon which the small-scale surface roughness is superimposed. σo Facet scattering Bragg scattering η Backscatter curve for natural surfaces illustrating the two scattering mechanisms:Facet scattering for steep incidence angles;Bragg scattering for shallow incidence angle. 13 Sheet 14 of 18 Ocean waves - Distinguish long wavelength ocean waves - wavelength d is 10’s to 100’s of metres - due to local wind conditions or distant storms - wave speed is approximately equal to wind speed - infer wind speed - monitor using ground HF radar frequency 10 to 30MHz; wavelength 30m to 10m,an ‘over-the-horizon’ radar that relies on reflection of signals from the ionosphere. - short wavelength waves - capillary waves - wavelengths are ~ cm to 10’s of cm - due to local wind conditions - use SARs operating in the 1 to 10GHz frequency range. - SAR gives wave direction ,wavelength and wave height information. - information is used to develop ,ocean wave forcast models which are important for weather forecasting and the prediction of changes in the global climate 14 Sheet 15 of 18 Geology - Bragg scattering is used to intepret the scattering from sparsely vegetated rock terrains. - the rock type and age can be inferred from the surface roughness. Eg sedimentary rocks will be more weathered-and hence rough than igneous rocks older rocks will be more weathered than rocks sedimentary and igneous rocks can be distinguished to some extent by the brightness of the scattering they give. Study of Forest Canopies - Volume scattering - due to ‘particle’ distribution in a non-homogenoeous medium. - the penetration of radiation into a medium - and hence the extent to which volume scattering is important - depends on the skin depth δ of the medium. δ = 1 σ is the conductivity ofthe medium; f is the microwave frequency π . µ. σ . f - the radar frequency is important because it determines. - the penetration of the signal into the region,and hence the volume sampled - the resolution of the objects within the region ,and - the ability of the radar to distinguish different materials within the region because their scattering cross-section depends upon the frequency. - volume scattering is used to study forest canopies - by comparing the returns at different frequencies and with different polarizations (vertical & horizontal) the distribution of vegetation at various heights within the canopy can be inferred. - the polarisation changes distinguish bewteen returns which arise from single scattering and multiple-scattering. - if changes in canopy density over a period are monitored the effects of environmental changes such as the increase in acid rain can be studied. 15 Sheet 16 of 18 Polar Ice - different types - and hence ages - polar ice can be distinguished by the scattering they give. - as with vegetation scattering ,more information about the distribution and development of ice floes is obtained by comparing the images formed at different microwave frequencies. - studies of the polar regions are important in investigations of global warming (due to,for example carbon dioxide emissions) because the effects could be most severe there. - the growth of sea-ice is a primary mechanism for the removal of carbon dioxide from the atmosphere. - knowledge of the position of ice is important for fishing and navigation in the polar regions. MULTIYEAR LOWSALINITY,LOWLOSS SURFACE-VOLUME SCATTERING FIRST YEAR: HIGH-SALINITY HIGH-LOSS SURFACE SCATTERING OPEN WATER: HIGH-LOSS SURFACE SCATTERING SEA LEVEL RIDGE SURFACE SCATTERING GLACIER ICE LAKE ICE VOLUME SCATTERING Scattering mechanisms for various ice types :multiyear,first year and open water 16 WATER Sheet 17 of 18 - more information about the distribution and development of ice floes is obtained by comparing the images formed at different microwave frequencies 102 T =-10°C 10 Pure Ice ε = 3 - j0.03 1 Multi-year ice Pentration depth (m) First year ice 10-1 ε = 3.3 - j0.25 10-2 1.5 2 3 4 5 6 7 8 9 10 15 20 Frequency (GHz) - studies of the polar regions are important in investigations of global warming (due to,for example,carbon dioxide emissions) because the effects could be most severe there - the growth of sea-ice is a primary mechanism for the removal of carbon-dioxide from the atmosphere . - knowledge of the position of ice is important for fishing and navigation in the polar regions. 17 Sheet 18 of 18 Soil Moisture - the moisture content of soil varies the relative permittivity of the soil and hence it’s microwave scattering backscatter cross-section. 24 22 20 18 ε‘ 16 14 D ie le c t ric C o n st a n t S O IL M O IS T U R E a t L -B A N D 12 10 8 6 4 ε ‘’ 2 0 20 10 30 S o il M o istu re ( % ) C -B and X -B and L -B and 90 RADAR W AVELENGTH 80 ε‘ 70 L iq u id W ater 60 50 R eal p art, ε‘ 40 D ie le c t r ic C o nstant 30 I m a g in a r y P a r t , ε ‘’ 20 ε ‘’ 10 0 100 30 3 10 1 0 .3 0 .1 W a v e le n g t h ( c m ) Dependance of complex dielectric constant on (a) Soil moisture at L-Band; and (b) Radar wavelength. (Ulaby et al,1982) - knowledge of the water content of soil is important for the development of models for the global hydrological cycle. 18