Wavemill Proof-of-Concept Campaign. Processing Algorithms and Results
Nestor Yague-Martinez, José Márquez. Starlab Barcelona SL, Spain
Martin Cohen, Dave Lancashire. EADS Astrium Ltd, United Kingdom.
Christopher Buck. European Space Agency / ESTEC, Netherlands.
Abstract
Wavemill is an interferometric SAR system concept for an instrument intended for measuring the ocean surface currents and mapping its surface topography. The outcome of the Wavemill feasibility study promoted by ESA was very promising. The purpo of the Wavemill Proof-of-Concept Campaign is to demonstrate the feasibility of the Wavemill concept by flying an interferometric airborne SAR instrument with squinted beams. This paper gives an overview of the needed SAR and InSAR algorithms to process the data and some preliminary results of the campaign.
1 The Wavemill Concept
Wavemill takes advantage of hybrid along- and across-track interferometry to produce a two-dimension
al map of coastal and Open Ocean surface currents. The spaceborne instrument is meant to look
to both the left and right sides of the sub-satellite track to produce two swaths of about 100 km each. The swaths are illuminated by both forward and aft squinted antennas, which allow the currents to be sampled from nearly orthogonal directions [1, 2].
2 Proof-of-Conce
p t
爱情面前谁怕谁Cam
p aign
In October 2011 an airborne InSAR experiment was performed over the Ground truth current monitoring region of Colwyn Bay (Liverpool away). The aim
of this experiment was to demonstrate the Wavemill principle. The data acquisition campaign was carried out by EADS Astrium Ltd, UK with the Astrium Air-borne Radar Demonstrator boarded in a Douglas DC-
3 G-ANAF, owned and operated by RVL Ltd.学茶艺
In the past a similar experiment was conducted [3] which consisted of a hybrid two-antenna InSAR sys-tem in monostatic configuration boarded on an air-plane. The aim of that experiment was the mapping of water surface currents and the generation of a digital elevation model. That system was equipped only with
a pair of antennas, thus the derived surface velocity was only possible in the radar line-of-sight direction. The Wavemill concept overcomes the limitation of measuring in only one dimension by combining two squinted beams, fore and aft [1]. This way a 2D sur-face currents mapping is possible.
In order to achieve both squinted beams, two pairs of antennas, looking fore and aft, have been mounted for the campaign. The acquisitions were done in single-pass mode left looking. Every pair of antennas works together: one of the antennas operates in trans-mit/receive mode (monostatic channel), while the -cond one in receive mode (bistatic channel). Since one of the objectives of the campaign is to proof that a two-dimensional surface velocity can be retrieved, the system was not replicated to the right looking direc-tion, as the Wavemill system forees. Table 1 sum-marizes the system parameters.
Operating frequency9.55 GHz
System Bandwidth100 MHz
Polarization VV
1977年属什么Track altitude2790 m
Platform Velocity80 m/s
Look angle at mid swath 24.9º
Boresight squint angle 21.88º
Table 1: Wavemill PCC System Parameters
The data acquisition campaign took place in two dif-ferent days with two different antennas arrangements:
EUSAR 2012
•Pure along-track configuration: the antennas are parated only in the flight direction.
•Hybrid configuration: the antennas are parated in along-track and across-track direction.
The hybrid configuration provides information about the water surface velocities and the topology. Due to the squinted configuration, and under certain circum-stances, we think that the pure along-track configura-tion might be nsitive to the topology. This is cur-rently under investigation.
The magnitudes of the antennas paration are de-tailed in Table 2.
Across-track antenna paration11 cm
Along-track antenna paration50 cm
Table 2: Antennas configuration
The acquisitions took place over three main sites: Liverpool Bay, Angley and Llangollen, in United Kingdom, the two first sites over water, to retrieve surface current velocities, and the third one over land.
3 Methodology
3.1 SAR
Processing
The SAR data processing tasks have been divided into two steps. EADS Astrium UK provides range-compresd data equidistant in along-track direction. Starlab is in charge of the azimuth focusing. The range compression issues are not addresd in this pa-per.
A direct back-projection algorithm [5] was lected for the azimuth compression. This algorithm performs a time-domain convolution of the range-compresd echoes. The benefits of this algorithm are exact inver-sion and ideal motion compensation capabilities. The drawback is the high computational load.
Bara et al. analyzed in [5] the SAR impul respon for squinted systems pointing out that a pha ramp appears in range direction when processing to zero-Doppler geometry. This can lead to an interferometric pha bias if range misalignments of both images are prent. Fornaro et al. [6] made a generalization of the SAR Processing geometry using a conical reference system and analyd the geometric, spectral and pha aberrations introduced depending on the processing geometry. They stated that processing to the so-called Acquisition Doppler (AD) geometry removes the in-terferometric pha bias pointed out in [5]. This is the approach that we have followed for the
processing of the campaign data. Figure 1 shows the conical refer-ence system, x and r are the target (P) coordinates, wherein the axis reprents the platform path and the conic aperture is given by the angle ϕ.
The angle ϕd is the acquisition squint angle, i.e. the angle defined by the point at the center of the illumi-nation interval and the target, r d is its associated range. As pointed out by Fornaro et al., if we t ϕ=ϕd and r=r d, the processing geometry will cor-respond to the acquisition geometry. Thus the aberra-tions discusd by Bara et al. will then be corrected.
Figure 1: Conical reference system.
For the monostatic channel, the back-projection signal in SAR AD processing geometry for a target defined by r d and ϕd can be written as:
s(t,τ;r d,ϕd)=
t
d
+T obs2
t d−T obs2
s rc(t ,τ;r d,ϕd)e j4πλ[r(t ;r d,ϕd)−r d]dt
(1)
where, t is the azimuth time, τ the slant range time,
s rc(t,τ;r d,φd) the received range-compresd echo at time t, r(t;r d,ϕd) the range from the monostatic channel pha center position at time t to the target (taking into account the platform motion history), t d the azimuth time at which the target is obrved with the nominal antenna boresight and T obs the target ob-rvation time.
泼水节的来历>就的古义
The range-compresd echoes have to be precily interpolated at the corresponding range.
3.2 Interferometric
Processing
Figure 2 prents the block diagram of the interfero-metric processor. The monostatic image is assigned to the master channel and the bistatic image to the slave channel.
A geometrical coregistration is performed using the navigation data. After the master and slave grids have been aligned the interferogram can be formed.
A coherence magnitude map is then generated by us-ing the estimator of two SAR images u1 and u2 [4]:
|ˆγ[i,k]|=
W
u1[i,k]u∗2[i,k]e−jφfep[i,k]
W
|u1[i,k]|2
W
|u∗2[i,k]|2
(2)
Being W a small window centered around pixel [i,k] and e−jφfep[i,k]the pha associated to the flat earth pha (to be considered in the acquisitions with across-track baline).
Figure 2: Wavemill-PCC Interferometric Processor For the further processing, depending on the acquisi-tion antennas configuration along-track or hybrid in-terferometry (i.e across and along-track interferome-try) can be performed.
For a pure along-track system, the interferometric along-track pha can be written as:
Δφ=2π
λ
ΔR=
2π
λ
dR
dt
Δt=
2π
λ
B x
U r
V
(3)
where U r is the surface velocity in the line-of-sight direction, V the platform velocity, B x the along-track baline and λ the radar wavelength.
The cross-track pha is given by:
ΔφXT I=−2π
λ
B2y+B2z sin(θ−α)(4)
where B y and B z are the across-track baline com-ponents in the horizontal and vertical plane respec-tively, θ is the look angle and α the angle the baline makes with respect to a reference horizontal plane. The hybrid interferometric pha can then be written as:
Δφ
hybrid=2
πλB x
U r
V七零八碎的意思
−
2π
λ
B2y+B2z sin(θ−α)(5)
4 Preliminary
results
Some preliminary results are prented below. Figure 3 shows amplitude, coherence and interferometric pha of a crop of a long data take over the Menait Strait, Angley, United Kingdom. The acquisition was performed with a hybrid baline configuration and has an extension of ca. 1000
m (azimuth) x 600 m (range). The low coherence over the water can be due to low backscattering of the water surface due to wind lack. There are also available acquisitions over Open Ocean, which we expect will have a better coherence. We are currently processing the images.
Figure 3: Menait Strait. Amplitude (top), coherence (middle), interferometric pha (bottom) Figure 4: Llangollen. Amplitude (top), coherence (middle), interferometric pha (bottom)
Figure 4 shows amplitude, coherence and interfero-metric pha of a crop of a data take over Llangollen, a hilly area in United Kingdom. The interferometric pha has been flattened. The height of the amiguity is ca. 400 m for mid swath. Note the prence of a hill with a height difference of about 200 m from early azimuth towards late azimuth.
5 Conclusions
The prented preliminary results have shown promis-ing capabilities of the Wavemill PCC for topography mapping, demonstrated for the acquisition over Llan-gollen area. The acquisition over Menait Strait did not show enough coherence over water and we suspect, that this is due to the lack of signal backscattering to-wards the nsor. We are working in the processing of acquisitions over Open Ocean, where we expect to have a better coherence.
随意和随便的区别More results will be prented at the workshop.
6 References
[1] C. Buck C, M. Caparrini, J. Márquez, B. Richards,
D. Lancashire. The Wavemill Concept for Direct M easurement of 2D Ocean Surface Currents. ESA Envisat Symposium, Montreux (Switzerland), 23 – 27 April 2007.
[2] C. Buck, M. Aguirre, C. Donlon, D. Petrolati, S D'Addio. Steps towards the preparation of a Wavemill Mission.
[3] R. Siegmund, M. Bao, S. Lehner, R. M ayerle. First demonstration of Surface Currents imaged by hybrid along- and cross-track interferometric SAR. IEEE Transactions on Geoscience and Remote Sens-ing, Vol 42, No 3, March 2004.
[4] R. Bamler and P. Hartl. Synthetic aperture radar interferometry. Inver Problems 14 (1998) R1–R54. [5] A. Yegulalp. Fast Backprojection Algorithm for Syntetic Aperture Radar. IEEE Radar Conference 1999.
祖国啊我亲爱的祖国教案
[Bara] M. Bara, R. Scheiber, A. Broquetas, A. Moreira. Interferometric SAR Signal Analysis in the Prence of squint. IEEE Transactions on Geoscience and Remote Sensing, Vol 38. No 5. September 2000.
[6] G. Fornaro, E. Sansosti, R. Lanari, M. Tesauro. Role of the P rocessing Geometry in SAR Raw Data Focusing. IEEE Transactions on Aerospace and Elec-tronic Systems Vol 38, No. 2. April 2002.
