Richards J.A., Jia X. Remote Sensing Digital Image Analysis: An Introduction

Подождите немного. Документ загружается.

A.3 Aircraft Scanners in the Visible and Infrared Regions 405

A.2.14

Ikonos

The Ikonos satellite was launched on 24 September 1999. It is in a sun-synchronous,

near polar orbit at an altitude of 681 km, with a repeat cycle of 35 days (although

a 1.5 day revisit capacity is possible with off-nadir pointing). It has a descending

equatorial crossing of 10:30 am, comparable to SPOT. Characteristics of its imaging

sensor are given in Table A.16.

Table A.16. Ikonos sensor characteristics

Spectral Bands IFOV Swath Dynamic Range

(

µm) (m) (km) (bits)

Panchromatic 0.45–0.90 1 × 111 11

Blue 0.45–0.53 4 × 411 11

Green 0.52–0.61 4 × 411 11

Red 0.64–0.72 4 × 411 11

Near IR 0.77–0.88 4 × 411 11

A.3

Aircraft Scanners in the Visible and Infrared Regions

A.3.1

General Considerations

Multispectral line scanners, similar in principle to the Landsat MSS and TM instru-

ments, have been available for use in civil aircraft since the late 1960’s and early

1970’s. As with satellite image acquisition it is the forward motion of the aircraft that

provides along track scanning whereas a rotating mirror or a linear detector array

provides sensing in the across track direction.

There are several operational features that distinguish the data provided by aircraft

scanners from that produced by satellite-borne devices. These are of signiﬁcance to

the image processing task. First, the data volume can be substantially higher. This is

a result of having (i) a large number of spectral bands or channels available and (ii) a

large number of pixels produced per mission, owing to the high spatial resolution

available. Frequently up to 1000 pixels may be recorded across the swath, with many

thousands of scan lines making up a ﬂight line; each pixel is normally encoded to at

least 8 bits.

A second feature of importance relates to ﬁeld of view (FOV) – that is the scan

angle either side of nadir over which data is recorded. This is depicted in Fig. A.4.

In the case of aircraft scanning the FOV, 2γ , is typically about 70 to 90

◦

. Such a

large angle is necessary to acquire an acceptable swath of data from aircraft altitudes.

By comparison the FOV for the Landsats 1 to 3 is 11.56

◦

while that for Landsats 4

and 5 is slightly larger at about 15

◦

. The consequence of the larger FOV with aircraft

406 A Missions and Sensors

Fig. A.4. The concept of ﬁeld of view (FOV) and instantaneous ﬁeld of view (IFOV)

scanning is that signiﬁcant distortions in image geometry can occur at the edges of

the scan. Often these have to be corrected by digital processing.

Finally, the attitude stability of an aircraft as a remote sensing platform is much

poorer than the stability of a satellite in orbit, particularly the Landsat 4 generation

for which the pointing accuracy is 0.01

◦

with a stability of 10

−6

degrees per second.

Because of atmospheric turbulence, variations in aircraft attitude described by pitch,

roll and yaw can lead to excessive image distortion. Sometimes the aircraft scanner

is mounted on a three axis stabilized platform to minimise these variations. It is more

common however to have the scanner ﬁxed with respect to the aircraft body and

utilize a variable sampling window on the data stream to compensate for aircraft roll.

Use of airborne multispectral scanners offers a number of beneﬁts. Often the user

can select the wavebands of interest in a particular application, and small bandwidths

can be used. Also, the mission can be ﬂown to speciﬁc user requirements concerning

time of day, bearing angle and spatial resolution, the last being established by the

aircraft height above ground level. As against these however, data acquisition from

aircraft platforms is expensive by comparison with satellite recording since aircraft

missions are generally ﬂown for a single user and do not beneﬁt from the volume

market and synoptic view available to satellite data.

A.3.2

Airborne Imaging Spectrometers

Since the mid 1980’s the availability of new detector technologies has made possible

the development of aircraft scanners capable of recording image data in a large

number, typically hundreds, of spectral channels. For a given pixel, enough samples

of its reﬂectance properties may be obtained by these instruments to allow very

accurate characterisation of the pixel’s spectral reﬂectance curve over the visible and

reﬂected infrared region. Because of the large number of channels the data sets are

often referred to as hyperspectral. These devices were the forerunners of instruments

such as Hyperion, treated in Sect. A.2.12.

A.4 Spaceborne Imaging Radar Systems 407

A good discussion of the development of imaging spectrometry may be found in

Goetz et al. (1985) and Vane and Goetz (1988). Much of the work on those devices

led to the development of similar spaceborne instruments.

Table A.17 summarises the characteristics of a selection of current aircraft imag-

ing spectrometers.

A.4

Spaceborne Imaging Radar Systems

A.4.1

The Seasat SAR

The ﬁrst earth observational space mission to carry a synthetic aperture imaging

radar was the Seasat satellite launched in June 1978. Although only short lived it

recorded about 126 million square kilometres of image data, including multiple cov-

erage of many regions. Several other remote sensing instruments were also carried,

including a radar altimeter, a scatterometer, a microwave radiometer and a visible

and infrared imaging radiometer. Relevant characteristics of the satellite and its SAR

are summarised in Table A.18. Polarization referred to in this table relates to the

orientation of the electric ﬁeld vector in the transmitted and received waveforms.

Free space propagation of electromagnetic energy, such as that used for radar, takes

place as a wave with electric and magnetic ﬁeld vectors normal to each other and

also normal to the direction of propagation. Should the electric ﬁeld vector be paral-

lel to the earth’s surface, the wave is said to be horizontally polarized. Should it be

vertical then the wave is said to be vertically polarized. A wavefront with a combi-

nation of the two will be either elliptically or circularly polarized. Even though one

particular polarization might be adopted for transmission, some rotation can occur

when the energy is reﬂected from the ground. Consequently at the receiver often both

vertically and horizontally polarized components are available, each having its own

diagnostic properties concerning the earth cover type being sensed. Whether one or

the other, or both, are received depends upon the antenna used with the radar. In the

case of Seasat, horizontally polarized radiation was transmitted (H ) and horizontally

polarized returns were received (H ).

Further details on the Seasat SAR will be found in Elachi et al. (1982).

A.4.2

Spaceborne (Shuttle) Imaging Radar-A (SIR-A)

A modiﬁed version of the Seasat SAR was ﬂown as the SIR-A sensor on the second

ﬂight of Space Shuttle in November of 1981. Although the mission was shortened

to three days, image data of about 10 million square kilometres was recorded. In

contrast to Seasat however, in which the ﬁnal image data was available digitally, the

data in SIR-A was recorded and processed optically and thus is available only in

ﬁlm format. For digital processing therefore it is necessary to have areas of interest

408 A Missions and Sensors

Table A.17. Imaging spectrometers

Instrument Spectral Spectral Dynamic IFOV Pixels per

range resolution range line

µm nm bits mrad

CASI-2 0.4 – 1 2.2 12 1.3 512

(Itres Research) (288 channels)

CASI-3 0.4 – 1.05 2.2 14 1.3 1490

(Itres Research) (288 channels)

DAIS 7915 0.4 – 1.0 15 – 30 15 3.3 512

(Georphysical (32 channels)

Environmental 1.5 – 1.8

Research Corp.) (8 channels) 45

2 - 2.5

(32 channels) 20

3–5

(1 channel) 2000

8 – 12.6

(6 channels) 900

AVIRIS 0.4 – 0.72 9.7 12 1 550

(Airborne Visible and (31 channels)

Infrared Imaging 0.69 – 1.30 9.6

Spectrometer – JPL) (63 channels)

1.25 – 1.87 8.8

(63 channels)

1.84 – 2.45 11.6

(63 channels)

MIVIS 0.433 - 0.833 20 12 2 765

(Daedalus (20 channels)

Enterprises Inc.) 1.15 – 1.55 50

(8 channels)

2.00 – 2.50

(64 channels)

8.20 – 12.70 ≤500

(10 channels)

HYDICE 0.4 – 2.5 7.6 – 14.9 12 0.5 320

(Hyperspectral (206 channels)

Digital Image

Collection Experiment

US Naval Research Labs)

HYMAP 0.44 – 0.88 16 12 2.5 ×2.0 512

(Integrated Spectronics 0.881 – 1.335 13

Pty Ltd) 1.4 – 1.81 12

1.95 – 2.5 16

(128 bands total)

A.4 Spaceborne Imaging Radar Systems 409

Table A.18. Characteristics of Seasat SAR, SIR-A, SIR-B, and SIR-C

digitized from ﬁlm using a device such as a scanning microdensitometer. A summary

of SIR-A characteristics is given in Table A.18, wherein it will be seen that the

incidence angle was chosen quite different from that for the Seasat SAR. Interesting

features of landform can be brought out by processing the two together.

More details on SIR-A will be found in Elachi et al. (1982) and E1achi (1983).

A.4.3

Spaceborne (Shuttle) Imaging Radar-B (SIR-B)

SIR-B, the second instrument in the NASA shuttle imaging radar program was car-

ried on Space Shuttle mission 41 G in October 1984. Again the instrument was

essentially the same as that used on Seasat and SIR-A, however the antenna was

made mechanically steerable so that the incidence angle could be varied during the

mission. Also about half the data was recorded digitally with the remainder being op-

tically recorded. Details of the SIR-B mission are summarised in Table A.18; NASA

(1984) contains further information on the instrument and experiments planned for

the mission. Because of the variable incidence angle both the range resolution and

swath width also varied accordingly.

A.4.4

Spaceborne (Shuttle) Imaging Radar-C (SIR-C)/X-Band Synthetic

Aperture Radar (X-SAR)

SIR-C/X-SAR, the third Shuttle radar mission, was carried out over two 10 day

ﬂights in April and September 1994. The SAR carried was the result of cooperation

between NASA and DARA, the German Aerospace Agency, and had the character-

istics indicated in Table A.18. Further details of the mission and the SAR will be

found in Stofan et al. (1995) and Jordan et al. (1995).

A.4.5

ERS-1,2

The European Remote Sensing Satellites ERS-1 and ERS-2 were launched in July

1991 and April 1995 respectively; they carry a number of sensors, one of which is a

410 A Missions and Sensors

Table A.19. Characteristics of free ﬂying satellite SAR systems

∗

5.3 GHz for ERS 1, 2 and Radarsat and 1.28 Ghz for JERS-1

synthetic aperture radar intended largely for sea state and oceanographic applications.

Characteristics of the radar are summarised in Table A.19.

A.4.6

JERS-1

The Japanese Earth Resources Satellite JERS-1 was launched in February 1992. It

carries two imaging instruments; one is an optical sensor and the other an imaging

radar. Table A.19 shows the design characteristics for the radar. The optical sensor,

called OPS, has 8 wavebands between 0.52 µm and 2.40 µm with a swath width of

75 km and a dynamic range of 6 bits. The optical pixel size is 18.3 m (across track)

× 24.2 m (along track). Provision is included for stereoscopic imaging.

A.4.7

Radarsat

Canada’s Radarsat was launched on 4 November 1995; its SAR is able to oper-

ate in the standard and six non-standard modes, one of which will give a 518 km

swath (Raney et al., 1991). Table A.19 lists the characteristics of the Radarsat SAR.

Radarsat-2, scheduled for launch in 2005, will have an ultra-ﬁne beam mode with

3 m resolution, and further polarisation options (VV, VH, HV).

A.4.8

Shuttle Radar Topography Mission (SRTM)

By deploying an outboard radar antenna 60 m from the space shuttle, along with the

main antenna in the cargo bay, two simultaneous images can be obtained of the same

region, but from different perspectives. Because of the coherent nature of the data,

the two images can be interfered to reveal topographic detail of the earth’s surface.

The Shuttle Radar Topography Mission in February 2000 used the SIR-C

(C band)/X-SAR system to acquire interferometric data from which approximately

80% of the earth’s land mass was imaged with 16 m absolute height accuracy and

20 m horizontal accuracy.

A.5 Aircraft Imaging Radar Systems 411

A.4.9

Envisat Advanced Synthetic Aperture Radar (ASAR)

The ASAR is an advanced version of the synthetic aperture radar from the ERS-1

and 2 missions. It operates at 5.331 GHz and incorporates a number of imaging modes

that provide a variety of resolutions, polarisations and swath widths. Generally, the

swath width is 100 km with the exception of wave mode (5 km) and wide swath

width and global monitoring (400 km) products.

A.4.10

The Advanced Land Observing Satellite (ALOS) PALSAR

ALOS is scheduled for launch in 2005, and is designed as a follow on to JERS-1 and

ADEOS (Midori). Besides PRISM (for stereoscopic mapping) and an AVNIR (see

Sect. A.2.7) ALOS will carry a phased array L band SAR, to be known as PALSAR.

The SAR will have a swath width of 70 km and a 2 look spatial resolution of 10 m

in its observation mode, and a swath width of 250–360 km with a spatial resolution

of 100 m in a scansar (wide swath width) mode.

A.5

Aircraft Imaging Radar Systems

Airborne imaging radar systems in SLAR and SAR technologies are also available.

As with airborne multispectral scanners these offer a number of advantages over

equivalent satellite based systems including ﬂexibility in establishing mission pa-

rameters (bearing, incidence angle, spatial resolution etc.) and proprietary rights to

data. However the cost of data acquisition is also high.

TableA.20 summarises the characteristics of three aircraft imaging radars, chosen

to illustrate the operating parameters of these devices by comparison to satellite

based systems. Note the band: wavelength designations – X: 0.030 m, C: 0.057 m,

L: 0.235 m, P: 0.667 m. Note also that interferometric operation is also possible with

the systems listed.

412 A Missions and Sensors

Table A.20. Representative aircraft synthetic aperture radar systems

CCRS

SAR DLR

ESAR JPL

AIRSAR

Wavebands X, C bands X, C,L&Pbands C,L&Pbands

Polarisation HH, VV multipolarisation multipolarisation

Range resolution 6 m, 20 m 1.5 m 10 m

Azimuth resolution 6 m, 10 m 4–12 m 1 m

Interferometry yes yes yes

Canada Centre for Remote Sensing

Deutsche Forschungsanstalt für Luft- und Raumfahrt

Jet Propulsion Laboratory

References for Appendix A

C. Elachi (Chairman), 1983: Spaceborne Imaging Radar Symposium, Jet Propulsion Labora-

tory, January 17–20, JPL Publication, 83–11.

C. Elachi, T. Bickell, R.L. Jordan and C. Wu, 1982: Spaceborne Synthetic Aperture Imaging

Radars. Applications, Techniques and Technology. Proc. IEEE, 70, 1174–1209.

NASA, 1984: The SIR-B Science Investigations Plan, Jet Propulsion Laboratory Publication,

84–3.

A.F.H. Goetz, G.Vane, T.E. Solomon and B.N. Rock, 1985: Imaging Spectrometry for Earth

Remote Sensing, Science, 228, 1147–1153.

R.L. Jordan, B.L. Huneycutt and M. Werner, 1995: The SIR-C/X-SAR Synthetic Aperture

Radar System. IEEE Trans. Geoscience and Remote Sensing, 33, 829–839.

R.K. Raney, A.P. Luscombe, E.J. Lanham and S. Ahmed, 1991: Radarsat. Proc. IEEE, 79,

839–849.

E.R. Stofen, D.L. Evans, C. Schmullius, B. Holt, J.J. Plaut, J. van Zyl, S.D. Wall and J.

Way, 1995: An Overview of Results of Spaceborne Imaging Radar-C, X-Band Synthetic

Aperture Radar (SIR-C/X-SAR). IEEE Trans. Geoscience and Remote Sensing, 33, 817–

828.

G. Vane and A.F.H. Goetz, 1988: Terrestrial Imaging Spectroscopy, Remote Sensing of Envi-

ronment, 21, 311–332.

Appendix B

Satellite Altitudes and Periods

Civilian remote sensing satellites are generally launched into circular orbits. By

equating centripetal acceleration in a circular orbit with the acceleration of gravity

it can be shown (Duck and King, 1983) that the orbital period corresponding to an

orbital radius r is given by

T = 2π



/µ (B.1)

where µ is the earth gravitational constant, with a value of 3.986 ×10

−2

. The

corresponding orbital angular velocity is

θ =



µ/r

rad · s

−1

(B.2)

The orbital radius r can be written as the sum of the earth radius r

and the altitude

of a satellite above the earth, h:

r = r

+ h (B.3)

where r

= 6.378 Mm. Thus the effective velocity of a satellite over the ground (at

its sub-nadir point) ignoring earth rotation, is given by

v = r

θ = r



µ/(r

+ h)

(B.4)

The actual velocity over the earth’s surface taking into account the earth’s rotation

depends upon the orbit’s inclination (measured as an angle i anticlockwise from the

equator on an ascending pass – i.e. at the so-called ascending node) and the latitude

at which the velocity is of interest.

Let the earth rotational velocity at the equator be v

(to the east). Then at latitude φ,

the surface velocity of a point on the earth will be v

cos φ cos i. Therefore the actual

ground track velocity of a satellite at altitude h and orbital inclination i is given by

= r



µ/(r

+ h)

± v

cos φ cos i (B.5)

where the + sign applies when the component of earth rotation opposes the satellite

motion (i.e. on descending nodes for inclinations less than 90

◦

or for ascending nodes

with inclinations greater than 90

◦

). Otherwise the negative sign is used.

414 B Satellite Altitudes and Periods

Fig. B.1. Satellite periods versus altitude above the earth’s surface, for circular orbits

Equations (B.1) to (B.5) can be used to derive some numbers of signiﬁcance for

remote sensing missions, although it is to be stressed here that the equations apply

only for circular orbits and a spherical earth.

Figure B.1 shows a plot of orbital period (in minutes) as a function of satellite

altitude plotted on logarithmic coordinates. This has been derived from (B.1) and

(B.3). Some signiﬁcant altitudes to note are (i) h = 907 km, at which T = 103

min; being the approximate period of the ﬁrst three Landsat satellites, (ii) h =

35,800 km at which T = 24 hours, being the so-called geosynchronous orbit – if

this is established over the equator then the satellite appears stationary to a point on

the ground; this is the orbit used by many communication satellites, (iii) h = 380

Mm at which T = 28 days – this is the orbit of the moon.

Consider now a calculation of the time taken for Landsat 1 to acquire a 185 km

frame of MSS data. This can be found by determining the local velocity. For the

Landsat satellite the orbital inclination is 100

◦

; at Sydney Australia the latitude is

◦

S. From (B.5) this gives

= 6.392 km s

−1

Therefore 185 km requires 28.9 s to record.

References for Appendix B

K.I. Duck and J.C. King, 1983: Orbital Mechanics for Remote Sensing. In: R.N. Colwell (Ed.).

Manual of Remote Sensing, 2e, American Society of Photogrammetry, Falls Church.