Guide to Magellan Image Interpretation

 

Chapter 2. Magellan Image Data

John P. Ford and Jeffrey J. Plaut

 

Introduction

[7] The Magellan radar mapping mission produced the first global, high-resolution (~ 100-m) image data set of Venus [Saunders et al., 1992]. Images were obtained from September 1990 to September 1992 during three mapping cycles (Figure 2-1). A total of 4225 usable SAR imaging orbits was obtained. The area covered by each orbit is typically 20 km wide and 17,000 km long. Ninety-eight percent of the surface was imaged and many areas were viewed more than once with different imaging geometries and/or directions of illumination. Magellan image data products consist of mosaics in a variety of formats and scales. Figure 2-2 is a global radar mosaic. Details concerning the characteristics of the mosaics and the images, the imaging cycles, and information about the availability of data and data products are presented in this chapter.

 

Image Data Records

Raw SAR data received from the Magellan spacecraft are processed into basic image data records (BIDRs). Each BIDR strip is accompanied by a corresponding altimeter/radiometer composite data record (ARCDR) for each orbit. In the altimeter mode, nadir-directed radar pulses measured the distance between the spacecraft and the Venusian surface. In the radiometric mode, the radar system monitored the microwave thermal emission of the surface. The primary image data product is the full-resolution BIDR (F-BIDR). Because the long, narrow format of the F-BIDR image is not useful for geologic mapping, adjacent image strips are assembled into mosaicked image data records (MIDRs).

Details of SAR image acquisition can be found in the Experimenter's Notebook. This computer file provides an investigator with the orbital elements, uplink commands,.....

 


Figure 2-1. Mission mapping baseline and gravity-data acquisition plan.

Figure 2-1. Mission mapping baseline and gravity-data acquisition plan.


[
8]

Figure 2-2. Global radar mosaic of Venus.

Figure 2-2. Global radar mosaic of Venus.

 

[9] ...downlink text observations, and radar imaging geometry (including incidence angles and image coordinates) that were effective during the course of the mission. The image data obtained between 89°N and 89°S latitude have been processed to a sinusoidal equal-area map projection. Within 10 deg of the poles, the data were also processed to an oblique sinusoidal projection.

 

Mosaicking

Orbit determinations based on short data arcs from a single day's tracking provide the basis for SAR processing, locating individual pixels, and mosaicking the image strips. The final ephemeris data are sufficiently accurate to allow mosaicking of image data strips by a process of dead reckoning without the use of tie points.

Mosaicking algorithms have been developed to produce smooth boundaries between adjacent orbits. In some mosaicked images, however, a dark or bright banding runs parallel to the orbit boundaries. This banding is the result of small differences in antenna pointing, spacecraft navigation, and topographic modeling that resulted in the application of slightly inaccurate radiometric factors in the processor.

Image mosaics consisting of 7168 lines x 8192 samples provide the base for standard Magellan photographic image products (MIDRs) (Figure 2-3). Full-resolution mosaicked image data records (F-MIDRs) use the full resolution (75 m) of the input F-BIDRs, and cover an area approximately 5 deg on a side. Each F-MIDR has a photomosaic identifier in the general form of F-MIDRP.xxHyyy;C, where P stands for photomosaic, H represents the hemisphere (N for north; S for south), xx is approximate latitude (deg) and yyy is approximate longitude (deg) at the center of the mosaic, and C denotes the cycle (1, 2, or 3) in which the data were acquired. In Cycle 1, C is a single digit corresponding to the version number. In Cycles 2 and 3, three-digit numbers are used: The first digit indicates the cycle and the last is the version number. For example, F-MIDRP.60N026;301 is an F-MIDR photomosaic centered near 60°N latitude and 26°E longitude from Cycle 3 and is version 1.

Mosaics of larger areas are obtained by compressing the resolution of the BIDRs. Three levels of compressed MIDRs have been produced (C l-MIDR, 15 deg on a side and 225-m resolution; C2-MIDR, 45 deg on a side and 675-m resolution; C3-MIDR, 120 x 80 deg and 2025-m resolution; see Figure 2-3). Each stage of compression involves the spatial averaging of 3 x 3 arrays of pixels from the next higher resolution image data record. Thus, Cl-MIDRs are produced from C1 -BIDRs. C2-MIDRs and C3-MIDRs are compressed from C1 -MIDRs. Each of the C1-, C2-, and C3-MIDR photomosaic identifiers denotes the coordinates of the mosaic center and the acquisition cycle as described above for the FMIDRs.

After mosaicking, the images are contrast enhanced, annotated, and masked to produce photographic output data for scientific analysis. Unenhanced digital data are stored on computer-compatible tapes (CCTs), write-once-read-many (WORM) optical disks, and compact disks read-only memory (CD-ROMs); in addition, these data have been written to film. The CD-ROMs also contain numerous documentation files.

 

Resolution

In the F-BIDRs, the image pixel size is sampled to 75 m in both cross-track and along-track directions. The spatial resolution of the images in the cross-track direction and the number of looks used to process the data vary with latitude. The number of looks corresponds with the number of independent observations made for each resolution element. Increasing the number of looks reduces the speckle noise that occurs in coherent radar imaging, but it also reduces the spatial resolution of the images. The number of looks and the pixel resolution as functions of latitude for the first mission cycle are listed in Table 2-1.

 


Table 2-1. Number of looks and pixel resolution for Cycle 1 image data
a

Latitude deg

Number of looks

Cross-track resolution, m

Along-track resolution, m

.

90

14

250

110

82

15

211

110

60

9

155

110

45

8

125

110

30

6

108

110

10

5

101

110

-10

5

108

110

-25

5

125

110

-40

11

155

110

-62

15

211

110

-70

14

250

110

.

a From Saunders et al., 1992.


 

Pixel values in the F-BIDRs represent the observed red, backscatter coefficient relative to an empirically derived formula [Muhleman, 1964] that describes the average backscatter math symbols: signa superscript 0 (theta) of the Venusian surface as a function of incidence angle 0. This formula states

 

mathematical equation

 

where the constants K1 = 0.0188 and K2 = 0.111.


[
10]

Figure 2-3. Image data products.

Figure 2-3. Image data products.

 

In lieu of these intended values, the SAR processor was implemented with erroneous values of K1 = 0.0118 and Greek letter theta =Greek letter theta + 0.5 deg. Thus, pixel values in the F-BIDRs are based on an approximation of the intended backscatter function Greek letter sigma superscript 0(a) where:

 

mathematical equation

 

The pixel DN values range from I to 251. A DN value of 101 corresponds to a measured backscatter function equal to that of the approximation model (a DN of 111 represents the expected value according to the correct Muhleman model). The relative backscatter coefficient Greek letter sigma superscript 0
of other pixel DN values is given by

 

Greek letter sigma superscript 0
= 0.2 (pixel DN- 101) dB

 

Non-SAR Data

Venusian topography, roughness, and compositional characteristics of the surface cannot be determined from image data alone; altimetry and radiometry measurements have provided important additional information necessary for [11] the interpretation of these parameters. Details of these nonSAR experiments are outlined in Chapter 3.

 

Image Brightness

Brightness variations in Magellan images result primarily from three different surface variables: (1) topographic effects, (2) roughness, and (3) electrical properties, each of which is influenced to some extent by variations in the radar incidence angle and/or the direction of illumination. A guide to the interpretation of these effects in the images is provided in Chapters 5 through 9.

At the scale of the radar resolution, topography may have a pronounced effect on image brightness; terrain that slopes toward the imaging sensor appears bright and spatially compressed relative to terrain that slopes away, which appears dark and spatially expanded. Where the incidence angle Greek letter theta
< 20 deg, small changes in slope give large changes in backscatter. In cases where broad regional slopes are too small to provide brightness contrast, altimetry data are needed to distinguish large-scale positive and negative relief.

In the absence of topographic effects, surface roughness at the scale of the radar wavelength may dominate the backscatter where the incidence angle Greek letter theta
> 20 deg and < 60 deg. Rough surfaces have a notably higher backscatter than smooth surfaces. Though the Magellan data have been normalized to a model that describes the average backscatter of Venus (see above), surface roughness can strongly influence image brightness.

In addition to topographic or surface-roughness effects, the intrinsic reflectivity properties of the surface can have a major influence on image brightness; high dielectric constants (controlled by composition and/or bulk density) enhance radar backscatter.

 

Cycles of Coverage

Each of the three mapping cycles was limited in the amount of image coverage that could be obtained. After the first mapping cycle, image acquisition was directed toward filling gaps and obtaining data from different viewing angles and directions. As the spacecraft orbited from north to south, the SAR when pointed east was "left-looking"; when pointed west it was "right-looking." Variations of incidence angle with latitude in each of the three cycles are listed in Table 4-1 and shown in Figure 4-3.

 

Cycle 1

The purpose of Cycle 1-from mid-September 1990 to mid-May 1991-was to acquire radar mapping data of 70% of the surface at a resolution better than 300 m (Figure 2-1). Because of the elliptical orbit, the spacecraft altitude varied from about 2000 km near the north pole to 290 km at 9.5°N latitude (periapsis). To maintain an acceptably high signal-to-noise ratio throughout each mapping orbit and compensate for the increasing signal loss as the distance to the surface increased, it was necessary to decrease the incidence angle progressively with increasing latitude on either side of periapsis.

Left-looking images were obtained at incidence angles that varied from about 45 deg at periapsis to about 16 deg at high latitudes (Figure 4-3). About 83.7% of the surface was covered (Figure 2-4). Superior conjunction (when Venus passed behind the Sun) precluded the acquisition of data from 110 of the 1790 orbits needed to obtain a full 360 deg of longitudinal coverage. Other gaps resulted from geometric limitations (1.5%) that precluded viewing the south polar region and data losses (8.6%) caused by overheating of the electronics system, failure of one of the onboard tape recorders, attitude-control anomalies, and erroneous attitude updates.

 

Cycle 2

Cycle 2-from mid-May 1991 to mid-January 1992- was dedicated to filling gaps caused by superior conjunction in Cycle I and obtaining coverage of the south polar area. The latter requirement necessitated reorientation of the spacecraft in a right-looking geometry. The images were obtained with a mostly constant incidence angle of about 25 deg in the latitudes from 65°N to 45°S and with smaller angles toward the south pole (Figure 4-3).

During the data playbacks in Cycle 2, the spacecraft electronic bays were overheated by solar illumination. To compensate for this, the spacecraft was cooled by pointing the high-gain antenna toward the Sun. Since tape playback could not be relayed to Earth during these Sun-pointing periods, the amount of mapping data obtained was significantly less than in Cycle 1. The surface coverage in Cycle 2 amounts to 54.5% (Figure 2-5); cumulative coverage of the surface rose to 96%.

An eight-orbit stereo test in mid-Cycle 2 showed that Magellan could revisit areas mapped in Cycle I with a different left-looking incidence angle and successfully produce same-side stereo image pairs. This stereo data test pointed the way for acquiring stereo coverage in mission Cycle 3.

An error in the implementation of the SAR processing algorithm caused a slight degradation in the quality of the images during Cycles 2 and 3. The error is manifested as a....

 


[
12]

Figure 2-4. Mosaic of Cycle-1 coverage (left-looking).

Figure 2-4. Mosaic of Cycle-1 coverage (left-looking).

 

[13] ....reduction in the dynamic range (gray levels) of the images. The lower end of the brightness scale was "clipped," such that all pixels intended to have DN values between I and about 75 were reassigned to a narrow range near 75. This resulted in a loss of detail in the dark regions of the images. The data were eventually reprocessed to restore the full dynamic range, but most image mosaics (F-MIDRs and C-MIDRs) from Cycles 2 and 3 were constructed from the reduced dynamic range data. The orbits affected by this problem range from 2773 to 4515.

Cycle 2 orbits 2689 through 2901 were originally processed with a topographic model different than that used for the remainder of the Magellan SAR data. The model was intended to improve the location of features, but unfortunately it produced greater distortion. The orbits were eventually reprocessed, but artifacts from the original processing appear in most of the mosaics that include images from these orbits. One area of study that can be affected by these distortions is height determination based on opposite-side stereo pairs (see Chapter 4).

 

Cycle 3

Cycle 3-from mid-January to mid-September 1992- emphasized the acquisition of stereo image coverage. The ability to obtain data during this period was marked by both success and failure. About 21.3% of the surface was covered by left-looking images at incidence angles mostly smaller than those used in Cycles 1 and 2 (Figures 2-6 and 4-3). Additional coverage of Maxwell Montes was obtained at incidence angles larger than those in Cycle 1. Failure of the primary downlink transmitter and deterioration of the secondary transmitter reduced the signal-to-noise ratio in the data and created numerous gaps in coverage. Loss of data also occurred during solar occultations. However, an additional 2% of the planet was covered at the nominal Cycle 1 incidence angle profile in the final two weeks of this cycle. The cumulative coverage of the surface rose to 98%.

 

Cycle 4

At the termination of radar mapping in Cycle 3, the mission was dedicated to gravity observations in Cycle 4. Gravity data can be collected only when periapsis is unocculted from Earth. This condition existed for the full duration of Cycle 4. Gravity data were acquired by pointing Magellan's high-gain antenna toward Earth and recording changes in the radio signal's Doppler shift. In this mode, it is not possible to acquire image or altimetry data. The gravity data were reduced to measure spacecraft velocity changes as small as 0.1 mm sec-1. Gravity maps will indicate density variations in the planet's interior. Because of the elliptical orbit, the gravity resolution varies as a function of altitude. It is highest at periapsis and lowest at the polar regions. Thus, high-resolution gravity data can be obtained only at low latitudes. An aerobraking maneuver to circularize the orbit will equalize the gravity resolution around the planet (Figure 2-1). This will enable the acquisition of uniformly high-resolution gravity data at all latitudes.

 

Availability of Data

To ensure that a complete and well documented data set is available to the scientific community and to the general public, Magellan CCTs and CD-ROMs have been distributed to the National Space Science Data Center (NSSDC). The NSSDC is the principal repository and distributor of Magellan CD-ROMs. Some SAR mosaics have been combined with digital elevation data to produce perspective views. Programs on video cassettes have been created that use thousands of perspective scenes incrementally to simulate flights over the surface of Venus. Information on the availability of these and other Magellan products is given in the appendix.

 


[
14-15]

Figure 2-5. Mosaics of Cycle-2 coverage: (a) left-looking;

Figure 2-5. Mosaics of Cycle-2 coverage: (b) right-looking.

Figure 2-5. Mosaics of Cycle-2 coverage: (a) left-looking; (b) right-looking.


[
16-17]

Figure 2-5. Mosaics of Cycle-3 coverage: (a) left-looking;

Figure 2-5. Mosaics of Cycle-3 coverage: (b) right-looking.

Figure 2-5. Mosaics of Cycle-3 coverage: (a) left-looking; (b) right-looking.

 

[18] References

- Muhleman, D.O., 1964, "Radar scattering from Venus and the Moon," Astron. J., v.69, p. 34-41.

- Saunders, R.S., et., 1992, "Magellan mission summary," J.Geophys. Res., v.97, no. E8, p. 13,067-13,090

 
previousindexnext