The Volcanic and Tectonic History of Enceladus

The Volcanic and Tectonic History of Enceladus

ICARUS 119, 385–404 (1996) 0026 ARTICLE NO. The Volcanic and Tectonic History of Enceladus JEFFREY S. KARGEL U.S. Geological Survey, Flagstaff, Ari...

676KB Sizes 7 Downloads 87 Views

ICARUS

119, 385–404 (1996) 0026

ARTICLE NO.

The Volcanic and Tectonic History of Enceladus JEFFREY S. KARGEL U.S. Geological Survey, Flagstaff, Arizona 86001 E-mail: [email protected] AND

STEFANIA POZIO Reparto di Planetologia, Viale dell’ Universita` 11, 00185 Rome, Italy Received April 17, 1995; revised September 28, 1995

Enceladus has a protracted history of impact cratering, cryovolcanism, and extensional, compressional, and probable strike–slip faulting. It is unique in having some of the outer Solar System’s least and most heavily cratered surfaces. Enceladus’ cratering record, tectonic features, and relief elements have been analyzed more comprehensively than done previously. Like few other icy satellites, Enceladus seems to have experienced major lateral lithospheric motions; it may be the only icy satellite with global features indicating probable lithospheric convergence and folding. Ridged plains, 500 km across, consist of a central labyrinthine ridge complex atop a broad dome surrounded by smooth plains and peripheral sinuous ridge belts. The ridged plains have few if any signs of extension, almost no craters, and an average age of just 107 to 108 years. Ridge belts have local relief ranging from 500 to 2000 m and tend to occur near the bottoms of broad regional troughs between swells. Our reanalysis of Peter Thomas’ (Dermott, S. F., and P. C. Thomas, 1994, The determination of the mass and mean density of Enceladus from its observed shape, Icarus, 109, 241– 257) limb profiles indicates that high peaks, probably ridge belts, also occur in unmapped areas. Sinuous ridges appear foldlike and are similar to terrestrial fold belts such as the Appalachians. If they are indeed folds, it may require that the ridged plains are mechanically (perhaps volcanically) layered. Regional topography suggests that folding may have occurred along zones of convective downwelling. The cratered plains, in contrast to the ridged plains, are heavily cratered and exhibit extensional structures but no obvious signs of compression. Cratered plains contain a possible strike–slip fault (Isbanir Fossa), along which two pairs of fractures seem to have 15 km of right-lateral offset. The oldest cratered plains might date from shortly after the formation of the saturnian system or the impact disruption and reaccretion of Enceladus. Another area of cratered plains has modified craters (e.g., Ali Baba and Aladdin), which some workers have explained by anomalous heat flow and viscous relaxation; lateral shear and shield-building volcanism also may have been important. A young rift-like structure (northern Samarkand Sulci) has few craters and a

concentration of cracks or grabens and flattened, flooded, and rifted craters. Pit chains and cratered domes suggest explosive volcanism. Smooth plains may have formed by cryovolcanic equivalents of flood-basalt volcanism. Pure H2O would be difficult to extrude through an icy crust and is cosmochemically improbable as a cryovolcanic agent. Density relations rule out eutectic brine lavas on Enceladus, but NH3 –H2O volcanism is possible. Current steady-state tidal dissipation may cause melting of ammonia hydrate at a depth of p25 km if the crust is made of ammonia hydrate or p100 km if it is made of water ice.  1996 Academic Press, Inc.

1. INTRODUCTION

Enceladus has a complex surface (Figs. 1–4) and a dynamic history despite its diminutive stature among icy satellites (its disk, diameter 499 km, could fit within Arizona). Pre-Voyager models suggested that large icy satellites should be geologically evolved but Enceladus should not be (Consolmagno and Lewis 1978, Peale et al. 1980). Hints to the contrary came in 1980 from (1) Earth-based observations, which showed that Saturn’s tenuous E-ring may have been ejected from Enceladus (Baum et al. 1981), and (2) low-resolution Voyager 1 images of Enceladus, which showed a high albedo and no large craters. Enceladus’ extraordinary nature was recognized in Voyager 2’s highresolution (1–3 km/pixel) images of p43% of the satellite, which showed evidence of a protracted geologic history with episodes of volcanism and tectonism (Smith et al. 1982) (Figs. 1–3). Many publications have contributed knowledge of Enceladus’ shape, mean density, surface features, and composition, photometry, geologic history, thermal history, resurfacing mechanisms, and associations with Saturn’s E-ring. There has been no previous complete tectonic mapping of Enceladus in areas covered by Voyager 2 images, no tectonic mapping at all in conventional projec-

385 0019-1035/96 $18.00 Copyright  1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

386

KARGEL AND POZIO

FIG. 1. Image of Enceladus (diameter 499.4 km). White lines show locations of photoclinometric profile a across Isbanir Fossa, profile b across a trough, and profiles c and d across Samarkand Sulci. Resolution 1.07 km per pixel. Voyager image 44004.32.

tions, and no comprehensive analysis of Enceladus’ cratering record as observed by Voyager 2. In this paper, we examine Enceladus’ tectonic and cratering histories and surface landforms more completely than ever before. Digital limb profiles (originally reported by Dermott and Thomas (1994) and generously provided by Peter Thomas) and photoclinometric profiles have been used in a detailed analysis and modeling of Enceladus’ regional topography and local relief. Enceladus’ cratering record is used to deduce a relative history of cryovolcanism and tectonism. 2. OBSERVATIONS

2.1. Methods Tectonic mapping. Tectonic features on Enceladus were mapped using the U.S. Geological Survey (1992) air-

brush map (Fig. 4) as a controlled base, processed images for interpretation and measurement of selected features, and limb and photoclinometric profiles for validation of tectonic mapping and measurement of relief. A tectonic map was prepared by each author, first independently and then collaboratively (Pozio and Kargel 1989). Topographic profiles produced by limb fitting and photoclinometry (Figs. 6 and 7) were used for confirmation or revision of our interpretations and production of the final tectonic and geologic terrain maps of Enceladus (Fig. 5). Crater counts. Abrupt changes in the density of craters on Enceladus occur along major tectonic contacts, a fact recognized during early studies of Voyager data (Smith et al. 1982). The most extensive previous study of Enceladus’ cratering record was by Plescia and Boyce (1983), who determined crater densities for four sample regions totaling

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

FIG. 2. Image showing terrain near terminator of Enceladus. White lines show locations of photoclinometric profiles across three craters with central domes (Ali Baba, profile f; Aladdin, profile g; and a smaller crater west of Aladdin, profile h). Arrows identify a pit chain discussed in text. Resolution 0.99 km per pixel. Voyager image 44004.36.

387

36,000 km2 (p5% of the surface) and containing 112 craters larger than 5 km. Extending this type of data to the whole surface of Enceladus visible at high resolutions, we attempted to count and measure every observable crater larger than 2.83 (2Ï2) km in diameter (Pozio and Kargel 1990). This lower limit is too small for complete counts, but it ensures that counts at some larger diameter are as complete as possible. Rim-to-rim diameters were measured and grouped by terrain (Fig. 5b) into Ï2 diameter bins. A summary of results for craters larger than 5.66 km (p6 pixels in the best images) is given in Table I and Fig. 8. Counts for craters $5.66 km in diameter (453 in all) appear to be virtually complete for almost all of the 43% (337,000 km2) of Enceladus’ surface covered by high-resolution images (counts near the subsolar point or too near the limb were not attempted). Counts for craters smaller than 5.66 km in diameter generally are incomplete due to resolution limits. Limb profiles. Limb measurements taken from Voyager 2 images of Enceladus were originally made and mathematically fitted by Dermott and Thomas (1994), who analyzed their data extensively in terms of the shape of Enceladus but gave little attention to geology. Those authors gave graphical presentations of 5 of their 11 profiles, but these were published at too small a scale to be very useful for geological analysis. Peter Thomas generously provided all 11 digital profiles to aid this study (Fig. 6). Figure 4 shows the locations of the profile lines, numbered in order of decreasing distance of Voyager 2 to Enceladus (increasing spatial resolution). The limb profiles are residuals after subtraction from the limb figures of best-fit triaxial ellipsoids, which account for .90% of variance from sphericity. Dermott and Thomas (1994) analyzed their fits with the assumption that Enceladus is in hydrostatic equilibrium; their analysis yielded a precise calculation of Enceladus’ mean density (either 1.00 6 0.03 g cm23 or 1.12 6 0.05 g cm23, depending on assumptions). The residuals are interpreted as local and regional topography. Vertical precision of nearby pixels is estimated as better than 2s p 0.2 pixels (200–700 m, image dependent). Uncertainties in positioning of reseau marks used in limb fitting is 60.1 pixels, which may have introduced low-frequency errors with amplitudes up to 0.2 pixels (200–700 m) and wavelengths of 80 pixels (85–285 km). The total error of residuals (elevation accuracy) is estimated as 300–1000 m (image dependent), which includes a high-frequency error component due to imprecision in limb locations and a low-frequency error due to reseaumark positioning. Locations of sharp features in the airbrush map (Fig. 4) and in limb profiles are mostly consistent to 62 km. Minor limb features can be identified in each of a set of closely spaced profiles. Major limb features commonly have corresponding markings in the shaded-relief map.

388

KARGEL AND POZIO

FIG. 3. Image centered on ridged plains. Arrows identify a cratered dome discussed in text. Resolution 1.8 km per pixel. Voyager image 44000.48.

These internal consistencies validate the accuracy of the high-frequency component (local relief) of the limb profiles and suggest that pixel-to-pixel errors may be better than the 0.2 pixels (2s ) estimated above. Some of the lowfrequency undulations in each set of adjacent closely spaced profiles are consistent and geologically correlatable, but in some cases a systematic discrepancy is present; if not due to intervening topography such as ridge belts, such discrepancies may be accounted for by errors in the positioning of reseau marks.

Photoclinometry. Selected features were analyzed by photoclinometry (transects in Figs. 1 and 2; profiles in Fig. 7). DN (image intensity) was measured on a transect orthogonal to the terminator and passing through the feature of interest, initially avoiding steep slopes. A fourthorder polynomial was fitted to DN vs distance for each image and transect to give the photometric response function. Root-mean-square departure between the measured and fitted values gave 2s errors in elevations of adjacent pixels (bars in Fig. 7); these errors are upper limits, because

FIG. 4. Shaded relief map of Enceladus (U.S. Geological Survey 1992). Annotations give place names. Limb profile lines, from Dermott and Thomas (1994), are discussed in Section 2.1. Circled areas marked a–d in unmapped or poorly mapped regions are where limb profiles consistently indicate the existence of major uncharted ridges, as discussed in text.

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

389

390

KARGEL AND POZIO

FIG. 5. (a) Tectonic map of imaged part of Enceladus. Bold lines represent prominent ridges, troughs, and scarps; thin lines represent more subtle features. (b) Terrain map of Enceladus. Text describes the major terrains and their subunits, including areas of cratered plains (Cp), ridged plains (Rp), rifted terrain (Rt), and banded terrain (Bt).

we assumed that differences between the observed DN and the polynomials are due entirely to noise, albedo variations, or other factors other than surface topography. DN then was measured along a smaller interval crossing the feature(s) of interest. To increase signal:noise, DN was averaged across 3-pixel-wide transects. The polynomial was used to convert DN to local slope relative to Sun and then a slope relative to horizontal. Elevation changes were integrated to obtain topographic profiles (Fig. 7). When ending and starting elevations were substantially different,

the profile was recalculated with a tilt to equate these elevations. Relative elevations for points more than about 20 pixels apart are meaningless. After an initial trial we decided against the use of a theoretical photometric function, such as the Lommel– Seeliger function used by Passey (1983), in favor of an empirical approach. Albedo variations on Enceladus are small (Verbiscer and Veverka 1994), making an empirical approach more attractive and capable of fitting the observations very well over the entire range of brightness longi-

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

391

FIG. 5—Continued

tudes. By contrast, the Lommel–Seeliger function fits the observations well at small brightness longitudes but not at brightness longitudes .45–508 (Passey 1983). Many features of special interest, such as Enceladus’ ridges, occur at these high brightness longitudes, where the Lommel– Seeliger function does not perform well. 2.2. Interpretation of Enceladus’ Tectonic Features Overview. We have delineated four major geologic terrains on the basis of tectonic relationships and large abrupt changes in crater density (Fig. 5). These terrains include cratered plains (Cp), ridged plains (Rp), rifted terrain (Rt), and banded terrain (Bt). Further subdivisions were made on the basis of minor changes in crater morphology or density and associated minor tectonic contacts. Some cratering variations are laterally gradational, so that the subdivision of terrains is subjective.

Figure 5a differs from earlier tectonic maps (Smith et al. 1982, Passey 1983) in that (1) it covers in standard map projections the entire area of high-resolution Voyager images, and (2) it distinguishes ridges, troughs, scarps, pit chains, and lineaments of uncertain character. Tectonic features appear to be either principally negative-relief forms (pit chains, troughs, and scarps, thought to be mainly extensional) or positive-relief ridges (thought to be mainly compressional). Interpretation of tectonic features and descriptions of terrains. Interpretations of tectonic features in Fig. 5a as positive or negative were verified whenever possible with limb profiles and photoclinometry. Limb profiles also were used to interpret lineaments that are barely visible on the disk but are readily seen on the limb. Positive relief features are observed in limb profiles more readily than negative ones. Figure 6 has guides drawn between representative

392

KARGEL AND POZIO

FIG. 6. Limb profiles of Enceladus originally measured by Dermott and Thomas (1994) from 11 images. Interpretations of features discussed in text. Profile lines shown in Fig. 4. Voyager 2 image numbers (FDS numbers) for limb profiles 1–11, in that order, are: 43993.12, 43993.16, 43993.20, 43996.02, 43997.25, 43997.29, 43997.37, 43999.00, 44000.42, 44000.44, and 44004.12. Sets with closely spaced profile lines are shown together on one graph. Tie lines connect representative features that appear in more than one profile of a given set. Ridges shown in the tectonic map (Fig. 5a) are indicated by ‘‘r,’’ troughs are indicated by ‘‘t.’’ The longitudinal extents of major geologic terrains and geographic features crossed by the profiles are indicated. The panel showing blown-up portions of profiles 5, 6, and 7 have bars indicating the positions of troughs (labeled ‘‘t’’) and ridges (unlabeled bars) that show up in the USGS shaded-relief map (Fig. 4).

limb features that appear in two or more profiles. Many of these features also appear in Figs. 4 and 5a. Passey’s (1983) previous photoclinometry of Enceladus, confirmed by us, indicate that (1) many craters are abnormally shallow compared to fresh lunar craters and (2) sinuous ridges range from p500 to 1500 m high. (We obtained maximum relief of p2000 m.) Passey’s (1983) results are difficult to use since they had poor resolution and were shown without vertical exaggeration or removal of planetary curvature. Our new measurements supplement Passey’s with analysis of several additional types of features. We did not analyze many craters because Passey (1983) presented numerous such measurements. Some peaks that consistently appear in two or more limb profiles in areas where these profiles nearly intersect

apparently correspond to ridges in unmapped or poorly mapped areas. These findings, marked in Fig. 4, include the following. (a) A set of peaks with marginal troughs in profiles 9 and 10 (Fig. 6) near lat 228 S, long 3448. These peaks, a possible southwestern extension of Samarkand Sulci, correspond to faint markings in Fig. 4. (b) Prominent peaks near lat 478 S, long 2408 in profiles 4, 5, 6, 7, and 8 (Fig. 6). Peaks are from p400 to 1000 m high; peaks in each profile occur atop a broad rise p200–250 km across. (c) A set of peaks on a broad rise near long 275–2808 in profiles 8, 9, and 10 (Fig. 6). (d) A mountainous area near lat 458 S, long 2158–2258

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

393

FIG. 6—Continued

FIG. 6—Continued

in profiles 9 and 10 (Fig. 6). These peaks are a possible southward extension of Harran Sulci (Fig. 4). Squyres et al. (1983) sketched major tectonic features and geologic terrains on one face of Enceladus. They recognized one substantial ridge; because it is bounded on both sides by troughs, they attributed to it an extensional origin. They also drew an analogy between Enceladus’ ridged plains and Ganymede’s grooved terrain. It is not obvious to us that this proposed analogy is very close. Our work, like that of Smith et al. (1982), indicates a complex tectonic style involving many ridges and many troughs. Although closely spaced ridges may be difficult to distinguish from closely spaced troughs, the limb and photoclinometry profiles shown in Figs. 6 and 7, as discussed further below, leave little doubt that many lofty ridges occur in some areas of Enceladus. Mappers have taken two approaches to mapping what we call cratered plains. Smith et al. (1982) broke this unit into smooth plains, cratered plains, and cratered terrain.

Passey (1983) identified the whole region as cratered terrain and subdivided it into subunits somewhat as we have done. Like all previous workers, we recognize the obvious lateral variations in crater density in the cratered plains. We subdivided this terrain accordingly. We do not perceive definitive places to draw sharp contacts in the cratered plains, which have statistically gradational variations in crater density and morphology, so that the subdivisions are subjective. Rectilinear troughs, scarps, and pit chains occur in the lightly cratered plains near the sub-Saturn point (near lat 08, long 08; Fig. 1). Several troughs (including Daryabar Fossa) are perpendicular to and truncated by a west-dipping, 300-m-high fault scarp, Isbanir Fossa (profile a, Fig. 7). Daryabar Fossa shows in limb profile 8 (Fig. 6) as a feature $300 m deep and 4 km wide. The parallel trough north of Daryabar Fossa (near lat 188N, long 3408) appears in limb profile 6 (Fig. 6); it is p500 m deep; this same trough has a width of nearly 5 km and a depth of p400 m near lat 228N, long 3358 in photoclinometry profile b, Fig. 7). We have not seen definitive ridges, cryovolcanic flow margins, or partly buried ‘‘ghost’’ craters in the cratered plains. (Small ghost craters would not be distinguishable from ordinary unflooded craters in available images.) The cratered plains region Cp12 has unusual craters but no obvious tectonic lineaments (see area near terminator in Fig. 2). Most large craters in Cp12 are anomalously shallow (Passey 1983), many have elliptical or irregular shapes, and many have bizarre central domes that are morphologically different from and larger than the central peaks of typical complex craters on icy satellites (Schenk 1989). The ratio of central dome diameter to crater diameter for the unusual craters in Cp12 ranges from 0.50 to 0.77, much greater than typical ratios (p0.2–0.3) for normal complex craters on icy satellites. Photoclinometric profiles were run across three large craters in Cp12 (Fig. 7). The

394

KARGEL AND POZIO

TABLE I Crater Counts of Terrains on Enceladus Cumulative crater statistics a Terrain b

Area counted (km2)

N5.66

N8.0

N11.31

Cp1 Cp2 Cp3 Cp4 Cp5

22,400 13,000 11,400 12,000 7,940

446 154 789 2083 4156

6 6 6 6 6

141 109 263 417 723

45 0 351 417 1637

6 45 6 175 6 186 6 454

0 0 0 83 6 83 379 6 218

Cp6 Cp7 Cp8 Cp9 Cp10

6,340 2,850 2,440 11,080 28,900

3785 4211 3687 5054 3218

6 6 6 6 6

773 1215 1229 675 334

2050 2105 2458 1715 1626

6 6 6 6 6

569 859 1003 393 237

473 351 409 451 692

6 6 6 6 6

Cp11 Cp12 Bt1 Bt2 Rt Rpc

53,000 21,400 9,370 15,900 12,900 109,000

1189 3505 1601 377 1783 0

6 6 6 6 6

150 405 413 154 372

340 1215 213 126 310 0

6 6 6 6 6

80 238 151 89 155

38 327 107 0 155 0

6 27 6 124 6 107

Cp1-2 Cp5-9 Bt1-2 Avg. Enceladus

35,400 30,650 25,270 339,920

339 4372 831 1341

6 6 6 6

98 378 413 63

28 1860 213 491

6 6 6 6

28 246 151 38

0 424 6 118 107 6 107 138 6 20

273 351 409 202 155

6 110

N16 0 0 0 0 0 158 351 0 181 277

N22.63 0 0 0 0 0

6 158 6 351 6 128 6 98

0 0 0 0 69 6 49

0 187 6 93 107 6 107 0 78 6 78 0

0 140 6 81 0 0 78 6 78 0

0 131 6 66 107 6 107 56 6 13

0 0 0 17.7 6 7

N32 0 0 0 0 0 0 0 0 0 0 0 47 6 47 0 0 0 0 0 0 0 2.9 6 2.9

a N values are the cumulative numbers of craters larger than or equal to the given diameter (in km) per 106 km2. Counts for craters smaller than 5.66 km are observationally incomplete due to resolution limits of the imagery. Uncertainties are tabulated at the 1s level (approximated as N 3 (Ïn/n), where n is the actual number of craters counted. The interested reader may recalculate n for each measurement by multiplying N 3 (area counted/106 km2). b Cp is the cratered plains, Bt is the banded terrain, Rt is the rifted terrain, and Rp is the ridged plains. The locations of these units and their subdivisions are shown in Fig. 5b. c Only two small craters were counted in Rp, giving it a crater density of N4 5 18 6 13.

summits of the central domes of these craters are as high as if not higher than the crater rims, a feature not normally seen in ordinary complex impact craters. In the case of Ali Baba, the central dome extends 600 m above the level of adjoining plains. Consistent with Passey’s (1983) findings, we observe that the depth to diameter ratios of these craters range from 0.010 to 0.045, shallower than normal complex craters of this size on small icy satellites. The ridged plains (Rp), 500 km across, consist of a central labyrinthine lineament complex 100 km across, a surrounding set of smooth plains (Sarandib and Diyar Planitiae), and a peripheral system of sinuous ridge belts totaling at least 750 km in length (Harran Sulci and southern Samarkand Sulci with its eastern extension). Possible southern extensions of the sinuous ridges were seen in limb profiles 9, 10, and 11 (Fig. 6) but were poorly imaged against the disk. The Rp appears to be centered on the trailing point or antapex of Enceladus’ orbital motion, where the tectonic ‘‘labyrinth’’ occurs (Fig. 4; Fig. 3, right). The nearly symmetric tectonic relation of this rectilinear grid to the ridged

plains and their bounding curvilinear ridges is intriguing but of uncertain geotectonic significance (Fig. 5a). Most lineaments in the labyrinth lack shadows or other clearly interpretable signs of relief because they are near the subsolar point in the best images (Fig. 3). Only one feature in the labyrinth is clearly expressed as a ridge when viewed against the disk of Enceladus. Limb profile 4 (Fig. 6) cuts directly across the labyrinth and shows that at least two lineaments in the airbrush map are in fact ridges, in one case over 1000 m high locally. Other lineaments crossed in the labyrinth by profiles 1–4 are too closely spaced to be certain of their character, although we suspect that most if not all of them are ridges. The labyrinth occurs on a broad regional high (profiles 4, Fig. 6); this and the linear character of the labyrinth’s ridges distinguish them from the sinuous ridges at the margins of the Rp. The major ridges of southern Samarkand Sulci and Harran Sulci characteristically have rounded or sharply crested tops (Fig. 7). Photoclinometry and limb profiling show that the major ridges are 500 to 2000 m high and have marginal troughs (Passey 1983, Dermott and Thomas 1994) (Figs.

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

395

FIG. 7. Photoclinometric topographic profiles of selected features on Enceladus. Locations of profile lines, identified by letter, are shown in Figs. 1 and 2. Profile e is a blown-up transect across the most prominent ridge shown in profile d. The single representative error bars are 2s uncertainties of the elevation differences between adjacent or nearby pixels.

6 and 7). Limb profiles also show that the ridges of Samarkand and Harran Sulci generally occur on or near the bottoms of broad regional troughs (Fig. 6), a finding that has been confirmed by careful visual inspection of Voyager

2 images FDS 43999.00 and FDS 44000.42 in the region where Samarkand Sulci crosses the limb. Despite the fact that the bases of the ridges commonly are depressed below the local ellipsoidal reference, the highest peaks in every

396

KARGEL AND POZIO

FIG. 8. Cumulative plot of crater densities of terrains on Enceladus. Statistics for craters smaller than 5.66 km in diameter show effects of observational loss due to resolution limits. Terrain symbols explained in the legend to Fig. 5.

profile rise above the reference ellipsoid and in many cases far above adjacent terrain. Samarkand Sulci has its highest peaks 1100 to 1400 m above reference (profiles 5, 6, and 7 in Fig. 8). These features clearly are mountainous ridges rather than closely spaced grooves or troughs. Nevertheless, some crests of minor ridges are only near the level of adjacent terrain and some of these relief features are very closely spaced, so the interpretation can be problematic. Lineaments named Samarkand Sulci on the U.S. Geological Survey (1992) map north of lat 338 N, long 3208 have a different character than features of the same name south of this point. Northern Samarkand Sulci consists of linear cracks or graben and appears to form a rift, whereas southern Samarkand Sulci appears to consist of sinuous ridges. The crater density of northern Samarkand Sulci, though low, is greater than that of southern Samarkand Sulci, which has virtually no craters. Sinuous ridges of southern Samarkand Sulci bend eastward near lat 338 N, long 3158 and appear to truncate the north-trending troughs of northern Samarkand Sulci. Thus, northern and southern Samarkand Sulci differ in their tectonic significane and time of origin (the rift formed before the ridges). Perhaps northern Samarkand Sulci should be renamed as fossae. The extension of sinuous ridges eastward from the bend (east of lat 338 N, long 3158) could be included with Samarkand Sulci or Harran Sulci (or given a new name under class sulci). Rifted terrain (Rt), dominated by troughs and scarps, constitutes northern Samarkand Sulci (Figs. 1–3, 5) between lat 308 and 808 N and long 3108 to 3358. Passey (1983) suggested that several large craters in terrain Rt were cleanly rifted and perhaps partly flooded. In addition to these unusual craters, terrain Rt is characterized by northtrending lineaments, apparently mostly troughs. The cra-

tered plains’ boundaries with the rifted terrain consist of north-trending, inward-sloping scarps. The western scarp is well expressed near the terminator in image FDS 44000.42. This scarp is p200 m high (by photoclinometry, profile not shown) and 4–6 km wide. The scarp has very low average slopes ,38 (locally it could be steeper), suggesting that it may be an eroded or mantled fault-line scarp of a normal fault. The banded terrain (Bt; center of disk in Fig. 3, mapped in Fig. 5) is crossed by en echelon bands of uncertain origin. The crater density is nonuniform, and the general appearance of Bt is one of an older surface unevenly mantled, perhaps by thick flows or allocthonous thrust sheets. The northern unit of terrain Bt partly envelopes an impact crater near lat 528 N, long 2858. 2.3. The Cratering Record of Enceladus Figure 8, a cumulative plot, and Fig. 9, an ‘‘R’’ plot, portray the crater densities of terrains on Enceladus and other icy satellites. The most heavily cratered terrains on Enceladus are almost as heavily cratered as the heavily cratered surfaces of Tethys and Rhea. The youngest known terrain on Enceladus, the ridged plains, is a factor of 4 less heavily cratered than Triton’s average surface. Among known terrains on icy satellites, only Europa and maybe the polar deposits on Triton (not shown in Fig. 9) are less heavily cratered than Enceladus’ ridged plains (Lucchitta and Soderblom 1982, Strom 1987, Croft et al. 1995). The ridged plains have only two small craters known in this

FIG. 9. Relative ‘‘R plot’’ of crater densities for averages of the youngest and oldest terrains on Enceladus compared with crater densities on other Saturnian satellites. Terrain symbols explained in the legend to Fig. 5. Data for Tethys and Rhea from Strom (1987); data for Triton from Croft et al. (1995). See Strom (1987) for a definition of R values. Statistics for Enceladus9 craters smaller than 5.66 km in diameter are not plotted so that observational loss is not a significant factor in this plot. Very limited data for Enceladus’ ridged plains (2 craters, both p4 km in diameter) would plot near or slightly below the curve for Triton.

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

entire area (109,000 km2); several more could be present under high solar incidence without detection, but many craters are visible in other areas under similar illumination at lower resolution. Thus, we doubt that there has been undercounting of craters in the Rp by much more than a factor of 2. The cratered plains vary in crater density by a factor of 30 (Table I) over a small longitude range (Fig. 5b); these variations may represent differences in relative ages more than a leading/trailing asymmetry in cratering rate. Crater densities increase northward from low latitudes (Cp1, Cp2, and Cp11) to high (Cp3 to Cp10). Ghost craters, such as those in lava-filled lunar basins and in Enceladus’ rifted terrain, have not been observed in the cratered plains, but small ghost craters might not be identifiable as such. Though many craters in Cp12 are heavily modified, this terrain has a crater size–frequency distribution similar to that of adjacent, normal regions of cratered plains (Fig. 8). Several large domes occur in cratered plains. Two large ones are in Cp12 in the craters Ali Baba and Aladdin. Another dome p20 3 38 km broad 3 1 km high (by photoclinometry) occurs outside of any large craters in Cp10 near lat 528 N, long 2218 (Fig. 3). The density of post-rifting (unstretched, unflooded) craters in the rifted terrain is nonuniform. Some small areas have higher densities of small craters than adjacent parts of rifted terrain. Similarly, the banded terrain has a low but variable crater density, indicating formation or modification over a substantial period. Terrain Bt1 has about the same crater density as terrains Rt and Cp4, whereas the sparse cratering of Bt2 is similar to that of Cp1-3 (Table I). 3. DISCUSSION

3.1. Enceladus’ Cratering History and Terrain Ages Crater populations. Smith et al. (1982) and Strom (1987) interpreted the cratering record of saturnian satellites to indicate two populations of craters and crater-forming projectiles, S1 and S2, which are distinguished by their crater size–frequency functions. Terrains dominated by S1 have large proportions of large craters and high crater densities; terrains dominated by S2 are relatively deficient in large craters and have low to moderate crater densities. S1 generally preceded S2. In Fig. 9, S1 is represented by the curves for the heavily cratered surfaces of Rhea and Tethys; S2 is best represented by the curve for the young terrains on Enceladus. The most heavily cratered terrains on Enceladus have crater size–frequency functions intermediate between the curves for old terrains on Tethys and Rhea and the young terrains on Enceladus. This suggests that a mixture of S1 and S2 craters may be present in Enceladus’ oldest cratered surfaces. Differences in the crater size–frequency functions of

397

Enceladus’ old and young terrains are shown in Fig. 9 by differing slopes at small diameters. The cumulative plot (Fig. 8) shows similar differences in slope between old and young terrains, thus suggesting a possible temporal shift (either abrupt or gradual) in the crater size–frequency distribution from S1 to S2. Strom (1987) observed a similar change of the crater size–frequency distribution on Miranda. Most saturnian planetocentric populations would have short time constants (p102 –104 years) for sweep-up or ejection (Stevenson et al. 1986, Farinella et al. 1990). Heliocentric objects (especially long-period comets) can have time constants greater than the age of the Solar System, and thus could produce a steady background flux of impacts. S1 could include bombardments by accretionary remnants or late-arriving pieces of a reaccreting Enceladus following its impact breakup, which may have occurred about four times (Smith et al. 1982). Contributions to S2 may have been made by short bursts of bombardment by planetocentric material such as (1) collisionally evolved remnants of S1 objects, (2) fragments of disrupted satellites of Saturn, and (3) ejecta of large impact basins on the satellites. Comets must also have made a major contribution to S2. The S2 impactors almost surely were not dominated by a single source of planetocentric objects. The variation in crater densities on Tethys, Dione, and Enceladus is considerable and mainly is geologically controlled. There is no special reason to suppose that the geologic dynamism on all these satellites would have occurred during a single short, intense blitz by planetocentric objects. Accretion could convert gravitational potential energy into thermal energy and drive a short post-accretional phase of geologic activity in the larger saturnian satellites, but this explanation does not serve tiny Enceladus (Squyres et al. 1988, Kossacki and Leliwa-Kopystynski 1993). Furthermore, terrain Rp is almost crater free; impacts of long-period comets alone would have produced more than the observed craters in 4 Ga. Hence, we conclude that the cratering record of Enceladus implies long-term or episodic resurfacing spanning billions of years. Cratering variations. The crater density of the cratered plains alone varies by a factor of 30. The observed crater density on the ridged plains at diameters .4 km is 1.8 3 1025 km22, p1/500 of the crater density of Enceladus’ most heavily cratered terrain (Cp9). Some of this difference may reflect a leading/trailing hemispheric asymmetry in cratering rate expected for tidally locked satellites (Shoemaker and Wolfe 1982). Plescia and Boyce (1983) examined their data for such an effect but found none; we have confirmed their findings. Reliable crater counts extend over a range from 0.6 to 1.9 rad from the antapex of motion; the predicted leading/trailing cratering asymmetry would amount to a factor of 2–5 difference in crater density if the surface had a uniform age. A leading/trailing asymme-

398

KARGEL AND POZIO

try could exist, but if so, it is obscured by large age differences. The crater density of the ridged plains is one-third of Triton’s average (Croft et al. 1995), which is the second lowest overall crater density among the icy satellites (Europa has the lowest). Poor statistics and poor imaging of about half of terrain Rp may have contributed to a low crater count. Allowing for statistical uncertainties, we estimate that the actual crater density of terrain Rp may be 2 to 4 times that counted; hence, the actual crater density of terrain Rp is probably (within a factor of 2) 4 3 1023 times that of the heavily cratered terrain Cp9. Absolute ages. If Tethys, Dione, and Rhea (which lack powerful nonradiogenic heating mechanisms) became inactive shortly after their formation, and if the change from S1 to S2 cratering occurred simultaneously throughout the saturnian system, then the youngest terrains on these objects and the oldest on Enceladus (which appear to include a component of S1) may date from this transition (p4 Ga?). Other terrains on Enceladus may be very much younger. As discussed above, the impact history of the saturnian satellites is unlikely to have been steady through time or monotonic in decline. Even so, the alternative assumptions of a constant or exponentially declining impact rate permits constraints to be placed on the absolute ages of terrains on Enceladus—the actual cratering history may well be approximated by one or the other assumption or something intermediate. This type of calculation is not very useful for old terrains, because order-of-magnitude uncertainties relating to the models’ assumptions may cause the calculated ages to be far younger or far older than the age of the Solar System. However, for sparsely cratered terrains the calculations, though still highly uncertain, invariably lead to ages that are far younger than the age of the solar system, a result with real geologic significance. The youngest terrain on Enceladus is the ridged plains. Smith et al. (1982) suggested that terrain Rp has an age ,109 years, thus implying that terrain Rp cannot have been resurfaced by a primordial cataclysm. This age was calculated from the rate of cratering by comets and an estimated crater density a factor of 50 less than on Dione’s smooth plains. The young age of terrain Rp places Enceladus in a class with Europa and Triton as icy satellites with ‘‘recent’’ geologic activity; tidal heating may play an important role in driving geology on all these objects. Table II summarizes three sets of model ages of terrains on Enceladus. Model 1 is based on the assumption that the cratering rate has been constant through time and equals the calculated rate of cratering at Enceladus due to comet impacts (Smith et al. 1982); it yields nonsensically high ages for most terrains, indicating that the present low cratering rate given by Smith et al. (1982) has not prevailed through time. Model 1 is nonetheless instructive, because

TABLE II Model Ages of Terrains on Enceladus Geologic unit Rp Cp2 Cp1 Bt2

Age, Model 1 (myr)a 200d 800 2,300 6,000

Age, Model 2 (myr)b 10d 140 400 400

Age, Model 3 (myr)c 200d 600 1,200 2,000

Cp3 Cp11 Bt1 Rt

4,000 18,000 11,000 16,000

700 1,100 1,400 1,600

1,700 3,000 2,500 2,900

Cp4 Cp10 Cp12 Cp8

22,000 80,000 60,000 130,000

1,900 2,900 3,200 3,300

3,200 4,500 4,200 4,900

Cp6 Cp5 Cp7 Cp9

110,000 80,000 110,000 90,000

3,400 3,800 3,800 4,600

4,700 4,500 4,800 4,600

a

Model 1 is based on the assumption that the present-day calculated rate of cratering of Enceladus by comets (Smith et al. 1982) has prevailed through geologic time. The crater densities used are N 10 values, extrapolated as necessary using the crater size-frequency function of Enceladus. This model yields untenably high ages (far greater than the age of the Solar System) for most units. Younger ages are probably upper limits. b Model 2, which also assumes a constant cratering rate through time, is forced to yield reasonable ages (younger than or about equal to the age of the Solar System) by assuming a cratering rate equal to the observed density of craters larger than 5.66 km in the average heavily cratered plains and dividing by the age of the Solar System (4600 million years). c Model 3 is based on the assumption that there has been an exponential decrease in the rate of cratering from 4600 million years ago, when the heavily cratered plains are assumed to have formed. The half-life of the cratering projectile population is 700 million years. The cratering curve is fixed by an exponential decline to the present calculated rate of cratering by comets (Smith et al. 1982). The crater densities used are N 10 values, extrapolated as necessary using the crater size–frequency function of Enceladus. d The model ages of the ridged plains have as a basis of calculation the density of 4-km craters (Table I). Due to observational limitations (solar incidence and image resolution), it is thought that only about half the craters were detected. Hence, the model ages are based on twice the crater density given in Table I, and this density was then extrapolated to 5.66 km using Enceladus’ crater size-frequency function.

it is possible that the present flux of comets has dominated cratering during the past few hundred million years and that the ages calculated for the younger terrains are roughly correct. Ages that are less than the Solar System’s age are upper limits, because Enceladus may have had other sources of cratering objects and because the cratering rate seems to have declined through time. Model 1 yields an age for terrain Rp of only 2 3 108 years (factor of 2 uncertainty). If the leading/trailing cratering asymmetry is con-

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

sidered, then terrain Rp may be a few tens of percent older. In any case, the 109-year upper age limit given by Smith et al. (1982) has been further constrained to just a few hundred million years. Terrains Cp1 and Cp2 also have Model-1 ages substantially younger than the Solar System. Model 2, like Model 1, makes the assumption that the cratering rate is constant, though realistic ages (younger than the Solar System) are forced by the assumption that the cratering rate is equal to the observed average density of craters in the heavily cratered plains divided by the age of the Solar System. Thus, Model 2 does not make assumptions regarding crater populations but only that the flux is constant. The Model 2 age for terrain Rp is 5 to 20 myr. Several other geologic units have Model 2 ages that are much younger than the Solar System. Model 3 makes the assumption that the cratering rate has been in monotonic, exponential decline. If the age of the heavily cratered plains is assumed to be the same as the age of the Solar System, and if the present cratering rate is assumed to be that due to long-period comets (Smith et al. 1982), then the half-life of the cratering projectile population is calculated to be 7 3 108 years. Several terrains have Model 3 ages much younger than the Solar System. Models 1, 2, and 3 yield widely differing ages for the younger terrains. Despite the age uncertainties implied by these models, these results indicate episodic resurfacing (at least of Rp, Cp1, and Cp2) long after the origin of the Solar System. Consideration only of Models 2 and 3 (more realistic than Model 1) expands the list of terrains that are much younger than the Solar System to Cp1, Cp2, Cp3, Cp4, Cp11, Bt1, Bt2, Rt, and Rp and implies that Enceladus has had multiple resurfacing episodes. Nothing can be said reliably of the absolute ages of the older terrains; they are probably billions of years old and appear to have large age differences. The difference between age estimates of the ridged plains (a few million to a few hundred million years) may be important for interpretations of the origin of Saturn’s E-ring and the chances that Enceladus is currently active. If terrain Rp has an average age of only a few million years, it would not be at all unlikely that regions of continuing activity exist and might account for Saturn’s E-ring by very recent explosive volcanism (Baum et al. 1981, Morfill et al. 1983, Poirier et al. 1983, Kargel 1984, Pang et al. 1984, Buratti 1988). If terrain Rp’s average age is hundreds of millions of years, then the chance that Enceladus is currently active is reduced, and one may have to call on nonvolcanic origins of the E-ring, such as spattering of impact melt, condensation of impact vapor (McKinnon 1983), or impact erosion of microsatellites (Hamilton and Burns 1994). Other geologic units are much younger than the heavily cratered plains but older than terrain Rp. These results indicate that Enceladus has had multiple episodes of geologic activity.

399

3.2. Tectonic Analogs and Volcano-Tectonic Mechanisms Crater-density variations in the cratered plains. Terrain Cp shows a pronounced northward increase in age; the statistics do not show whether the increase is gradational or stepped. Gradational variations could be consistent with a plate-spreading origin of the cratered plains (Kargel 1983); stepped variations could be due to a stochastic or systematic northward shift in volcanism. An absence of large ghost craters in the cratered plains is consistent with wholesale regeneration of lithosphere or deep volcanic flooding (though small ghost craters could exist without detection as such). The depth : diameter ratio for complex craters on Saturn’s satellites is p0.1 for craters 20 km in diameter (Schenk 1989), which indicates that volcanic deposits would have to be .2 km deep to bury completely all 20-km craters in the younger parts of the cratered plains. The preservation of small craters in Cp12 and the shallow, domed floors of large craters could indicate viscous relaxation (Passey 1983). The irregular shapes and huge central domes of these craters are inconsistent with viscous relaxation as the sole modifying process. Schenk and Moore (1995) proposed that these crater domes and similar ones on Ganymede are volcanic. High heat flow in Cp12 (Passey 1983) may have promoted volcanic–tectonic modifications to these craters. Origin of troughs and scarps. As mapped in Fig. 5, the cratered plains are dominated by features interpreted as extensional. Some fractures are fresh and others are cratered, suggesting a long-term extensional regime. Troughs on icy satellites have been interpreted as grabens and extension cracks (Parmentier et al. 1982, Squyres et al. 1983). Extension cracks have been suggested because normal faulting usually occurs when the tensile strength of the fractured medium is exceeded and lateral confining stress exceeds tensile strength by several times. The widths of Enceladus’ troughs are 2–4 km, so that confining stresses at the bottoms of normal faults would be a few bars, two orders of magnitude less than the tensile strength of cold, intact ice. Ice tensionally stressed with little lateral confinement was thought to produce open fissures rather than normal faults or grabens (Squyres et al. 1983). Subsequent images of Miranda, Ariel, and Triton revealed many normal faults, and our interpretation suggests that normal faults occur on Enceladus, too. The structural mechanical arguments cited by Squyres et al. (1983) can be reconciled with normal faulting because the effective tensile strength of prefractured ice, such as in the regolith of an icy satellite, is much less than the tensile strength of unfractured ice. The rifted terrain near Samarkand Sulci is characterized by sub-parallel and quasiperiodic extensional features (normal faults, grabens, and/or extension cracks), possible cryovolcanic flood deposits, and stretched and rifted cra-

400

KARGEL AND POZIO

ters. Terrain Rt may have dropped down along inwardsloping scarps, probably expressions of normal faults, which appear to be heavily eroded or mantled. Cratered plains adjoin terrain Rt to the east and west along scarps. The large modified craters in terrain Rt suggest that heavily cratered plains formerly extended across here. The rift may have replaced cratered plains by (1) modest tectonic extension, down-dropping, and volcanic flooding, or (2) complete rifting and generation of new lithosphere. Herrick and Stevenson (1990) suggest that periodic grooves or ridges (such as in the rifted terrain and ridge belts) could form on Enceladus by necking instabilities given high enough heat flow. They suggested a model in which necking may occur when intrusion of warm material occurs at shallow levels (one to a few kilometers) along with a brief episode of extension (lasting p105 years) or compression (lasting p103 years). Cratering evidence would require either a more extended period or multiple episodes of tectonic development. A west-dipping scarp, Isbanir Fossa, truncates and appears to translate several orthogonal troughs (Fig. 4; bottom of Fig. 1). Isbanir Fossa may be a strike–slip fault along which two pairs of troughs (Daryabar Fossa and unnamed troughs parallel to and north of Daryabar Fossa) have p15 km right-lateral and p300 m vertical displacement (Kargel 1983). The unnamed troughs involve an eastern branch that curves, so the strike–slip interpretation is not positive, though the sense of curvature (concave north) is consistent with deformation that could occur within a right-lateral strike–slip fault zone. This rectilinear fracture system also includes a pit chain parallel to Isbanir Fossa (near lat 258–458 N, along 88; Figs. 1, 2, and 5a); Kargel (1984) speculated that these pits may have formed by tectonically controlled explosive cryovolcanism. Some possible tectonic analogs of Enceladus’ sinuous ridges on other icy satellites include trough-bounded ridges on Ariel near lat 508 S, long 3008, the ridged plains of Elsinore Corona on Miranda, and cratered ridged plains on Dione near lat 458 S, long 558. Ariel’s and Miranda’s ridges were attributed to (1) viscous cryovolcanic effusions of possible NH3 –H2O-rich liquids or slurries) by Croft and Soderblom (1991), Kargel et al. (1991), and Schenk (1991), or (2) possible solid-state effusions of warm ice (Jankowski and Squyres 1988). Either interpretation could be applied to Enceladus’ ridges; a third possible explanation for Enceladus involves folding. An extensional tectonic analog between Enceladus’ ridges and Ganymede’s grooves was proposed by Squyres et al. (1983) on what we believe was their mistaken belief that the sinuous ridges of Enceladus are negative-relief grooves. A better analog for Ganymede’s grooved terrain would be Enceladus’ rifted terrain. Ganymede’s grooved terrain is believed to have developed as rifts with cycles of extensional tectonism (groove formation) and cryovol-

canism (e.g., Parmentier et al. 1982). The apparent presence of flooded, rifted craters in Enceladus’ rifted terrain (Passey 1983) further supports the Ganymede model for the rifted terrain but not Enceladus’ ridged plains. The tectonic features on the 43% of Enceladus that was mapped suggests a global tectonic pattern. According to our interpretation, extensional features and cratered plains dominate the anti-Saturn region, the north polar region, and the Saturn-facing region; compressional features and the youngest plains on Enceladus dominate the region near the orbital trailing point. Compressional deformation formed fold-like belts of sinuous ridges p608 from the trailing point. The prograde region was poorly imaged, but its low crater density may be consistent with global geologic–tectonic symmetry. Variations in crater densities complicate the tectonic interpretation; the tectonic development of Enceladus may have occurred continuously or episodically over a long period. For instance, the sinuous and rectilinear ridges in terrain Rp must themselves be extremely young, whereas some cratered fractures in terrain Cp must be much older. The global tectonic pattern may reflect (1) a constant, purely spatial stress pattern, (2) a moving locus and evolving sign (e.g., extension to compression) of tectonic stress and deformation, or (3) temporal and spatial stress variations. In accordance with Passey (1983), we suspect that the last possibility is the most probable. Dione and Europa also have global tectonic patterns, but incomplete and nonuniform imaging makes intersatellite comparisons problematic. Europa’s and Dione’s tectonic patterns (and by analogy, maybe Enceladus’) may have been caused by tidal stresses or despinning (Lucchitta et al. 1982, Moore 1984), though none of the patterns match idealized predictions (Melosh 1977, 1980). Heating mechanisms. Radiogenic heating should be less in Enceladus than in other icy satellites of comparable size because of its lower density (hence, its lower content of K-, U-, and Th-bearing rock). Tidal dissipation in a layered Enceladus probably drives Enceladus’ activity (Poirier et al. 1983, Squyres et al. 1983, Ross and Schubert 1989). Enceladus’ tidal heat flow may be as great as 5 mW m22 (Ross and Schubert 1989). For a crust with the thermal conductivity of water ice, this heat flow would drive a thermal gradient of p0.9 K km21, enough to allow NH3 –H2O melting p100 km below the surface. For a crust of ammonia hydrate, which is less thermally conductive than water ice (Ross and Kargel 1995), the thermal gradient could be as much as 4 K km21, enough to allow melting 25 km beneath the surface. 3.3. Enceladus’ Ridges Interpreted as Folds Extrusion of ridges on Ariel, Miranda, and Triton has been proposed (Croft et al. 1995) and could be applied to

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

FIG. 10. Critical compressive stresses required to buckle layered strata on Earth and layered ices on Enceladus.

Enceladus’ ridges. Squyres et al. (1983) favored an extensional tectonic origin. We favor a model involving folding because of the ridges’ quasisinusoidal plan and cross profiles. Folding may have been a response to (1) volcanism, lithospheric loading, and downwarping analogous to the origin of lunar wrinkle ridges (Solomon and Head 1980), or (2) lithospheric compression along zones of mantle downwelling. The second possibility is supported by differing forms and vertical scales of the two types of ridge folds. Curvilinear ridge belts on Enceladus are similar in many respects to fold belts on Earth and Venus (Gwinn 1970, King 1970, Hatcher et al. 1989, Smrekar and Solomon 1992, Solomon et al. 1992). The length, width, and wavelength (in plan view) of sinuous ridge belts on Enceladus are about 1/6 of the equivalent values for the Appalachians. Enceladus’ ridges have cross-sectional wavelengths averaging slightly less than those of major Appalachian folds, and the ridge amplitudes are about 1/3 as great on Enceladus as in the pre-erosional Appalachians. Folding on Earth affects thinly bedded sedimentary and volcanic rocks, strongly foliated metamorphic rocks, and debris-layered glacier ice. Unbedded rocks rarely fold but respond to compression by thrusting or ductile shortening. Folding involves buckling, shearing, and recrystallization. We have applied the solution of Turcotte and Schubert (1982) to the initiation of buckling on Enceladus. The critical deviatoric compressive stress, s c , required to initiate buckling is

s c 5 h[Ehrg]/[3(1 2 n 2)]j1/2,

(1)

where E is Young’s modulus (p1011 dynes cm22 for cold ice), h is layer thickness, r is density (0.95 g cm23 for cold ice), g is gravity (8 cm sec22), and n is Poisson’s ratio (p0.3 for cold ice). Enceladus’ low gravity and the low density of ice tend to make folding easier than for rock on Earth (Fig. 10). Massive or thickly bedded rocks normally do not

401

fold under the range of horizontal compressive deviatoric stresses in convergent plate boundaries on Earth. (Deviatoric compressional stresses in Earth’s crust are limited to ,100 bars by brittle failure.) Shale is sufficiently elastic and thinly layered that the threshold buckling stress commonly is exceeded. Buckling may become permanent and grow by strain recrystallization and grain-boundary or bedding-plane shear. Stratified ice on Enceladus may fold if compressed (Fig. 10). Compressive stresses in Enceladus’ crust are unknown. Thermal convection in small icy satellites may produce deviatoric stresses of p0.01 bar to 0.1 bar (Ellsworth and Schubert 1983, Squyres and Croft 1986); shell deformation due to despinning may induce stresses p0.05 bar in small icy satellites (Squyres and Croft 1986); and sinker tectonics may produce horizontal compressive surface stresses from 1 to 10 bars (Janes and Melosh 1988). Thus, compressive stresses in Enceladus probably are smaller than in Earth, but folding would be possible if Enceladus’ crust is layered by interbedding of volcanic flows with horizons of E-ring particulates, impact breccia, or vesicular material. 3.4. Possible Composition of Enceladus and Its Lavas Strong geological, geophysical, and cosmochemical reasons make us doubt that Enceladus and its flows are composed of pure H2O. But Enceladus is the Solar System’s most reflective large object, and only pure water ice has been observed spectroscopically there (Clark et al. 1983). Its low mean density (1.00–1.12 g cm23 (Dermott and Thomas 1994)) and pure-ice surface may indicate a unique composition. Pure ice is less dense than Enceladus, but pure liquid water overlain by a thin ice shell would satisfy constraints imposed by density and surface composition. Nebular condensation and accretion could produce objects of pure H2O only with special fractionations. More likely, Enceladus contains a small allotment of rocky minerals with a larger-than-usual allotment of water ice and other ices (Lewis 1972, Fegley and Prinn 1989). Kargel (1991, 1992) reviewed reasons for considering H2O–NH3 , H2O–NH3 –CH3OH, and H2O–salt solutions as possible cryovolcanic lavas. These liquids have viscosities ranging from p0.1 poises for eutectic MgSO4-rich brines, 1 poise for H2O–CH3OH, 40 poises for H2O–NH3 , to 4 3 104 poises for H2O–NH3 –CH3OH (Kargel et al. 1991). All but the last liquid have viscosities low enough to form volcanic plains of low relief. The ternary H2O–NH3 –CH3OH liquid, partly crystallized H2O–NH3 , or nearly solid masses of extruded water ice could form viscous dome-building extrusions. A liquid-water mantle with a thin ice crust would have virtually no ability to convect, expand, contract, or impart significant shear stresses to the crust, thus making it difficult to account for the tectonically riven surface. Melting or

402

KARGEL AND POZIO

freezing of pure H2O (29% volume change on melting) could produce intense shell deformation, but it might result in a nearly isotropic stress field unlike that inferred from observations of Enceladus. The H2O–NH3 peritectic melting reaction causes a 11% change in volume (Croft et al. 1988]), so that melting or freezing of ammonia–water would be even less capable of explaining Enceladus’ tectonics. Salt-water eutectic brines generally freeze to mixtures of water ice and hydrated salts; volume changes during melting or freezing are commonly of the same sign but half the magnitude that of pure water. Thus, we doubt that volume changes attending melting and refreezing is the sole explanation for Enceladus’ complex tectonic pattern. Extrusion of pure liquid water onto the surface of a less dense crust of pure ice would be difficult but not impossible. More likely, volcanism occurred because partial melts were buoyant relative to crust and mantle. H2O–NH3 peritectic (32.6% NH3) and eutectic (35.4% NH3) liquids are less dense than their frozen ices (Croft et al. 1988) and are less dense than Enceladus; hence, H2O–NH3 volcanism is more likely than H2O volcanism. Enceladus’ potential for extrusion of high-density eutectic salt-water brines (denser than Enceladus) is reduced on the same basis. H2O– CH3OH or H2O–NH3 –CH3OH would be plentiful in a satellite containing comet-like volatiles; density relations favor extrusion of these liquids (Hoban et al. 1991, 1993, Kargel 1992). Saturn’s E-ring (Baum et al. 1981, Stevenson 1982, Morfill et al. 1983, Pang et al. 1984, Showalter et al. 1991) and a crater chain on Enceladus (Kargel 1984) (Figs. 2 and 5a) may have formed by explosive volcanism. (Note alternative models for the origin of the E-ring; McKinnon 1983, Hamilton and Burns 1994). Phreatomagmatic explosions caused by interactions of liquid H2O–NH3 with liquid H2O, or of liquid CH4 –CO–N2 with H2O or H2O–NH3 , could produce both features, although phreatomagmatic interactions are not needed for explosive volcanism. Kargel and Strom (1990) modeled exsolution of CH4 from saturated highpressure H2O–NH3 on Triton and found that it could cause explosive volcanism; the same process could work on Enceladus. The improbability of pure H2O magmatism is highlighted by the chemical complexity of condensate assemblages in the Solar System and the high chemical activity of water. So why is the visible surface of Enceladus apparently made of nothing but ice? Flotation of water ice in flows or plutons, sinking of dense phases such as salts and ammonia dihydrate, diapiric solid-state ascent of segregated ice bodies, surface coatings by icy E-ring particles, or spectroscopic masking of impurities by water ice may have contributed to make a true ice-dominated surface or to make it appear that way (Croft et al. 1988, Buratti 1988, Hogenboom et al. 1994, 1995]. Despite these caveats, the seemingly pureice surface of Enceladus is a rare, hard constraint on com-

position and the possible chemical nature of volcanism there. If additional concerted efforts to detect even traces of impurities fail, we may be forced to review arguments and assumptions regarding the improbability of water volcanism and pure-water satellites, especially in light of a recent downward revision in the mean density of Enceladus nearly to that of pure water (Dermott and Thomas 1994). Following a quarter century of speculation that H2O– NH3 volcanism may be important on small icy satellites (Lewis 1971, Consolmagno and Lewis 1978, Stevenson 1982, Squyres et al. 1983, Croft et al. 1988, Kargel et al. 1991, Schenk 1991, Kargel 1992), we also conclude that H2O–NH3 , with or without other substances, is a likely resurfacing agent on Enceladus. 4. CONCLUSIONS AND GEOCHRONOLOGY OF ENCELADUS

The geologic history of Enceladus may be divided into four partly overlapping stages. (1) Early global resurfacing late during population S1 cratering removed most S1 craters. The last stage of S1 again cratered Enceladus, but not to saturation. The oldest surviving remnants of the ancient surface, overprinted by S2 craters, is at high northern latitudes in Cp5-9. Model ages of these ancient terrains (Models 2 and 3 in Table II) have no reliability because the ages are based partly on the assumption that these terrains date from close to the origin of the Solar System. (2) Rifting resulted in fracturing, downdropping, cryovolcanic flooding, and viscous relaxation and stretching of impact craters in the rifted terrain. This event was roughly coeval with plains formation in Cp4 and emplacement of Bt1. Models 2 and 3 (Table II) suggest absolute ages dating from the middle of Enceladus’ geologic history p1400 to 2900 myr. (3) The youngest subunits of cratered plains (Cp1-3) were formed by continued cryovolcanism and extensional fracturing. The younger subunit of banded terrain (Bt2) formed at about the same time. Models 2 and 3 (Table II) suggest a protracted or episodic development of these terrains from 140 to 2900 myr. (4) Ridged plains were formed by compressional tectonics and widespread cryovolcanism involving a lava of relatively low viscosity. Models 2 and 3 (Table II) indicate an average age between 10 and 200 myr. Continuing explosive volcanism may be renewing Saturn’s E-ring. In sum, Enceladus has a complex history with multiple episodes of activity spread over billions of years. Longterm or episodic geologic activity has occurred and requires high levels of tidal heating. Current steady-state tidal dissipation may cause melting of ammonia hydrate at a depth

THE VOLCANIC AND TECTONIC HISTORY OF ENCELADUS

of just 25 km if the crust is made of ammonia hydrate or p100 km if it is made of water ice. ACKNOWLEDGMENTS We thank Guy Consolmagno, Doris Weir, Derek Hirsch, and three anonymous reviewers for scientific and editorial reviews. We are indebted to Peter Thomas, who generously made available a digital database of limb residuals (Dermott and Thomas 1994) and engaged us in fruitful discussions concerning limb profile data. J.S.K. acknowledges logistical support from CNR (Italy), which made the early stages of this work possible.

REFERENCES BAUM, W. A., T. KREIDL, J. A. WESTPHAL, G. E. DANIELSON, P. K. SEIDELMANN, D. PASCU, AND D. G. CURRIE 1981. Saturn’s E ring. I. CCD observations of March 1980. Icarus 47, 84–96. BURATTI, B. 1988. Enceladus: Implications of its unusual photometric properties. Icarus 75, 113–126. CLARK, R. N., R. H. BROWN, M. L. NELSON, AND J. N. HAYASHI 1983. Surface Composition of Enceladus, Abstract 853, 15th Annual DPS Meeting. CONSOLMAGNO, G. J., AND J. S. LEWIS 1978. The evolution of icy satellite interiors and surfaces. Icarus 34, 280–293. CROFT, S. K., AND L. A. SODERBLOM 1991. Geology of the uranian satellites. In Uranus (J. T. Bergstralh, E. D. Miner, and M. S. Matthews, Eds.), pp. 561–628. Univ. of Arizona Press, Tucson. CROFT, S. K., J. I. LUNINE, AND J. S. KARGEL 1988. Equation of state of ammonia–water liquid: Derivation and planetological application. Icarus 73, 279–293. CROFT, S. K., J. S. KARGEL, R. L. KIRK, J. M. MOORE, P. M. SCHENK, AND R. G. STROM 1995. The geology of Triton. In Neptune and Triton (D. P. Cruikshank, Ed.), Univ. of Arizona Press, Tucson. DERMOTT, S. F., AND P. C. THOMAS 1994. The determination of the mass and mean density of Enceladus from its observed shape. Icarus 109, 241–257. ELLSWORTH, K., AND G. SCHUBERT 1983. Saturn’s icy satellites: Thermal and structural models. Icarus 54, 490–510. FARINELLA, P., P. PAOLICCHI, R. G. STROM, J. S. KARGEL, AND V. ZAPPALA` 1990. The fate of Hyperion’s fragments. Icarus 83, 186–204.

403

tor of the 3.52-em emission feature in several comets. Icarus 93, 122–134. HOBAN, S., D. C. REUTER, M. A. DISANTI, M. J. MUMMA, AND R. ELSTON 1993. Infrared observations of methanol in comet P/Swift-Tittle. Icarus 105, 548–556. HOGENBOOM, D. L., J. S. KARGEL, T. C. HOLDEN, AND J. GANASAN 1994. The ammonia–water phase diagram and phase volumes to 4 kbars. Proc. Lunar Planet Sci. 25th, 555–556. HOGENBOOM, D. L., J. S. KARGEL, J. P. GANASAN, AND L. LEE 1995. Magnesium sulfate–water to 400 MPa using a novel piezometer: Densities, phase equilibria, and planetological implications. Icarus 115, 258–277. JANES, D. M., AND H. J. MELOSH 1988. Sinker tectonics: An approach to the surface of Miranda. J. Geophys. Res. 93, 3127–3143. JANKOWSKI, D. G., AND S. W. SQUYRES 1988. Solid-state volcanism on the satellites of Uranus. Science 241, 1322. KARGEL, J. S. 1983. Enceladus: An analog of terrestrial plate tectonism? Proc. Lunar Planet. Sci. 14th, 363–364. KARGEL, J. S. 1984. A crater chain on Enceladus: Evidence for explosive water volcanism. Proc. Lunar Planet. Sci. 25th, 427–428. KARGEL, J. S. 1991. Brine volcanism and the interior structures of asteroids and icy satellites. Icarus 94, 368–390. KARGEL, J. S. 1992. Ammonia–water volcanism on icy satellites: Phase relations at 1 atmosphere. Icarus 100, 556–574. KARGEL, J. S., AND R. G. STROM 1990. Cryovolcanism on Triton (abstract), Proc. Lunar Planet. Sci. 21st, 599–600. KARGEL, J. S., S. K. CROFT, J. I. LUNINE, AND J. S. LEWIS 1991. Rheological properties of ammonia–water liquids and crystal-liquid slurries: Planetological applications. Icarus 89, 93–112. KING, P. B. 1970. Epilogue. In Studies of Appalachian Geology: Central and Southern (G. W. Fisher, F. J. Pettijohn, J. C. Reed Jr., and K. N. Weaver, Eds.), pp. 437–439. Interscience, New York. KOSSACKI, K. J., AND J. LELIWA-KOPYSTYNSKI 1993. Medium-sized icy satellites: Thermal and structural evolution during accretion. Planet. Space Sci. 41, 729–741. LEWIS, J. S. 1971. Satellites of the outer planets: Their chemical and physical nature. Icarus 16, 241–252. LEWIS, J. S. 1972. Low temperature condensation from the solar nebula. Icarus 16, 241–252. LUCCHITTA, B. K., AND L. A. SODERBLOM 1982. The geology of Europa. In Satellites of Jupiter (D. Morrison, Ed.), pp. 521–555. Univ. of Arizona Press, Tucson.

FEGLEY, B., JR., AND R. G. PRINN 1989. Solar nebula chemistry: Implications for volatiles in the solar system. In The Formation and Evolution of Planetary Systems, (H. Weaver, F. Paresce, and L. Danly, Eds.), pp. 171–211. Cambridge Univ. Press, Cambridge.

LUCCHITTA, B. K., L. A. SODERBLOM, AND H. M. FERGUSON 1982. Structures on Europa. Proc. Lunar Planet. Sci. Conf. 12th, 1555–1567.

GWINN, V. E. 1970. Kinematic patterns and estimates of lateral shortening, valley and ridge and great valley provinces, central Appalachians, south-central Pennsylvania. In Studies of Appalachian Geology: Central and Southern, (G. W. Fisher, F. J. Pettijohn, J. C. Reed Jr., and K. N. Weaver, Eds.), pp. 127–146. Interscience, New York.

MELOSH, H. J. 1977. Global tectonics of a despun planet. Icarus 31, 221–243.

HAMILTON, D. P., AND J. A. BURNS 1994. Origin of Saturn’s E ring: Selfsustained, naturally. Science 264, 550–553. HATCHER, R. D., W. A. THOMAS, AND G. W. VIELE, Eds. 1989. The Appalachian–Ouachita Orogen in the United States, The Geology of North America, Vol. F-2. Geol. Soc. Am., Boulder, CO.

MCKINNON, W. B. 1983. Origin of the E-ring: Condensation of impact vapor. . . or boiling of impact melt? Proc. Lunar Planet. Sci. 14th, 487–488.

MELOSH, H. J. 1980. Tectonic patterns on a tidally distorted planet. Icarus 43, 334–337. MOORE, J. M. 1984. The tectonic and volcanic history of Dione. Icarus 59, 205–220. MORFILL, G. E., E. GRU¨ N, AND T. V. JOHNSON 1983. Saturn’s E, G, and F rings: Modulated by the plasma sheet? J. Geophys. Res. 88, 5573–5579.

HERRICK, D. L., AND D. J. STEVENSON 1990. Extensional and compressional instabilities in icy satellite lithospheres. Icarus 85, 191–204.

PANG, K. D., C. C. VOGE, J. W. RHOADS, AND J. M. AJELLO 1984. The E ring of Saturn and satellite Enceladus. J. Geophys. Res. 89, 9459–9470.

HOBAN, S., M. MUMMA, D. C. REUTER, M. DISANTI, R. R. JOYCE, AND A. STORRS 1991. A tentative identification of methanol and the progeni-

PARMENTIER, E. M., S. W. SQUYRES, J. W. HEAD, AND M. L. ALLISON 1982. The tectonics of Ganymede. Nature 295, 290–293.

404

KARGEL AND POZIO

PASSEY, Q. R. 1983. Viscosity of the lithosphere of Enceladus. Icarus 53, 105–120. PEALE, S. J., P. CASSEN, AND R. T. REYNOLDS 1980. Tidal dissipation, orbital evolution, and the nature of Saturn’s inner satellites. Icarus 43, 65–72. PLESCIA, J. B., AND J. M. BOYCE 1983. Crater numbers and geological histories of Iapetus, Enceladus, Tethys, and Hyperion. Nature 301, 666–670. POIRIER, J. P., L. BOLOH, AND P. CHAMBON 1983. Tidal dissipation in small viscoelastic ice moons: The case of Enceladus. Icarus 55, 218–230. POZIO, S., AND J. S. KARGEL 1989. The tectonic and igneous evolution of Enceladus. Proc. Lunar Planet. Sci. 20th, 864–865. POZIO, S., AND J. S. KARGEL 1990. The cratering record and geological history of Enceladus. Proc. Lunar Planet. Sci. 21st, 975–976. ROSS, M. N., AND G. SCHUBERT 1989. Viscoelastic models of tidal heating in Enceladus. Icarus 78, 90–101. ROSS, R. G., AND J. S. KARGEL 1996. Thermal conductivity of solar system ices, with special reference to martian polar caps. In Solar System Ices, (C. de Bergh, B. Schmitt, and M. Festou, Eds.), Kluwer Academic, in press. SCHENK, P. M. 1989. Crater formation and modification on the icy satellites of Saturn and Uranus: Depth/diameter and central peak occurrence. J. Geophys. Res. 94, 3813–3832. SCHENK, P. M. 1991. Fluid volcanism on Miranda and Ariel: Flow morphology and composition. J. Geophys. Res. 96, 1887–1906. SCHENK, P. M., AND J. M. MOORE 1995. Volcanic constructs on Ganymede. J. Geophys. Res. submitted. SHOEMAKER, E. M., AND R. F. WOLFE 1982. Cratering time scales for the Galilean satellites. In Satellites of Jupiter (D. Morrison, Ed.), pp. 277–339. Univ. of Arizona Press, Tucson.

SMREKAR, S. E., AND S. C. SOLOMON 1992. Gravitational spreading of high terrain in Ishtar Terra. Venus, J. Geophys. Res. 97, 16,121– 16,148. SOLOMON, S. C., AND J. W. HEAD 1980. Lunar mascon basins: Lava flooding, tectonics, and evolution of the lithosphere. Rev. Geophys. Space Phys. 18, 107–141. SOLOMON, S. C., et al. 1992. Venus tectonics: An overview of Magellan observations. J. Geophys. Res. 97, 13,199–13,256. SQUYRES, S. W., AND S. K. CROFT 1986. The tectonics of icy satellites. In Satellites (J. A. Burns and M. S. Matthews, Eds.), pp. 293–341. Univ. of Arizona Press, Tucson. SQUYRES, S., R. T. REYNOLDS, P. M. CASSEN, AND S. J. PEALE, 1983. The evolution of Enceladus. Icarus 53, 319–331. SQUYRES, S. W., R. T. REYNOLDS, A. L. SUMMERS, AND F. SHUNG 1988. Accretional heating of the satellites of Saturn and Uranus. J. Geophys. Res. 93, 8779–8794. STEVENSON, D. J. 1982. Volcanism and igneous processes in small icy satellites. Nature 298, 142–144. STEVENSON, D. J., A. W. HARRIS, AND J. I. LUNINE 1986. Origins of satellites. In Satellites (J. A. Burns and M. S. Matthews, Eds.), pp. 39–88. Univ. of Arizona Press, Tucson. STROM, R. G., 1987. Mercury, In The Geology of the Terrestrial Planets, NASA SP-469, 13–55. STROM, R. G. 1987. The solar system cratering record: Voyager 2 results at Uranus and implications for the origin of impacting objects. Icarus 70, 517–535. TURCOTTE, D. L., AND G. SCHUBERT 1982. Geodynamics: Applications of Continuum Physics to Geological Problems. Wiley, New York.

SHOWALTER, M. R., J. N. CUZZI, AND S. M. LARSON 1991. Structure and properties of Saturn’s E ring. Icarus 94, 451–473.

U.S. GEOLOGICAL SURVEY 1992. Pictorial map and controlled photomosaic of Enceladus. U.S. Geol. Surv. Misc. Invest. Ser. Map I-2156, scale 1:2,000,000, 2 sheets.

SMITH, B., et al. 1982. A new look at the Saturn system: The Voyager 2 images. Science 215, 504–537.

VERBISCER, A. J., AND J. VEVERKA 1994. A photometric study of Enceladus. Icarus 110, 155–164.