Morphologic characteristics, flow stratigraphy, and radar backscatter properties of five lava flow fields on Venus (Turgmam Fluctus, Zipaltonal Fluctus, Tuli Mons/Uilata Fluctus, Var Mons, and Mylitta Fluctus) were examined to understand flow field emplacement mechanisms and relationships to other surface processes. These analyses indicate that the flow fields studied developed through emplacement of numerous, thin flow units, presumably over extended periods of time. Although the Venusian fields display flow morphologies similar to those observed within terrestrial flow fields, the Venusian flow units are significantly larger and have a larger range of radar backscatter coefficients. Both simple and compound flow emplacement appear to have occurred within the flow fields. A potential correlation between flow rheology and radar brightness is suggested by differences in planform morphology, apparent flow thickness, and apparent sensitivity to topography between bright and dark flows. Distributary flow morphologies may result from tube-fed flows, and postemplacement modification by processes such as flow inflation and crustal foundering is consistent with discrete zones of increased radar brightness within individual flow lobes. Mapping of these flow fields does not indicate any simple evolutionary trend in eruptive/resurfacing style within the flow fields, or any consistent temporal sequence relative to other tectonic and volcanic features.
 Previous investigations of lava flows have examined morphologies diagnostic of emplacement style and processes, constraining flow field development using terrestrial examples. Walker  distinguishes lava flows based on whether they are divisible into multiple flow units (compound) or are not (simple). These terms were originally applied to small-scale features observed primarily in cross-section; pahoehoe flows are considered to be almost always compound and a'a flows are commonly compound, whereas higher viscosity lavas (andesite and dacite) are more typically simple [Walker, 1972]. For a given lava composition, extrusion rate is suggested to be the primary control determining which type of flow is produced, with higher rates favoring simple flow emplacement. Guest et al.  describe constraints on flow morphology that are due to limited supply and to cooling. Volume-limited flows advance until lava supply from the local source ceases. Cooling-limited flows advance until cooling at the flow front is sufficient to prohibit further movement. This may produce overflows and breakouts from the original lobe, which change its gross planform morphology. Planform morphometric properties (average thickness and ratio of maximum width to maximum length) have been related to eruption duration and underlying slope, which is suggested to be independent of flow field size [Kilburn and Lopes, 1988; Lopes and Kilburn, 1990]. Due to the complexity of subsurface transport through multiple paths in compound pahoehoe flow fields, analyses of volume- and cooling-limited flows have been primarily limited to surface-fed lava flows [e.g., Kilburn and Lopes, 1988; Pinkerton and Wilson, 1994].
 Morphologic analyses of Venusian lava flows rely largely on synthetic-aperture radar (SAR) data collected during the 1990–1994 Magellan mission. Three radar-mapping cycles resulted in the collection of single-wavelength (S-band, 12.6 cm), single-polarization (HH, horizontal transmit-horizontal receive) data at ∼100 m/pixel for over 98% of Venus' surface. That data has been resampled to 75 m/pixel in the full-resolution radar mosaic data sets (FMAPs).
 A previous survey of large (>50,000 km2) lava flow fields, termed “great flow fields,” on Venus focused on overall flow field morphologies, radar textures, and eruptive settings [Lancaster et al., 1995]. Lancaster et al.  developed a morphologic classification of great flow fields based on 50 of the 208 large flow fields identified by Magee [1994; see also Magee and Head, 2001]. Flow fields are described as sheet, transitional, or digitate; digitate flow fields are further classified as apron, fan, or subparallel. The morphologies are attributed to differences in emplacement style, source characteristics, and local topography. The sheet flow field morphology (e.g., from Lauma Dorsa) lacks discrete sub-units and is interpreted to indicate ponded, volume-limited flow emplacement. Transitional flow fields (e.g., Neago Fluctus) are composed of multiple large sheets, creating a more digitate planform and greater length to width ratios than sheet flow fields.
 Digitate flow fields (e.g., the lava fan on Derceto Plateau, the southeast flank of Ozza Mons, and the flow fields examined herein) are the most common morphologic type on Venus. They are composed of multiple distinct flow units that are attributed to cooling-limited flow emplacement. The apron and fan types of digitate flow fields were found to have been extruded from centralized sources (such as coronae or clusters of shields) and are distinguished by whether the flow field completely surrounds the source vent (apron) or not (fan), reflecting the local topography. Subparallel flow fields were found to have been emplaced from less centralized sources (such as portions of rifts). It should be noted that sheet and digitate morphologies represent, respectively, simple and compound flows, although the scale of features for which these terms were defined is drastically different. This is the case because the ability to distinguish flow units within a flow field depends on the type and resolution of data available for flow mapping. Terrestrial studies of flow fields using airborne and spaceborne radar remote sensing data suggest that the ability to discriminate individual lava flow units within Venusian flow fields may not be possible for the smallest scales at which they occur, given the limitations of Magellan SAR resolution and lack of multiple wavelength and polarization data [e.g., Greeley and Martel, 1988; Gaddis et al., 1989; Plaut, 1991]. These studies indicate that features such as a'a flows and large-scale topographic elements may be distinguished using radar data; low-backscatter features, such as pahoehoe flows and cinder cones, were not reliably identified.
 The classification scheme of Lancaster et al.  is modified in a recent study of the three large Venusian volcanoes Sif, Gula, and Kunapipi Montes [Stofan et al., 2001]. Stofan et al.  divide flow fields into digitate, fan, and sheet types, which are further broken into the following subtypes based on appearance in the Magellan SAR data: dark, bright-intermediate, dark-bright edges, mosaic, hummocky, or complex. The variations in flow planform morphology and surface texture, distribution of these flow field types, and shapes of the volcanoes were examined to constrain the development of each edifice. Planform morphology was related to variations in eruption style, with digitate flows resulting from high effusion rate, short duration eruptions relative to sheet flows. Surface textures evident in the radar data were interpreted to reflect local emplacement processes, such as fracturing of the surface crust during flow. This analysis indicated that the volcanoes developed through emplacement of frequent, short duration, low effusion rate summit eruptions and infrequent, large volume flank eruptions. Previous investigations had suggested that lava flows from a volcanic center display a temporal progression from long, broad sheets to shorter, digitate flows over the duration of activity [Keddie and Head, 1994, 1995]. Volcanic histories determined by Stofan et al. , however, indicate that large sheets and smaller digitate flows are both emplaced throughout the duration of edifice growth, consistent with terrestrial volcanoes.
 Previous detailed mapping of Venusian flow fields has been limited. One such study mapped flow fields in Kawelu Planitia around Sekmet Mons, with intraflow field mapping conducted for Strenia Fluctus [Zimbelman, 2000]. From that study, it was concluded that flows within Strenia Fluctus emanate from multiple sources, are interfingered, and that at least some flows are compound. No obvious pattern in flow field stratigraphy was observed.
 For the current study, lava flow fields were chosen that display numerous discrete lobes and exhibit a significant range of radar brightness, giving preference to those imaged at similar incidence angles to terrestrial airborne radar (∼30–45°). Five examples (Figure 1) were selected that are covered well in the Magellan cycle-1 (left-looking) FMAPs, include rift and centralized vents, and represent a diversity of geologic and tectonic settings in various regions across Venus: Turgmam Fluctus, Zipaltonal Fluctus, the Tuli Mons/Uilata Fluctus flow complex, the Var Mons volcanic center, and Mylitta Fluctus (Figure 2). Based on stratigraphic relationships, these flow fields represent some of the most recent resurfacing (locally to regionally) within the extensive lowland plains [e.g., Crown et al., 1994; Zimbelman, 2000; Rosenberg and McGill, 2001; D. A. Crown et al., Geologic map of the Guinevere Planitia Quadrangle of Venus, map in preparation, 2002 (hereinafter referred to as Crown et al., map in preparation, 2002)], providing important constraints on the development of volcanic surface units.
 Flow field base images were produced using the Integrated Software for Imagers and Spectrometers (ISIS, available from the U.S. Geol. Surv., Flagstaff, AZ) to reproject and mosaic Magellan cycle-1 FMAP framelets. Calibration of the radar data follows Ford and Plaut  and Campbell . To apply the calibration for incidence variations between each successive data row, we fit a 2nd-order polynomial for incidence angle vs. latitude for each flow field. See appendix for additional details on Magellan radar calibration.
 Synthetic stereo anaglyphs were generated for each flow field to aid interpretation and provide qualitative constraints on topographic effects. The anaglyphs were generated using Magellan SAR data in conjunction with the Global Topography Data Record (GTDR) [Kirk et al., 1992; D. Young, personal communication, refer to http://www.geology.smu.edu/∼tectonics/young/young.html]. The GTDR represents elevation differences of 1 m, although the altimeter footprint is large (≥8 km). Slope information was obtained by taking representative transects along flows and calculating average slopes over a 100 km baseline (except where noted). This baseline length was selected because it is large enough to include a significant number of data points, but small enough to show variations within each flow field.
 Observations and mapping of the lava flow fields focused on flow morphologies, flow stratigraphy, and radar backscatter characteristics. Morphologic analysis was conducted to constrain lava emplacement regimes, including whether lava was emplaced in broad lobes or in channel/levee systems and whether individual flows are simple or compound at the scale of the Magellan SAR data, as well as to constrain lava transport styles by identifying morphologic characteristics representative of surface-fed and tube-fed flows. Flow stratigraphy was examined to support morphologic observations, provide further constraints on flow emplacement, and identify, if present, evolutionary trends in eruptive style. Crosscutting relationships were also examined between flows and other spatially associated features. Radar backscatter characteristics were analyzed within and between flow units and flow fields and compared to terrestrial flow field radar data.
4. Flow Field Descriptions
4.1. Turgmam Fluctus
 Turgmam Fluctus (58.4–52.7°N, 215.8–225.1°E) extends from Iris to Tikoiwuti Dorsa (Figure 2a), covering ∼140,000 km2 of the region between Vinmara and Ganiki Planitiae (to the west) and Virilis Tesserae (to the east). Mapping completed as part of the current analysis differs somewhat from the recently published 1:5 M geologic map of the Pandrosos Dorsa Quadrangle [Rosenberg and McGill, 2001]. Specifically, two radar-dark outcrops within the flow field that were mapped as smooth local plains material (ps) are included herein as part of the flow field, and the flow field margin is different in a few locations, although the differences in mapping do not significantly affect interpretations of flow field emplacement. The flow field was imaged at 28.7–31.4° incidence and displays backscatter coefficients ranging from −1.5 to −20.9 dB. Turgmam Fluctus superposes relatively radar-dark plains that exhibit numerous small volcanic constructs, fractures, and ridges. Rosenberg and McGill  indicate that Turgmam Fluctus superposes their radar-dark regional plains material (prc), densely lineated material (ld), and linear belt material (bl) units. The plains deformation that produced ridges in the region primarily occurred before emplacement of the flow field; ridges are breached by lava flows and the flow field lacks features associated with shortening. Fractures are typically buried by the flows, although some fractures crosscut the uppermost surface of the flow field. The flow field emanates from the fracture system, the most recent volcanic activity having produced radar-dark flows along an 8 km portion (centered at ∼56.0°N, 217.8°E) of a NNW-SSE trending fissure set. The flow field exhibits shallow slopes (<0.15°) and was emplaced through a series of episodes that produced multiple overlapping arms, resulting in a digitate fan morphology.
 Turgmam Fluctus is spatially associated with Marake Colles (located SSE of the most recently active fissure vents), which is made up of linear ridge segments (probably genetically related to Iris Dorsa and subsequently modified by Turgmam flows) adjacent to a cluster of small shields and domes (1–5 km diameter). Similar small volcanic constructs are found throughout the flow field. Turgmam Fluctus is superposed by Lonsdale crater (44 km diameter) and is located near Elizabeth crater (10.5 km diameter).
4.2. Zipaltonal Fluctus
 Emanating from Mist Chasma, Zipaltonal Fluctus (39.2–34.9°N, 247.2–254.1°E) covers almost 80,000 km2 of eastern Kawelu Planitia (Figure 2b). The flow field was imaged at 38.2–40.2° incidence and displays backscatter coefficients ranging from −0.2 to −24.4 dB. Zimbelman  mapped Zipaltonal Fluctus as having been emplaced on older lava flows as well as radar-dark to intermediate plains (including lineated plains), and embaying tessera along its western margin. The most recent eruptive vents are located along a 100 km portion of the Mist Chasma fissure system. The flow field exhibits shallow slopes (<0.10°) and displays a digitate subparallel morphology. Although radar brightness is variable throughout the flow field, the proximal area primarily displays relatively radar-dark flows, the medial region contains relatively radar-bright flows, and the distal portion exhibits flows of intermediate radar brightness.
 An intermediate-size shield or dome (37.2°N, 250.1°E) within the medial portion of the flow field appears to be partially buried by Zipaltonal flows. The edifice is ∼20 km in diameter and displays a radial surface texture with central radial fractures. A few tens of isolated small shields are also associated with the flow field. These shields are typically 1.5–2.5 km diameter and up to 6.5 km diameter. Other nearby features include Strenia Fluctus (which also emanates from Mist Chasma) to the north [Zimbelman, 2000], Phaedra crater (16 km diameter) to the south, and Junkgowa Corona (∼280 km diameter) to the east of Zipaltonal Fluctus. The surface of the flow field's distal-most portion is crosscut by radial fractures associated with Junkgowa Corona, and also appears to have been uplifted in conjunction with corona formation.
4.3. Tuli Mons/Uilata Fluctus Flow Complex
 Within Guinevere Planitia, a large volcanic flow complex covering ∼350,000 km2 developed through coalescence of Uilata Fluctus, flows from Tuli Mons, and other localized sources within 20.5–11.8°N, 311.4–318.6°E (Figure 2c). The flow field was imaged at 44.9–46.0° incidence and displays radar backscatter coefficients of −4.6 to −28.6 dB. Crown et al. [1994, map in preparation, 2002] mapped the flow complex as having been emplaced on mottled, lineated plains of intermediate radar backscatter, radar-dark plains with wrinkle ridges, and lava flows emanating from Madderakka Corona. Along its western margin, the flow complex may be interfingered with flows from Mehseti Patera. Tuli Mons displays ∼600 m of relief and a field of low, small edifices at its summit [Crown et al., 1994]. Slopes on the Tuli Mons edifice are shallow (<0.35°). A cluster of small edifices (shields, cones, and/or domes) is also associated with the Uilata Fluctus source region, located on a low shield-shaped construct displaying less relief and shallower slopes (<0.30°) than Tuli Mons. Other small edifices are located within the volcanic complex. The flow complex is composed of coalesced digitate aprons. Lobes are distributed radially from their sources on the two constructs, reflecting the topography of the edifices. Peripheral to both edifices' flanks, lobes are generally oriented toward the north, reflecting the regional slopes (∼0.035° over ∼900 km baseline) from Laufey Regio northward into Guinevere Planitia.
 Vard (6 km diameter) and Neda (8 km diameter) craters are superposed on the northern portion of the complex. Near the flow complex are Madderakka Corona to the south from which Koti Fluctus emanates, Atanua Mons to the southwest, Hulda Corona to the west, and Shih Mai-Yu crater (22 km diameter) to the east.
4.4. Var Mons Volcanic Center
 Var Mons, a large volcanic center, is located between Guinevere, Undine, and Navka Planitiae at 1.2°N, 316.2°E (Figure 2d). The volcano, which exhibits 1.7 km of relief, displays a complex summit region with at least three distinct eruptive centers and flows that embay its prominent rift zone, Nang-byon Chasma [Crown et al., 1994; Brian et al., 1999]. The flow field (8.1°N-5.7°S, 309.2–320.6°E) covers ∼820,000 km2, although much of its margin is poorly defined. The flow field was imaged at 43.9–46.1° incidence and displays radar backscatter coefficients of −3.5 to −27.5 dB. The flow field superposes mottled, lineated plains of intermediate radar backscatter, radar-dark plains with wrinkle ridges, and lava flows from Madderakka Corona (Crown et al., map in preparation, 2002). The Laufey Regio dark halo material, which is associated with Potter crater (46.9 km diameter), appears to superpose portions of the flow field, although some Var Mons lava flows superpose the halo material. Lava flows extend from the summit region of Var Mons, and particularly from a vent located 250 km NE of the highest point on the volcano. The flow field displays an overall digitate apron morphology, although of the five flow fields examined it is the closest to the transitional flow field morphology of Lancaster et al. . Maximum slopes within the complex are ∼2.35° (the steepest measured in this study), although slopes in the apron are typically <0.10°. Despite having a wide range of radar backscatter, the Var Mons flow field displays the least well defined lobes of the flow fields examined. Kala crater (17.4 km diameter) is superposed on the flank of Var Mons.
4.5. Mylitta Fluctus
 Located in southern Lavinia Planitia, Mylitta Fluctus (49.1–61.8°S, 347.8–359.3°E) covers ∼390,000 km2 (Figure 2e). The flow field was imaged at 20.4–23.6° incidence and displays radar backscatter coefficients of −2.0 to −17.8 dB. Fracturing and deformation in the region formed Nike and Enyo Fossae and Kalaipahoa Linea. The flow field superposes plains units of Lavinia Planitia and the uppermost flow surface was emplaced after the regional deformation, as fractures do not crosscut the flow field. The flow field extends up to 960 km from Tarbell Patera (58°S, 351°E; 24.5 × 40.3 km wide), which is located on an ∼395 × 430 km diameter shield. This edifice is adjacent to, and overlies a portion of, Jord Corona, with Mylitta flows resurfacing fractures on the corona. Maximum slopes on the shield are <1.00°, and slopes in the medial and distal portions of the flow field are <0.10°.
 Previous mapping characterized the general flow field setting and characteristics [Roberts et al., 1992]. The primary source vent is Tarbell Patera, and secondary vents may be inferred from the regional sketch map and flow maps [Roberts et al., 1992, Figure 3 and Plate 1]. Flow lobes were described as typically radar-bright with uniform surface textures and lobate margins. Central channels were identified and often associated with levees and overflows. Six smaller flow fields were mapped within Mylitta Fluctus, each of which was interpreted to represent a major eruptive event in Mylitta's history based on radar brightness and morphology. Lancaster et al.  classified Mylitta as a digitate subparallel flow field.
 The current analysis (which includes flows not previously mapped as part of the flow field) shows that flow lobes within Mylitta Fluctus display a wide variety of morphologic and radar backscatter properties. The previous division of the flow field into smaller discrete flow fields, based on C1-MIDRs (compressed-once mosaicked image data records, resampled to 225 m/pixel), has been reevaluated in the current FMAP-scale analysis. The current study indicates that Mylitta Fluctus is composed of an abundance of individual flow lobes forming an extensive flow field.
 Two clusters of volcanic shields are spatially associated with Mylitta Fluctus: one is located toward the northern end of the flow field near Chasca Vallis and associated with lineaments, and the other cluster is located on Jord Corona and extends SE along the general trend of the fossae. Alcott crater (∼65 km diameter) is located along the eastern margin of the edifice. Ejecta from Alcott is superposed on older Mylitta flows and the crater is flooded by subsequent flows.
5. Lava Flow Field Characteristics
 Interpretations of lava flow emplacement processes for the five flow fields examined are based on current knowledge of terrestrial analogs and are inherently limited by the difficulty in distinguishing surface texture changes along or across a flow from superposition of units with different surface textures. Additionally, although lava flow surfaces have been successfully modeled using scattering theory, small-scale surface morphology cannot be uniquely determined based on radar data [Campbell et al., 1993]. Magellan radar data interpretation is further complicated because each pixel represents averaged surface roughness over large (>5500 m2) areas. Although morphologies observed in the Magellan SAR data are similar to terrestrial flow morphologies, the Venusian features are often orders of magnitude larger.
5.1. General Flow Field Characteristics
 All five study areas display a digitate flow field morphology (Table 1). Two of the flow fields emanate from fissure systems and three were emplaced from centralized vents. Fissures tend to produce elongated flow fields and centralized sources typically generate apron morphologies, due to topographic effects, but vent geometry does not appear to be related to emplacement processes occurring within the flow fields. Each flow field is fairly well defined over most of its extent, though all display some diffuse margins. Flow fields range from 475 to 1450 km long and 200 to 900 km wide. Each flow field displays radar-bright and -dark lobes as well as radar-dark lava channels, although the entire range of morphologic variation is not displayed at any individual flow complex. Proximal regions tend to show the most variation of radar brightness within each flow field, as well as the greatest concentration of radar-dark flows (Figures 2 and 3). The range of radar backscatter observed is interpreted to result from the emplacement of near-vent flows over a wider range of conditions than flows in the medial and distal portions of the flow fields. However, modification of medial and distal flow surfaces during the final stages of, or following, emplacement, by processes such as flow inflation and foundering of the surface crust, may preferentially reduce the abundance of dark flows by disrupting smooth surfaces and thus may reduce the range of radar backscatter (see section 5.2). Flow fields show a significant difference in radar backscatter between roughest and smoothest flows (∼16–24 dB over a 1–4° range of incidence), although the overall backscatter values tend to decrease with increasing incidence angle. No consistent trends in radar backscatter are apparent across a flow field, nor with time (based upon the observed stratigraphy, discussed in sections 5.5 and 5.6).
Table 1. Summary Information for Flow Field Study Areas
 The five flow fields studied display numerous discrete, elongate flow lobes. These lobes are generally radar-bright and have lobate margins (Figure 2). Lobe margins typically exhibit similar radar brightness to that of the surface of the whole flow lobe, although this interpretation is somewhat biased because lobe boundaries are generally identified by a contrast in radar brightness at the 75 m/pixel scale. Bright lobes with dark, lobate margins (Figure 4a), dark lobes with bright, digitate margins (Figure 4b), and dark units with lobate margins are also present in the flow fields.
 Some lobes display discrete interior zones of increased radar brightness (Figure 4c), which may: 1) indicate postemplacement modification of portions of the lobe interior, 2) reflect changes in flow conditions during lobe emplacement, 3) represent older flows (kipukas), and/or 4) result from differential mantling. In the first case, broad lobes are interpreted to have been emplaced as relatively smooth flows. After emplacement, the lobe interior becomes rougher due to processes such as inflation that forms cracks (cm- to dm-scale) [Hon et al., 1994] and/or tumuli (m- to Dm-scale) [Rossi and Gudmundsson, 1996] at sub-pixel scales, or foundering of the surface crust after draining of the lobe interior. In the second case, flow roughness is interpreted to vary as the flow front advances. This may indicate that portions of the surface crust is fractured or deformed during emplacement due to differential flow within the lobe, that smooth-surfaced breakouts are emplaced at flow margins, or that the flow is responding to differences in underlying slope and/or substrate roughness. In the third case, kipukas of underlying radar-bright material are visible within the radar-dark lobe. In the last case, relatively fine-grained material may accumulate in topographic lows within and between flows. Although the last two scenarios may occur within the flow fields, they are not sufficient explanations for most occurrences of this morphology and are not consistent with the observation that the radar-dark portions are fairly common and tend to display relatively uniform margin widths along the flow lobes. Specifically, width of the radar-dark margin is not expected to be uniform along an individual flow if that flow is merely bifurcating around a kipuka, nor is mantling expected to commonly produce uniform widths of radar-dark zones adjacent to flow lobes. Both postemplacement modification and variations in surface roughness during flow advance (cases 1 and 2) have probably occurred within the flow fields and are consistent with this morphology representing a transition to the bright interior/dark, lobate margin morphology (Figure 4a). Dark lobes with discrete radar-bright zones occur predominantly in medial to distal portions of the flow fields, which are relatively flat in comparison with proximal regions. This association with topography further supports the interpretation of flow inflation, which commonly affects terrestrial basalt flows on near-horizontal slopes [e.g., Hon et al., 1994; Rossi and Gudmundsson, 1996].
 The length of individual lobes shows no simple relationships with average radar backscatter, backscatter variation within the lobe, nor flow widening and branching. Also, lobes tend to have as great a range of radar backscatter across a flow as along it. Lobe lengths are variable within and between flow fields, with maximum observed lobe lengths reaching 100 km within Turgmam Fluctus, 75 km within Zipaltonal Fluctus, 500 km within the Tuli Mons/Uilata Fluctus complex, 140 km within the Var Mons flow field, and 300 km within Mylitta Fluctus.
5.3. Lava Channels
 Lava channels are common within the flow fields and normally display relatively radar-dark surfaces and smooth (low sinuosity) margins (Figures 5 and 6). Channels generally occur as simple linear to sinuous monofilament conduits, although channels may display distributary morphologies. Typically, channels are up to a few tens of kilometers long and a few kilometers wide. Channels may occur as central features within a lobe (Figures 5a and 5b) or may not be associated with a distinct flow lobe (Figure 5c). Zipaltonal Fluctus displays a lava channel system that appears discontinuous (Figure 5d). This may be due to lava tube formation, changes in surface roughness along the channel surface, narrowing of the channel beyond the limits of resolution, partial burial by subsequent flows/overflows, and/or the development of independent channels along a similar trend.
 Many channels are associated with adjacent radar-bright surfaces, some of which are interpreted to be levees (Figures 5c and 6). Some ambiguities are present in levee identification, as it is not possible in all cases to distinguish levees that formed synchronously with a channel from preexisting topography affecting the course of a subsequent channel.
 Some lobes appear to have had a central channel that was later inundated by a relatively smooth-surfaced flow (Figure 7). This process may be identified by its tendency to cause the interior lobe margin (bounding the inundated portion) to have a sinuosity that is intermediate to that of channels and typical lobe peripheries. This interpretation is supported by previous analysis of a flow lobe within Mylitta Fluctus that indicated the peripheral margin (see Figure 7a) had a higher fractal dimension and greater range of fractal scales than the interior margin [Bruno and Taylor, 1995]. It is not possible in any of the channel inundation examples to determine the length of time between channel formation and flooding. In some examples, it is possible that the inundation occurred as channel overflows while the channel was still active, but after the channel margins had stagnated.
5.4. Flow Branching
 The lava flow fields display branching within both lobes and channels. Some lobe branching morphologies observed in Magellan 75 m/pixel data suggest simple flow emplacement, especially near source regions (Figures 3 and 8a). This is consistent with higher extrusion rates expected near the vent, which tend to produce simple flows [Walker, 1972]. Other branching morphologies suggest compound flow emplacement (Figure 8b) based on the interpretation that different portions of the lobe flow front were contemporaneously active, though in some cases flows may be responding to topography that is too small to resolve in the Magellan altimetry data. It should be noted that the ability to definitively distinguish between simple and compound flows is limited due to the resolution of the data set and the nature of radar data. Lobes in many places are divisible into flow units down to the limits of resolution, suggesting that compound flow emplacement may be widespread. This would, in fact, be expected of Venusian flow fields by analogy to terrestrial pahoehoe flows, which are emplaced in a compound manner at the meter-scale (orders of magnitude smaller than resolvable features in the Magellan data set).
 In some locales, distributary flow morphologies appear to be associated with a local source (Figure 9). This local source could be a flank vent associated with a minor rift, as is suggested for the example shown in Figure 9a, or a lava tube, as interpreted for the example shown in Figure 9b. The ability to differentiate between the two types of sources is limited by the resolution of the Magellan SAR data.
5.5. Flow Stratigraphy and Topographic Effects
 Gross topographic effects are evident in the tendency for narrow lobes to be emplaced on relatively steep slopes (proximal regions), whereas broader lobes are associated with shallower slopes (distal regions) (Figure 2). Qualitative assessment of the influence of topography on lava emplacement suggests that flows are responding to a smaller scale of topography than that represented in the Magellan GTDR.
 Flow units within the Venusian flow fields commonly display multiple overlapping relationships at a given locality (Figures 5b and 10), suggesting that flow field emplacement was complex rather than developing during a few discrete events. For radar-bright lobes, the high degree of overlap and the general lack of flow deflection indicate that the scale of topography to which these flows respond is generally larger than the thickness of individual flow units. Radar-dark flows appear to be much more responsive to topography (Figures 7 and 8a) and to the presence structural features (Figure 11). These relationships suggest a correlation between radar brightness and flow rheology, specifically that the radar-dark units are less viscous (and less thick), although postemplacement modification may alter the final surface roughness of flow units (see section 5.2 for discussion).
 It is expected that radar-facing margins will be brighter than those facing away from the radar if flows have appreciable thickness, although the minimum flow thickness required to produce this effect will vary depending on the margin profile and the imaging geometry. This effect is not observed within any of the flow fields examined, nor are flow thicknesses sufficient to be observed in Magellan altimetry. These observations indicate that individual flow lobes are relatively thin, consistent with previous estimates (e.g., 10–30 m for portions of Mylitta Fluctus [Roberts et al., 1992]). Preliminary modeling suggests that the lack of a bright radar-facing margin places an upper limit of ∼15 m on flow thickness for the flow fields examined [Kreslavsky and Vdovichenko, 1997].
 Based on the flow stratigraphy, no temporal progression in emplacement style is observed, suggesting that the full suite of observed flow morphologies may form throughout flow field development. This finding is consistent with eruptive histories of terrestrial and other Venusian volcanic centers [Stofan et al., 2001].
5.6. Spatially Associated Features
 In addition to complex intraflow field stratigraphy, lava flows exhibit complex superposition relationships with other features that are spatially related to the flow fields. Small shields and/or lava domes are commonly associated with the flow fields and typically range from 2–10 km diameter. Small edifices display a wide range of morphologies at various scales (Figure 6). At the large scale, edifices may display margins that are scalloped or smooth and lobate, and may have summits that are peaked or flat, and display one or more pits. At the small scale, surfaces may be rough or smooth at the radar wavelength. Some shields and domes are covered by flows, whereas others appear to be superposed on the flow fields. At a few locations within the flow fields, flows appear to be emplaced both before and after construction of an individual edifice (Figures 8b and 12), although in most cases temporal relationships are ambiguous.
 Other features, such as coronae (Figure 2e), paterae (Figure 2c), and rifts (Figures 2a, 2b, 2d, and 2e), have developed preceding, during, and/or after flow field development. Temporal relationships are frequently unclear, and the chronology of feature formation may be complex. For example, along an individual trend, some rifts associated with the Zipaltonal Fluctus proximal region display differences in radar brightness that are spatially correlated with lava flows (Figure 11). This appearance is interpreted to indicate that portions of these fractures were periodically resurfaced as the fractures developed, producing differences in the fracture offset along an individual trend. More generally stated, this indicates that eruptions occurred periodically over the duration of rifting, suggesting that both volcanic and tectonic activity occurred over protracted durations. All of the flow fields examined superpose volcanic plains units locally, although only the uppermost flow field surface may be inferred to be younger than the plains.
5.7. Radar Backscatter Characteristics and Comparison to Terrestrial Lava Flows
 Radar backscatter characteristics of flow units within the five Venusian flow fields are compared, using data averaged over a 3 × 3 pixel grid to provide representative values (Figure 13). The flow fields display a significant range of backscatter coefficients between the roughest and smoothest surfaces, from ∼16 to 24 dB over a 1–4° range of incidence. Although each flow field displays numerous radar-bright and -dark flows, the actual calibrated radar backscatter coefficients vary, showing a dependence on the incidence angle. This behavior is characterized by the empirically derived formula of Muhleman , which describes Venus' average surface backscatter coefficient as a function of incidence angle (Figure 13) [see also Pettengill et al., 1988; Ford and Plaut, 1993; Campbell, 1995]. Each of the Venusian flow fields examined in the current study display lobes with higher and lower radar backscatter coefficients with respect to incidence angle than the global average. Although Figure 13 suggests a dichotomy between radar-bright and -dark flows, lobes within each flow field display a continuum of radar backscatter coefficients. Channels within the flow fields tend to have intermediate to low radar backscatter coefficients.
 In order to compare radar backscatter characteristics of Venusian flow features with those of terrestrial lava flows, AIRSAR (airborne multipolarization, multifrequency synthetic-aperture radar) data were utilized for lava flows in and around Kilauea Caldera (Kilauea Volcano, Hawaii, USA) (J. J. Plaut et al., The unique radar properties of silicic lava domes, manuscript in preparation, 2002) (hereinafter referred to as Plaut et al., manuscript in preparation, 2002) and from Pisgah Crater (Mojave Desert, California, USA) [Plaut, 1991]. This analysis compares Magellan SHH data that have been averaged over 3 × 3 pixel grids with AIRSAR data that derives SHH values using a linear interpolation of pixel-averaged CHH (5.7 cm) and LHH (25 cm) data (Figure 13). For a given incidence angle, the Venusian flows have lobes that display similar to lower radar backscatter coefficients than the terrestrial flows. This is significant for interpretation of morphologic analogs. For example, dark-surfaced lobes with bright margins are interpreted to be sheet flows whose margins have undergone significant deformation, on average over the spatial extent of the individual pixels. Higher backscatter coefficients associated with the flow margins may result from minor deformation across an entire pixel, small areas of significant surface roughness averaged with smooth surfaces, or some intermediate case. Deformation probably resulted from increased flow viscosity due to cooling. As an alternative to deformation, flow margins may display higher backscatter coefficients due to an abundance of rough-surfaced breakouts along the lobe periphery. The range of radar backscatter coefficients are also significant because they indicate that the lava flow fields examined in this study have similar surface characteristics to flows at other Venusian volcanic centers [e.g., Campbell and Campbell, 1992; Campbell and Rogers, 1994], suggesting that the emplacement processes for the flow fields examined herein may be common on Venus.
 Analyses of the morphology, stratigraphy, and surface roughness properties of five Venusian lava flow fields were conducted to investigate flow field emplacement processes and the development of volcanic centers. The flow fields all display digitate morphologies and exhibit shallow slopes (maximum slopes 0.10–2.35° over 100 km baselines) over large areas (∼80,000 to 820,000 km2). Flow field surfaces display complex patterns due to emplacement of abundant, typically radar-dark channels and numerous radar-dark, -intermediate, and -bright lobes of various sizes and planform morphologies. Flow field surfaces tend to show the greatest variation in radar brightness within proximal regions, and surfaces of individual lobes within the flow fields may display uniform or varied radar brightness. However, at the Magellan FMAP wavelength and pixel size (12.6 cm, 75 m/pixel), individual flow lobes are similar or smoother than terrestrial lava flows. The Venusian flow fields display similarities in radar backscatter patterns and planform morphologies to terrestrial lava flow fields, although the larger scale and larger range in surface roughness of Venusian features suggest that simple analogies may not be sufficient.
 The flow fields developed through emplacement of both simple and compound flow lobes, based on lobe morphologies and branching patterns. In addition to surface-fed lava extending directly from vent regions, the presence of tube-fed flows is suggested by distributary morphologies from local sources within the flow fields and the presence of discontinuous channels. Flow inflation may have also occurred within medial and distal portions of some flow fields, as lobes displaying discrete zones of increased radar brightness are observed. Alternatively, these zones may be attributed to differences in emplacement conditions across a flow lobe.
 The flow fields display a variety of lobe and channel morphologies, characterized by variations in branching, surface roughness patterns, and margin sinuosity. Channels commonly show radar-dark surfaces with adjacent radar-bright levees. Several lines of evidence suggest that flow lobes within the five selected flow fields are relatively thin (less than a few tens of meters), such as the high degree of lobe overlap, lack of flow deflection by preexisting flows (for radar-bright lobes), and absence of lobe margin radar brightness differences due to the radar look direction. Some correlation appears to be present between radar brightness and planform morphology, flow thickness, and insensitivity to underlying topography, suggesting that increased radar backscatter may be indicative of higher flow viscosities, although postemplacement modification (such as by flow inflation) complicates this relationship. This interpretation is further supported by the tendency of radar-dark flows to occur in proximal regions and in channel systems, where we expect hotter, less viscous flows to have been emplaced.
 The selected Venusian volcanic centers developed compound flow fields with complex emplacement histories, indicated by overlapping relationships of numerous, discrete lobes and channel networks. Complicated superposition relationships within the flow fields and with associated features, such as shields, domes, and coronae, suggest the volcanic centers were active over protracted durations. The uppermost flow surfaces are among the youngest geologic units locally, although flow fields both bury tectonic features and are deformed by tectonic activity. The development of these flow fields does not indicate any evolutionary trend in eruptive styles within a flow field, nor any consistent temporal sequence relative to other tectonic or volcanic features.
 To apply this correction, the incidence angle is required for each pixel. Since incidence of the Magellan SAR is a function of latitude, image mosaics may be calibrated by applying a line-by-line correction. Incidence angle vs. latitude is provided at 1° latitude increments by Campbell . A second-order polynomial was fit to incidence vs. latitude (Φ) data for the latitude range covered by each image mosaic in this study, such that:
where A, B, and C are constants.
 A linear scaling was used to obtain the latitude of each line in each image mosaic:
where y is the line number (starting at y = 0), ymax is the maximum line number in the image, S is the southernmost latitude in the image, N is the northernmost latitude in the image, and latitudes south of the equator are expressed as negative numbers. Note that this is accurate for sinusoidal projections (as were used for the flow field mosaics in this study), but not all others, such as Mercator projections.
 The authors would like to thank Jeff Plaut for providing AIRSAR data and assistance in Magellan radar calibration and interpretation, Bruce Campbell for reviewing a preliminary version of the manuscript, Sarah B.Z. McElfresh for aid in processing FMAPs using ISIS, and Duncan Young for help generating synthetic stereo anaglyphs. This manuscript has benefited from reviews by Ellen Stofan and Misha Kreslavsky, as well as from discussions with Steve Anderson, Mark Bulmer, and Tracy Gregg. This research was supported by NASA grants NAG5-4037 and NAG5-10561 from the Planetary Geology and Geophysics Program.