Flood, R.D., Piper, D.J.W., Klaus, A., et al., 1995Proceedings of the Ocean Drilling Program, Initial Reports, Vol. 1553. MORPHOLOGY AND STRUCTURE OF AMAZON CHANNEL1Carlos Pirmez2 and Roger D. Flood3ABSTRACTAmazon Channel displays a relatively smooth, concave-up, longitudinal talweg depth profile that suggests a system in equilibrium. The gradual seaward decrease in channel slope occurs despite large variations in the gradient of the valley over whichthe channel is built. Equilibrium was apparently reached by adjustment of channel slope by two basic processes: changes ofchannel sinuosity and entrenchment/aggradation of the channel talweg. Channel equilibrium was disrupted in the past, when aknickpoint formed at a channel bifurcation site, associated with the formation of a new channel down-fan and causing a relativebase-level drop. The magnitude of base-level drop increases with pre-bifurcation channel sinuosity and with aggradation of thetalweg above the adjacent fan. Present-day channel morphology shows that knickpoints would be most pronounced for a bifurcation occurring on the middle fan, and less pronounced on the upper and lower fans. The present longitudinal profile indicatesthat past knickpoints have largely been erased from the profile. The mechanisms associated with channel equilibrium involvedsudden sinuosity changes and channel entrenchment upstream of the bifurcation site, and marked aggradation downstream as anew channel levee formed. Channel bifurcation is related to periods of enhanced channel progradation interpreted to resultfrom increased influx of terrigenous sediment to the fan.INTRODUCTIONSubmarine fans display a network of distributary channels that arethe pathways for turbidity currents. The overflow of channelized turbidity currents leads to the formation of levees adjacent to the channel and the accumulation of lens-shaped deposits termed channellevee systems. Fan growth results from the stacking and overlappingof channel-levee systems, resulting from lateral channel switching,interspersed with mass-flow deposits. It has been suggested from thestudy of several fans that only one channel-levee system is active atany time during fan growth (Damuth et al., 1983b, 1988; Droz andBellaiche, 1985; Weimer, 1989). During periods of little sediment input to the fan, such as during the present sea-level highstand, pelagicand hemipelagic sediment blankets the fan surface because terrigenous sediment is prevented from reaching the fan (e.g., Damuth et al.,1988).Submarine channels on deep-sea fans usually have a sinuous planform. On the mud-rich fans such as the Amazon (Damuth et al.,1983a), Rhone (Droz and Bellaiche, 1985), Mississippi (Garrison etal., 1982; Weimer, 1991), and Indus (Clark et al., 1992), channel sinuosity is commonly greater than 2 with recurving, and some cutoff,meander loops that resemble those of subaerial rivers (Flood and Damuth, 1987; Pirmez, 1994). Flood and Damuth (1987) showed thatAmazon Fan channels become more sinuous when traversing steepervalley gradients, resulting in a gradual down-fan decrease of alongchannel slope. Flood and Damuth (1987) and Damuth et al. (1988)suggested that channel meandering is largely controlled by the gradient of the valley over which the channels are built, but that other factors such as the frequency and type of sediment load of turbiditycurrents may also constitute important controls on channel meandering. This suggests that the channels seek a graded, or equilibrium,'Flood, R.D., Piper, D.J.W., Klaus, A., et al., 1995. Proc. ODP, Init. Repts., 155:College Station, TX (Ocean Drilling Program).2Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY 10964,U.S.A.'Marine Sciences Research Center, State University of New York, Stony Brook, NY11794-5000, U.S.A.profile in which slope is adjusted so that the turbidity currents are inequilibrium with the sediment load. Because each fan channel is generally perched atop its own channel-overbank deposits (Damuth etal., 1988) and the channel floor elevated in respect to the surroundingfan surface, a bifurcation may result in the disruption of the equilibrium profile with the introduction of a knickpoint at the site of channel switching. Knickpoints on the channel depth profile will result inslope changes along the turbidity current pathways that may directlyaffect the flow and consequently the manner in which sediments aredistributed on the fan (Flood et al., 1991). A detailed analysis of themorphology and internal structure of the channel may provide cluesas to the effects of past bifurcations and possibly on the mechanismsthrough which the channel morphology adjusts to varying conditions.Amazon Channel is the youngest channel-levee system within thedistributary channel network of Amazon Fan (channel 1 of Damuthet al., 1983a; Fig. 1). The channel was surveyed in detail, with almost100% SeaBeam coverage, closely spaced seismic reflection profiles(water-gun and 3.5 kHz), and several piston cores (Manley andFlood, 1988; Flood et al., 1991; Pirmez, 1994). It is a continuouschannel system that is directly connected to Amazon Canyon on theouter shelf and that extends for at least 900 km into abyssal depths.The path of Amazon Channel developed as a result of numerouschannel bifurcations (Fig. 1). In each of these events, turbidity currents flowed to the low-lying valleys adjacent to the channel-leveesystem, and a portion of the channel became abandoned as a newchannel-levee system developed downstream. The upstream segmentof the channel remained active after a bifurcation, and was continuously reused as a pathway for the turbidity currents redirected to thenew channel system downstream. As a result, different segments ofthe channel reflect a different growth history, and the upstream portions of the channel underwent a longer period of development compared to the lower segments. Bifurcations were interpreted to occurby avulsion (Damuth et al., 1983b, 1988; Manley and Flood, 1988),that is, by sudden abandonment of a portion of the channel as a resultof levee breaching.In this paper we present the results of measurements of AmazonChannel morphology combined with the interpretation of seismic reflection profiles. We first describe briefly the main morphologic parameters of the channel and how they relate to each other. The main

C. PIRMEZ, R.D. FLOODAmazon Submarine FanSeaBeam bathymetryO C25-14»Piston Cores:k L D E O pre-1984Amazon Φ FrenchΦShiptrackRelative channel stratigraphyAmazon1A.1B, 1C. 1D. 1E.1FBrownAquaPurpleBlueOraπge-1Orange-249 WFigure 1. Amazon Fan bathymetry including SeaBeam coverage of AmazonChannel. Channel stratigraphy according to the nomenclature of Manley andFlood (1988) and Pirmez (1994). Bathymetry in uncorrected meters (v 1.5km/s). Map modified from Flood et al. (1991).goal of investigating the spatial variation and cross-correlation of different morphologic parameters is to gain insight into the processesthat shape channel morphology. We then interpret seismic reflectiondata to reconstruct the channel geometry surrounding bifurcationsites and to map the spatial and temporal distribution of sedimentarysequences. The geometry and acoustic character of the seismic sequences may provide insights into the causes and effects of channelbifurcation on the channel morphology and depositional sequences.DATA AND METHODSAmazon Channel morphologic parameters were measured usingdigital SeaBeam bathymetric profiles (center beam) and large scale(40 in./degree, -1:109,368 at the Equator) bathymetric maps produced from SeaBeam data. The length of the channel was calculatedby adding the straight line distances between digitized points ( O.lkm apart on average) in the center of the channel. Channel measurements performed on individual bathymetric and seismic profiles werelinked to the coordinates of the corresponding (or closest) ship trackcrossing of the channel (Fig. 2A). The measurements could then bedisplayed in respect to distance along the sinuous channel.Channel parameters measured and calculated are (Fig. 2A, -B):talweg depth (Z, negative down), channel width (W) and relief (H),talweg topographic (t) and total (T) aggradation, channel (Sc) and valley (Sv) slope, sinuosity (P), and channel form ratio (F). W and H aremeasured between levee crests and represent bankfull values, chan-Figure 2. Scheme of channel measurement procedures. A. Channel lengthwas determined by adding the distance between every point digitized (openand filled circles). Channel parameters (talweg depth Z, width W, relief H, etc.) were measured at ship track crossings (filled circles). Valley slope isequal to ΔZ/D and channel slope is equal to Z/M. B. Channel relief (H) andwidth (W) are measured along a line perpendicular to the overall channelpath at the highest point outside the channel (levee crest). Channel relief isaverage of left (LH) and right (RH) values. Basal levee unconformity(defined by onlap of channel-levee reflections, thick line at bottom), servesas the reference for total aggradation (T). Topographic aggradation (t) ismeasured where a line perpendicular to the overall channel path reaches thelowest point outside the channel (interchannel low). The levee backside profile is determined by the path of steepest descent outside the channel (alsoprojected in A). Depths and thickness are converted from two-way timeusing a constant sound velocity (1.5 km/s).nel width represents the average of the left (LH) and right (RH) sidesof the channel (Fig. 2B). F is the width:relief ratio of the channelcross section (F W/H). Valley slope is calculated between two Zmeasurements separated by a straight "ruler" with length D 12 km(S v ΔZ/D; Fig. 2A). Channel slope is calculated between the sameZ values but using the along-channel distance M (Sc ΔZ/M). Channel sinuosity is simply the slope ratio SV:SC, which is equivalent to thedistance ratio M:D. P, Sv, and Sc are assigned to the midpoint of thechannel segment measured. The ruler is advanced along the channelby a fixed distance (4 km along channel) and the calculation repeatedin a sliding scheme. The chosen ruler length corresponds to 2 to 3times the average meander wavelength (Pirmez, 1994). Increasing D

MORPHOLOGY AND STRUCTURE OF AMAZON CHANNELbeyond 12 km results in stronger smoothing of the channel parametervariations along channel. Conversely, sliding distances smaller than4 km and small ruler lengths tend to enhance small errors in the digitized channel shape and depth measurements. Topographic aggradation represents the net elevation of the talweg (or levee crest) inrespect to the adjacent fan surface, measured along a line perpendicular to the valley orientation (Fig. 2B). Total aggradation correspondsto the thickness of deposits beneath the talweg to the base of the channel-levee system (basal unconformity on seismic profiles). The patha parcel of flow would follow once leaving the channel is characterized by first drawing the path perpendicular to the levee-backsidebathymetric contours and then measuring its length and depth (Fig.2A, -B). The levee backside profiles begin at a point on the channeltalweg, but have distance-depth values that are independent of thechannel talweg profile. Details of methodology, evaluation of potential errors and other measurements such as meander shape, wavelength, amplitude, radius of curvature, and cross-channel reliefdifference are discussed elsewhere (Pirmez, 1994).Seismic reflection data were acquired with a dual water-gunsource (2 80 in.3) and a four-channel streamer. Seismic data wereprocessed with band-pass filtering (30-180 Hz), automatic gain control (100 ms window, 10% gain applied), trace amplitude equalization (100 trace window), and displayed with variable area (positivepeaks only). This processing highlights lateral continuity of reflections at the expense of preservation of true amplitudes, although relative changes in amplitude are still preserved.CHANNEL MORPHOLOGYAt several locations Amazon Channel abruptly changes orientation (Figs. 1, 3A-3D), corresponding to the sites where a bifurcationoccurred. The abandoned channel segments from which AmazonChannel bifurcated were given color names by Manley and Flood(1988) and are identified in Figures 1 and 3. Several small channelshave been identified on the lower portion of the channel and arenamed IF to 1A in order of decreasing age (Figs. 1, 3D; Pirmez,1994). The youngest channel path is called Amazon Channel. The bifurcation sites and paths of the abandoned channels are deduced fromthe channel morphology and interpretation of the seismic reflectiondata (Manley and Flood, 1988; Pirmez, 1994; and below).Longitudinal Depth ProfileThe longitudinal depth profile is a relatively smooth, concave-upfunction of distance along the channel, suggesting a system in equilibrium. The measured channel extends for 807 km between the firstand last SeaBeam survey crossings. Distances along the channel arereferenced to the first survey crossing (0 km mark; Z -928 m). Thesystem extends an additional 60 km from the 0 km mark to the -100m bathymetric contour (Figs. 3A, 4). The canyon crosses the shelfbreak at the -12 km mark (Z -700 m, Fig. 4), and its surface expression on the shelf extends up to the -75 m bathymetric contour, landward of which it becomes buried by the pro-deltaic sediments of theAmazon subaqueous delta (Nittrouer et al., 1986). On the lower fan,small unleveed channels of low relief ( 10 m) are observed beyondthe SeaBeam survey (Damuth et al., 1983a; Moyes et al., 1978), butthese channels cannot be confidently connected to Amazon Channel.AggradationThe talweg cuts below the adjacent shelf and slope until the 60 kmmark, defining the seaward limit of Amazon Canyon (Figs. 3 A, 4,5).Between 60 and 100 km levees develop above the adjacent fan surface, marking the canyon-channel transition region (Fig. 5). AmazonChannel can be subdivided by its topographic aggra