Basin Research (2001) 13, 269±292
Fluvial response to sea-level changes: a quantitative analogue, experimental approach M. W. I. M. van Heijst and G. Postma Faculty of Earth Sciences, Utrecht University, PO Box 80021, 3508 TA, Utrecht, The Netherlands. E-mail:
[email protected]
ABSTRACT
Quantitative evaluation of ¯uvial response to allogenic controls is crucial for further progress in understanding the stratigraphic record in terms of processes that control landscape evolution. For instance, without quantitative insight into time lags that are known to exist between sea-level change and ¯uvial response, there is no way to relate ¯uvial stratigraphy to the sea-level curve. It is dif®cult to put ®rm constraints on these time-lag relationships on the basis of empirical studies. Therefore, we have started to quantify time-averaged erosion and deposition in the ¯uvial and offshore realms in response to sea-level change by means of analogue modelling in a 4r8-m ¯ume tank. The rate of sea-level change was chosen as an independent variable, with other factors such as sediment supply, discharge and initial geometry kept constant over the course of 18 experiments. Our experimental results support the common view that neither fall nor rise in sea level affects the upstream ¯uvial system instantaneously. An important cause for the delayed ¯uvial response is that a certain amount of time is required to connect initial incisions on the newly emergent shelf (canyons) with the ¯uvial valley. Lowering of the ¯uvial longitudinal pro®le starts only after the connection of an active shelf canyon with the ¯uvial valley; until that moment the pro®le remains steady. We quanti®ed the process of connection and introduced the quantity `connection rate'. It controlled, in conjunction with the rate of sea-level fall: (1) the amount of ¯uvial degradation during sea-level fall; (2) the total sediment volume that bypasses the shelf edge; (3) the percentage of ¯uvial relative to shelf sediment in the lowstand delta; (4) the volume of the transgressive systems tract and (5) the amount of diachroneity along the sequence boundary. Our experiments demonstrate also that the sequence-stratigraphic concept is dif®cult to apply to continental successions, even when these successions have been deposited within the in¯uence of sea level.
INTRODUCTION Progress in unravelling the ¯uvial record in terms of allogenic controls is hindered by lack of quantitative insight into the processes that control erosion and deposition over time scales of 100±1000 years, i.e. landscape evolution at a large scale. It is generally understood that rivers play an important role in sediment delivery to the shelf, especially during sea-level lowstands (e.g. Suter & Berryhill, 1985; Bartek et al., 1990; Coleman & Roberts, 1990). It was also recognized that ¯uvial response to sealevel change has signi®cant implications for sediment delivery and depositional geometries on the shelf and slope (Butcher, 1990; Wescott, 1993). However, little is known about the processes that control longitudinal river pro®le adjustments and the timing of shelfal and adjacent ¯uvial deposition in relation to climate and relative #
2001 Blackwell Science Ltd
sea-level change. This hinders, for instance, application of the sequence stratigraphic concept to the ¯uvial realm, where different allocyclic causes (climate, tectonics, eustasy) and different river processes can produce similar stratigraphy in the ¯uvial realm (the problem of convergence, e.g. Schumm, 1991). For example, any downward adjustment in longitudinal valley pro®le will cause an erosive surface, irrespective of whether this bounding surface relates to a fall in sea level, to tectonism or to climate change controlling the amount of discharge and sediment load in the river (Shanley & McCabe, 1994; Quirk, 1996). Reviews on hydrodynamics (Thorne, 1994) and ®eld studies (e.g. Blum & Price, 1998; ToÈrnqvist, 1998) refer to this complicated response of the ¯uvial system. The longitudinal pro®le grades towards base level, which is equal to sea level in coastal regions, although we recognize that rivers can locally erode below sea level 269
M. W. I. M. van Heijst and G. Postma
(Salter, 1993; Schumm, 1993; Best & Ashworth, 1997). Longitudinal-pro®le adjustment forced by a fall in sea level will start at the coastline and progresses landward by headward erosion resulting in knickpoint migration (Salter, 1993; Leeder & Stewart, 1996; Quirk, 1996), a view which is substantiated by analogue experiments by Wood et al. (1993) and Koss et al. (1994). Migration of one single knickpoint, or of an array of local knickpoints (Gardner, 1983), retards the adjustment of the valley pro®le's gradient. The point of intersection of the old and new graded longitudinal pro®le, i.e. the most landward knickpoint, would put an upward limit on the stratigraphic effect of the sea-level fall (cf. Leopold & Bull, 1979; Posamentier & Allen, 1993; Schumm & Ethridge, 1994). A drop in sea level is felt only several kilometres upstream for small, high-gradient rivers, whereas large, low-gradient rivers with larger drainage basins seem to adjust their pro®les 100s of kilometres upstream (Blum & ToÈrnqvist, 2000). For instance, in response to the last glaciation the small Obitsu River incised about 15 km upstream (Saito, 1995), the Colorado River nearly 100 km (Blum, 1993), and the Mississippi 300 km (Saucier, 1996). Fisk (1944) previously inferred even up to 1000 km for the Mississippi River (see Table 1). The morphologic concept of river-pro®le adjustment by headward erosion thus implies a time lag between the onset of a sea-level fall and the upstream adjustment of the river's longitudinal pro®le (Butcher, 1990). Our understanding of the time lag is poor (Shanley & McCabe, 1994; Quirk, 1996; Dalrymple et al., 1998). The time lag may cause erosional and depositional cycles in the coastal zone to be out-of-phase with the sea-level cycles (Ethridge et al., 1998), which illustrates also a major dif®culty in the application of sequence-stratigraphic concepts to ¯uvial strata. Through numerical and physical experimental studies we may be able to resolve some of this ¯uvial complexity by carefully testing the impact of each parameter on stratigraphy (Ethridge et al., 1998; Marriott, 1999; Blum & ToÈrnqvist, 2000). In this paper, we present quantitative results of analogue ¯ume modelling on the ¯uvial response to various rates of sea-level fall. Sea level is the isolated variable in our study, while initial topography, discharge, sediment
supply and tectonic subsidence were held constant. We performed a series of experiments that are designed to represent a common Quaternary passive margin setting with a shelf gradient steeper than the downstream reach of the ¯uvial valley and a nearly ¯at coastal plain gradient.
METHODOLOGY Experiment facility The set-up consists of an experimental tank of 4 r 4 r 1 m size serving as a depositional basin and is connected to a rectangular duct of 4 r 0.11 r 0.5 m in length that serves as the ¯uvial valley (Fig. 1, Table 2). The tank contains a sediment table with sidewalls that support a sand sheet, which forms the coastal plain, shelf, slope and basin con®guration (Table 3). A water tap with ¯ow meter provides discharge, and a sediment feeder with adjustable conveyor-belt speed controls the sediment supply rate. Both are located at the upstream end of the ¯uvial valley and act as a surrogate for the drainage basin. The applied sediment is unimodal, medium quartz sand (Table 2) that is supplied by the feeder and is used as substrate. An adjustable level of over¯ow controls the water level (sea level) in the main tank. An automatic positioning system, with x- and y-axes, is attached to the ceiling above the main tank. It carries a Dynavision SPR-2 laser sensor to collect altitude data (z-axis) of the coastalplain±shelf±slope±basin topography. The data are measured according to a 20 r 20-mm grid and have an accuracy of 0.4 mm for all three dimensions. Changes of the ¯uvial valley's stream pro®le are measured by means of rulers attached to the valley's glass wall at 10-cm spacing. Scaling real-world to experimental time±space dimensions The relative dimensions of ¯uvial-valley length vs. shelf width and vertical exaggeration are scaled following Hooke's (1968) similarity of process approach. This is the only possible way to model landscape evolution, since prototype dimensions are too large to keep up with
Table 1. Quaternary river±shelf characteristics (extended from Blum & ToÈrnqvist 2000).
River
Drainage basin (km2)
River gradient
Shelf gradient
Lowstand river extension (km)
Obitsu River Hawkesbury River Colorado (TX) Brazos Mississippi
274 22 000 110 000 118 000 3344 000
0.002 0.0005 0.0004 0.0002 0.00002
0.006 0.06 0.0008 0.0003 0.00025
30 40 100 100 150
270
#
Upstream limit of in¯uence from last glacial sea-level lowstand with respect to the present shoreline (km) 15 (Saito, 1995) 140 (Nichol et al., 1997) 90 (Blum & Valastro, 1994) 70 (Anderson et al., 1996) 300±400 (Saucier, 1996) 1000 (Fisk, 1944)
2001 Blackwell Science Ltd, Basin Research, 13, 269±292
Modelling ¯uvial response to sea-level changes
Fig. 1. (a) Experimental set-up consisting of a water- and sediment-®lled basin margin (main tank) and a ¯uvial valley (rectangular duct with the sediment feeder at its upslope end). An automatic bed pro®ler (laser) measures the topography of the sedimentary basin. (b) Schematic plan view and (c) cross-section of the experimental set-up showing the x-, y- and z-axis of the bed pro®ler that is used as the coordinate system.
#
2001 Blackwell Science Ltd, Basin Research, 13, 269±292
271
M. W. I. M. van Heijst and G. Postma Table 2. Facts on the set-up and the experimental method. Experimental set-up
Properties
Dimensions
Main tank: 4r4r1 m, table with shelf-slope con®guration: 3r3.4 m Fluvial valley, Duct: 4r0.5r0.11 m x, y and z axes with values in mm (Fig. 1) Main tank: automated bed pro®ler, accuracy of x, y and z data within 0.4 mm. Applied data point spacing 20 mm. Fluvial valley: manual stream pro®le measurement with rulers spaced 100 mm apart (accuracy 2 mm). 400 dm3 hx1 1 dm3 hx1 (y1.85 kg of dry sediment per hour) Unimodal medium sand used as uniform substrate (bed material) and as supply for the ¯uvial valley. D50; median grain diameter=250 mm D90; 90 percentile grain diameter=700 mm And 40 mm
