Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction 9 – 11 October 2003 Abstract Volume -1- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Sponsors : SHELL Research Rijswijk NITG-TNO Utrecht University NSG IAMG International Association for Mathematical Geology International Association of Sedimentologists -2- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Contents Contribution Page POROSITY PREDICTION THROUGH MICRO-SCALE SIMULATION OF GRAIN REARRANGEMENT ........................................................................................................................................... 7 ALBERTS, L. ........................................................................................................................................................ 7 2D AND 3D SUBSIDENCE ANALYSIS IN A COMPLEX FORELAND BASIN: THE VENETIAN BASIN (NE ITALY). ............................................................................................................................................. 8 BARBIERI C. 1 & GARCIA-CASTELLANOS D. 2 .................................................................................................... 8 QUANTITATIVE 3D MODELING OF SEDIMENTARY TRANSPORT MECHANISMS IN THE PERIALPINE FOREDEEP (UPPER MARINE MOLASSE, LOWER MIOCENE) .............................................. 10 BIEG, U.1, SUESS, M.P.1, KUHLEMANN, A. 1 & THOMAS, M. 2 ........................................................................ 10 MODELING ALLUVIAL DEPOSITS .............................................................................................................. 11 BRIDGE, J. ......................................................................................................................................................... 11 LIMITS TO PREDICTABILITY ARISING FROM NON-LINEAR AND CHAOTIC BEHAVIOUR: RESULTS FROM A NUMERICAL STRATIGRAPHIC FORWARD MODEL OF CARBONATE SYSTEMS ............................................................................................................................................................. 13 BURGESS, P.1 & EMERY, D.2 ............................................................................................................................. 13 FROM QUINCUNX TO DELTA: A STOCHASTIC CELLULAR AUTOMATA APPROACH TO THE CREATION OF FANS, VOLCANOES AND DELTAS.................................................................................. 14 BURROUGH, P. .................................................................................................................................................. 14 MODELLING OF THE MIDDLE TO LATE PLEISTOCENE RHINE-MEUSE SYSTEM IN THE CENTRAL NETHERLANDS ............................................................................................................................ 15 BUSSCHERS, F.S. & VAN BALEN, R.T. .............................................................................................................. 15 THREE-DIMENSIONAL MODELLING OF THRUST-CONTROLLED FORELAND BASIN STRATIGRAPHY ............................................................................................................................................... 16 CLEVIS, Q. ........................................................................................................................................................ 16 MODELLING THE STRATIGRAPHY AND PRESERVATION POTENTIAL OF MEANDERING STREAM DEPOSITS .......................................................................................................................................... 17 CLEVIS, Q. & TUCKER, G. ................................................................................................................................ 17 NUMERICAL AND ANALOGUE MODELLING OF SEDIMENTARY BASIN FORMATION ............. 18 CLOETINGH, S., GARCIA-CASTELLANOS, D., TER VOORDE, M. & SOKOUTIS, D. ......................................... 18 FORWARD MODELING OF MEANDERING CHANNELIZED RESERVOIRS ...................................... 19 COJAN, I., LOPEZ, S. & RIVOIRARD, J. ............................................................................................................. 19 PALEOZOIC BASIN STUDY: STRUCTURE, STRATIGRAPHY AND QUANTITATIVE MODELLING IN THE SOUTHERN CANTABRIAN MOUNTAINS, NW-SPAIN ........................................................... 20 DIETRICH, B. .................................................................................................................................................... 20 -3- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 QUANTIFIED CARBONATE PLATFORM DEVELOPMENT: THE ROSENGARTEN TRANSECT (MIDDLE TRIASSIC, DOLOMITES) ............................................................................................................... 21 EMMERICH, A. ................................................................................................................................................. 21 INCORPORATION OF NUMERICAL AND PHYSICAL FORWARD MODELLING INTO UNCERTAINTY ANALYSIS METHODS ....................................................................................................... 22 FEURER, J., & VAN DER ZWAN, C.J. ................................................................................................................ 22 ROLE OF FLUVIAL TRANSPORT DURING OROGENESIS: NEW NUMERICAL AND ANALOGUE MODELLING TECHNIQUES .......................................................................................................................... 23 DANIEL GARCÍA-CASTELLANOS, KATARINA PERSSON, DIMITRIOS SOKOUTIS, IVONE JIMENEZ-MUNT ............ 23 A STRATEGY FOR AUTOMATED INVERSION OF PROCESS-BASED MODELS ..................................... 24 GEEL, C.R.& WELTJE, G.J. ............................................................................................................................... 24 3D SIMULATION OF SEDIMENTARY FACIES: APPLICATION TO LANGHIAN REEF BUILDUPS IN VALLÈS-PENEDÈS BASIN (CATALUNYA). .......................................................................................... 25 GRATACOS, O. ................................................................................................................................................. 25 STRATIGRAPHIC FORWARD MODELLING OF DEEP MARINE SYSTEMS USING SEDSIM ........... 26 GRIFFITHS, C. ................................................................................................................................................... 26 THE TECTONIC EVOLUTION OF THE NORTHERN NORTH SEA FROM THE MID-JURASSIC FORM 2D SUBSIDENCE ANALYSIS.............................................................................................................. 27 HANNE, D. ....................................................................................................................................................... 27 FINE-SCALE FORWARD MODELLING - EXAMPLE FROM A DEVONIAN REEF OF THE CANNING BASIN ............................................................................................................................................. 28 MODELLING TURBIDITY CURRENTS BY CFD SIMULATIONS ............................................................ 29 HEIMSUND, S.1, 2, HANSEN, E.W.M.3, BAAS, J.H.4 & NEMEC, W.1 ................................................................ 29 FORWARD MODELLING & PREDICTION OF AEOLIAN SYSTEMS USING FUZZY LOGIC, CONSTRAINED BY DATA FROM RECENT AND ANCIENT ANALOGUES........................................ 30 HERN, C.1, NORDLUND, U.2, ZWAN, C.J. VAN DER1 & LADIPO, K. 3.............................................................. 30 USING SUB-REGIONAL SCALE FORWARD MODELS TO CONDITION RESERVOIR-SCALE STOCHASTIC SCENARIOS ............................................................................................................................. 31 HERN, C., LAMMERS, H., BURGESS, P. & NIJMAN, M. ................................................................................... 31 INVESTIGATION OF GRAVITATIONAL MASS TRANSPORT PROCESSES - USING THE DISTINCT ELEMENT METHOD AS A MODERN TOOL IN SEDIMENT TRANSPORT MODELLING ................ 32 HUHN, K. & KOCK, I. ...................................................................................................................................... 32 MODELLING CYCLES OF FLUVIAL AGGRADATION AND DEGRADATION USING A PROCESSBASED ALLUVIAL STRATIGRAPHY MODEL ............................................................................................ 33 KARSSENBERG, D.1, BRIDGE, J.S.2, STOUTHAMER, E.1, KLEINHANS, M.G.1 & BERENDSEN, H.J.A.1 ............. 33 CHANNELISED TURBIDITY CURRENTS: NEW INSIGHTS INTO FLOW STRUCTURE AND SECONDARY CIRCULATION ........................................................................................................................ 34 KEEVIL, G. ........................................................................................................................................................ 34 PREDICTING DISCHARGE AND SEDIMENT FLUX OF THE PO RIVER, ITALY SINCE THE LGM 35 KETTNER, A.J. & SYVITSKI, J.P.M.................................................................................................................... 35 -4- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 MODELLING BEDDING AND PETROPHYSICAL CHARACTERISTICS OF TIDAL HETEROLITHIC RESERVOIRS USING A PROCESS-ORIENTED APPROACH .................................................................... 43 MARTINIUS, A.W............................................................................................................................................. 43 TECTONIC CONTROL ON PAST CIRCULATION OF THE MEDITERRANEAN SEA ........................ 44 MEIJER, P.1, SLINGERLAND, R.2 & WORTEL,R.1 .............................................................................................. 44 NUMERICAL MODELING OF THE SEDIMENTARY SYSTEMS OF LARGE-SCALE COASTAL SYSTEM TRACTS .............................................................................................................................................. 45 NIEDORODA, A.W., REED, C.W.1, DAS, H.1, DONOGHUE, J.2, FERRAGHAZZI, S.2 , WANG, Z.B.3 & STIVE, M.4.................................................................................................................................................................... 45 3-D PHOTOREALISTIC MODELS OF GEOLOGIC OUTCROPS, A TOOL TO COLLECT QUANTITATIVE DATA; METHODOLOGY AND HISTORIES EXAMPLES IN UTAH AND WYOMING ......................................................................................................................................................... 47 OLARIU, C. ....................................................................................................................................................... 47 UPSCALING THE TIME PARAMETER IN STRATIGRAPHIC SIMULATION MODELS: EFFICIENT USE OF HIGH-MAGNITUDE LOW-FREQUENCY EVENTS ..................................................................... 48 OVEREEM, I.1 & STORMS, J.2 ............................................................................................................................ 48 ANALOGUE MODELLING AS A TOOL FOR INVESTIGATING CAUSALITY IN ALLUVIAL STRATIGRAPHY ............................................................................................................................................... 50 GEORGE POSTMA ............................................................................................................................................. 50 2D SUBSIDENCE MODELLING OF MIDDLE TRIASSIC CARBONATE PLATFORM DEVELOPMENT IN THE LOMBARDIAN ALPS (ITALY) ......................................................................... 52 SEELING, M. ..................................................................................................................................................... 52 THE COMMUNITY SURFACE-DYNAMICS MODELING SYSTEM: AN ENVIRONMENT FOR DEVELOPING, SHARING, AND USING SEDIMENT MODELS .............................................................. 53 SLINGERLAND, R. ............................................................................................................................................ 53 AVULSION, AUTOGENIC OR ALLOGENIC CONTROLLED? ................................................................ 55 STOUTHAMER, E. & BERENDSEN, H.J.A. ........................................................................................................ 55 VARIATION IN DIP OF LATERAL ACCRETION SURFACES IN SUBRECENT FLUVIAL DEPOSITS, PANNONIAN BASIN, HUNGARY: A REFLECTION OF CLIMATIC FLUCTUATIONS OR JUST MEANDERING EXCURSIONS? ...................................................................................................................... 56 SZTANÓ, O. 1 & MÉSZÁROS F. 2 ....................................................................................................................... 56 THE ROLE OF NUMERICAL SEDIMENTARY PROCESS MODELS IN HYDROCARBON EXPLORATION AND RESERVOIR CHARACTERIZATION .................................................................... 61 TETZLAFF, D. ................................................................................................................................................... 61 MODELS THAT TALK BACK ....................................................................................................................... 62 TIPPER, J.C. ...................................................................................................................................................... 62 MODELLING THE IMPACTS OF CLIMATE CHANGE ON EROSION, SEDIMENT PRODUCTION, AND LANDSCAPE EVOLUTION .................................................................................................................. 65 TUCKER, G. ...................................................................................................................................................... 65 SCALED MODELS OF SYNTECTONIC SEDIMENTATION – REVIEW OF PRINCIPLES AND FIRST RESULTS ............................................................................................................................................................. 66 URAI, J.L. & KUKLA, P.A. ............................................................................................................................... 66 -5- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 OUTCROP AND SEISMIC CONSTRAINTS FOR 3D NUMERICAL STRATIGRAPHIC FORWARD MODELING ........................................................................................................................................................ 67 VAN BUCHEM, F., GRANJEON, D. & ESCHARD, R. ......................................................................................... 67 IMPACT OF DISCHARGE AND SEDIMENT FLUX ON BASIN MARGIN ARCHITECTURE: AN EXPERIMENTAL APPROACH ....................................................................................................................... 69 VAN DEN BERG VAN SAPAROEA, A.P., POSTMA, G. & DALMAN, R.............................................................. 69 HIGH RESOLUTION 3D FORWARD STRATIGRAPHIC MODELLING OF CRETACEOUS CARBONATE PLATFORM SYSTEMS............................................................................................................ 71 VAN DER ZWAN, C.J.1, MASSE, J.-P.2, BORGOMANO, J.1, 2, LAMMERS, H.1 & FENERCI-MASSE, M.2 ................. 71 COMBINATION OF STRUCTURAL BALANCING, REVERSE BASIN AND FORWARD STRATIGRAPHIC MODELLING: SOUTHERN CANTABRIAN BASIN (NW-SPAIN) ......................... 73 VESOLOVSKY, Z. .............................................................................................................................................. 73 NON-UNIQUE SEQUENCE STRATIGRAPHIC ARCHITECTURES ........................................................ 74 WALTHAM, D., UDOFIA, M. & NICHOLS, G. .................................................................................................. 74 INSIGHTS INTO CARBONATE PLATFORM DROWNING MECHANISMS USING STRATIGRAPHIC FORWARD MODELING................................................................................................. 75 WARRLICH, G. & BURGESS, P.M. .................................................................................................................... 75 MODELLING SOURCE ROCK DISTRIBUTION AND QUALITY VARIATIONS: THE NEW OF-MOD 3D TECHNOLOGY ............................................................................................................................................ 76 ZWEIGEL, J. ...................................................................................................................................................... 76 -6- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Porosity prediction through micro-scale simulation of grain rearrangement Alberts, L. Delft University of Technology, Department of Applied Earth Sciences, Mijnbouwstraat 120, 2628 RX DELFT, The Netherlands, l.j.h.albert@citg.tudelft.nl Reservoir quality is commonly quantified in terms of the spatial distribution of porosity and permeability, where the latter critically depends on the former. Analytical methods of porosity prediction in natural sediments are not available, on account of the difficulties of describing the geometrical arrangement of multi-sized particles of irregular shape. Available techniques to estimate porosity are empirical equations that predict an exponential decrease of pore volume as a function of effective stress, or focus on cementation processes. Such methods do not incorporate uncertainty or variability of estimates, because they are tailored to data collected in a specific area. A more generally applicable method of porosity prediction is needed to enhance the usefulness of process-based stratigraphic simulators for reservoir modelling. We have developed a numerical model that simulates the compaction of particle populations at a micro-scale. The model comprises the interparticulate mechanics that occur after deposition and during shallow burial, i.e. prior to the phase in which chemical processes start to dominate compaction. Spatial continuity of the particle pack is simulated through the use of periodic boundary conditions in X and Y, whereas elastic boundaries ('cushions') are implemented in the Z direction. The particle pack thus represents an infinite layer of limited thickness, much like a single lamina. The grain-size distribution of the pack is the key parameter in the model. The model is object based, which allows the mechanics of each particle to be calculated, whereas the properties of the entire pack serve as global state variables during the rearrangement. With this model we can simulate realistic trajectories of porosity decline for natural particle-size distributions under several burial scenarios. Calibration of the model to physical experiments and sensitivity analyses will be discussed, to provide a basis for the evaluation of uncertainty in model predictions. The resulting particle pack can be used for simulations of permeability, conductivity and wave-propagation studies. -7- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 2D and 3D subsidence analysis in a complex foreland basin: The Venetian Basin (NE Italy). Barbieri C. 1 & Garcia-Castellanos D. 2 Department of Earth Sciences, University of Pavia, via Ferrata 1, 27100 Pavia, Italy, chiara.barbieri@manhattan.unipv.it 2 Department of Tectonics, Vrije Universiteit, de Boelelaaan 1085, 1081 HV Amsterdam, The Netherlands. 1 The Venetian basin (NE Italy) is characterized by a complex geometry due to the partial superposition of three foredeeps, different in both age and polarity, associated to the development of the surrounding belts, namely the Dinarides, the Eastern Southern Alps and the Northern Apennines, which underwent their main orogenic phases through Tertiary time. Therefore, it represents an interesting area to observe and better understand the interplay between the evolution of mountain belts and the response of the adjacent foreland. The main goal of this work is to recognize and quantify the contribution of surface loads (i.e. mountain belts, initial water depth, sedimentary infill) and of possible deeper-sourced (hidden) loads, to the subsidence observed in the basin. This study constitutes the base for the analysis of the sedimentary infill. To this purpose, a former 2D flexural numerical modelling is applied along two key transects trending NE-SW and N-S respectively, and a 3D flexural analysis is also performed to solve possible misfits occurring in the sectors which have been influenced by the combined effect of mountain belts. Geological and geophysical data concerning the surrounding belts have been collected from literature. In particular, data recently published in the context of the TRANSALP project, together with previously available information on both subsurface and surface studies, provide an accurate data set to study the effect of the Eastern Southern Alps on the Venetian basin. Moreover, interpretation and depth conversion of seismic lines and paleobathymetric analyses have been performed in this work on some industrial transects and wells respectively (courtesy of ENI), to integrate and improve the already existing data. Two dimensional forward modelling has been performed according to the numerical method discussed in Zoetemeijer et al. (1990). Gravimetric curves have been also calculated to provide an independent constraint to test the reliability of the models, which show the present geometries occurring in the basin. The model calculated on the basis of NE-SW transect aims to test the flexural effect of the Dinaric belt which underwent the main orogenic phases during Paleocene-middle Eocene. The best fit has been obtained for a continuous plate condition and an Effective Elastic Thickness value varying along section. Nevertheless, the tectonic phases occurred afterwards with different direction make difficult to match all the points representing the base of the Dinaric foredeep. The second model, calculated along the N-S trending section, is focussed on the flexure related to the main orogenic phases of the Eastern Southern Alps (late Miocene-early Pliocene). -8- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 In this case a good fit has been obtained for a broken plate condition and a constant Effective Elastic Thickness value of about 20 km. The resulting calculated flexure is strongly influenced by the initial water depth and no hidden load was needed to fit the observed flexure unlike a previously proposed model (Royden, 1993). In spite of the good fit in front of the Eastern Southern Alps, a few metres high forebulge is expected to the South from the model but it is not observed in the present Adriatic Sea. In order to better understand the meaning of this mismatch, 3D modelling has been applied by means of the software tao3D (Garcia-Castellanos, 2002). To this purpose, flexure of the plate has been analysed in two steps by progressively adding the loads due to the Eastern Southern Alps and the Northern Apennines. Results show that a forebulge form in the same area as predicted by the 2D model if the Northern Apennine load is not taken into account. Subsequently this bulge is shifted to the South by the Apennine load and it migrates eventually toward the Dalmatian region by including hidden loads, which are necessary to fit the flexure presently observed in front of this belt. In conclusion, on the basis of the calculated models, a forebulge would exist at the end of Miocene in the area corresponding to the present Venetian coastline, which subsequently migrated as a response to the Apennine-related flexure. This suggests also that the Apennines would have affected the area of Venice since at least Pliocene – early Quaternary. Garcia-Castellanos D. (2002). Interplay between lithospheric flexure and river transport in foreland basins. Basin Research, 14, 89-104. Royden L. (1993). The tectonic expression slab pull at continental convergent boundaries. Tectonics, 12, 2, 303-325. Zoetemeijer, R., Desegaulx, P., Cloething, S., Roure, F. and Moretti, I., 1990. Lithospheric dynamics and tectonic-stratigraphic evolution of the Ebro Basin. Journal of Geophysical Research, 95, B3, 2701-2711. -9- Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Quantitative 3D modeling of sedimentary transport mechanisms in the peri-Alpine foredeep (Upper Marine Molasse, Lower Miocene) Bieg, U.1, Suess, M.P.1, Kuhlemann, A. 1 & Thomas, M. 2 1 University of Tübingen, Department of Geoscience, Sigwartstr. 10, 72076 Tübingen, Germany Ulrich.bieg@uni-tuebingen.de 2 Technical University of Dresden, Institute for Planetary Geodesy At the Aquitanian-Burdigalian boundary the Molasse basin was flooded by a shallow tide- and wave- dominated seaway, linking the Rhône Basin through, the peri-Alpine foredeep with the Vienna Basin and thus with the eastern Paratethys. Mass balancing of sediments derived from the developing alpine orogen and the foredeep suggests a major export of sediments towards the West during the Upper Marine Molasse (OMM, Burdigalian). Therefore it is of great interest which paleoceanographic processes were involved and potentially controlled the westwarddirected transport. The widespread occurence of meso- to macrotidal facies successions in the Molasse Basin adverts that tidal activity was a major force driving paleocurrents. The aim of our study is therefore to provide a numerical model of sediment particle transport conditions within a shallow-water environment. We use QUODDY to model paleocurrents in this shallow foredeep. QUODDY is a Fortran implementation of a 3-D finite-element shelf circulation model by the Dartmouth College (Lynch & Werner, 1991; Lynch & Naimie, 1993; Lynch et al. (1996). It is a free-surface, tide-resolving model based on conventional 3-D shallow water equations. Atmospheric forcing terms for wind stress and temperature flux can be implemented, as wells as point source terms for rivers and adjacent alluvial plains. Further boundary conditions of the tides in the Paratethys were obtained by calculating an initial tidal mode, based on a 1-degree reconstruction of the global bathymetry in the Miocene. Based on paleogeographic data derived from Martel, Allen & Singerland (1994) and new published palaeogeographic maps of the alpine foredeep by Kempf & Kuhlemann (2002) we constructed an irregular triangular mesh with a varying grid resolution, to describe the paleobathymetry of the basin. First successful benchmarks were obtained by comparing with the model of Martel et al. (1994). Kuhlemann, J., Kempf, O. (2002): Post-Eocene evolution of the North Alpine Foreland Basin and its response to Alpine tectonics; Sedimentary Geology, Vol. 152, pp 45 – 78. Lynch, D.R., Werner, F.E. (1991): Three-dimensional hydrodynamics on finite elements. Part II: Nonlinear time-stepping model; Int. Numer. Meth. Fluids., 12, pp 507-533. Lynch, D.R. and Naimie C.E. (1993): The M 2 tide and its residual on the outer banks of the Gulf of Maine; J. of Phys. Oceanogr., Vol. 23. Lynch D.R., Ip, J.T.C., Naimie, C.E., Werner, F.E. (1996): Comprehensive coastal circulation model with application to the Gulf of Maine; Continental Shelf Research, Vol. 16, No.7, pp. 875-906. Martel, A.T., Allen, P.A., Slingerland, R. (1994): Use of tidal-circulation modeling in the paleogeographical studies: An example from the Tertiary of the Alpine perimeter; Geology, Vol. 22, pp 925-928. - 10 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modeling alluvial deposits Bridge, J. Binghamton University, Department of Geological Sciences, NY 13902-6000,Binghampton, United States, jbridge@binghampton.edu Understanding of fluvial sedimentary processes and deposits has come from field studies of modern fluvial environments, laboratory flume studies, and construction of idealized models based on these studies. Understanding and prediction of ancient alluvial deposits is based on direct modern analogs and/or idealized models (graphic or quantitative). Field studies of modern fluvial environments are difficult to undertake during floods, over large areas, and for a long time, and it is difficult to describe deposits below the water table. Recently, some of these problems have been overcome by: using remote-sensing images and DEMs for studying changes in channel and floodplain geometry; measurement of water flow and sediment transport during floods using new types of equipment; description of deposits using GPR in combination with coring, and OSL dating. Laboratory flumes have been used to study fluvial processes and deposits over a wide range of physical scales. However, scaling problems can limit the applicability of these studies to the real world, and these problems increase as the scale of the physical model decreases relative to the real-world prototype. In particular, all superimposed scales of bedform and associated strata cannot be produced in flumes, and rates of sedimentary processes are unrealistically high. Idealized models are ultimately based on studies of modern sedimentary processes and deposits: therefore, shortterm, small-scale models are well developed and most easy to test. As the spatial and temporal scale increase, useful models are more difficult to construct and test. Linkages between models of different scales are also lacking. Quantitative models for two scales of fluvial deposit are discussed here: deposits of alluvial valleys (or fans or deltas) with channel belts and floodplains, and; deposits of channel belts. The most common approach to modeling deposits of alluvial valleys (so-called alluvial architecture) is: determine geometry, proportion and location of distinct sedimentary bodies (objects) using well logs, cores, seismic or GPR; gain more information from presumed outcrop analogs; use stochastic (structure-imitating) models to simulate the geometry, orientation and location of objects between wells; simulate rock properties within objects using stochastic models. Other types of stochastic models are not object-based, and make use of transition probabilities/Markov chains or indicator variograms. Problems with the stochastic modeling approach include: difficulty in defining input parameters for the model; over-reliance on outcrop analog data of dubious quality and usefulness; unrealistic shapes and locations of objects; limited understanding of origin of deposits; no predictive value outside data region. Stochastic models are used because commercial software is available, simulations can be conditioned with available data (particularly from wells and cores), and forward (process-imitating) models are considered to be under-developed and very difficult to fit to subsurface data. However, forward (process-based) models can give much better understanding of deposits and have - 11 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 predictive ability beyond the data region. Furthermore, it is very likely that they can be fitted to subsurface data by trial-and-error with optimization (inversion). The simple 3-D alluvial architecture model of Mackey and Bridge (1995) has been substantially developed by Karssenberg and others, and can now be considered to be a quantitative 3-D fluvial sequence stratigraphy model. In addition to the processes simulated by Mackey-Bridge, this new model simulates geometric evolution of the channel network, upstream migration of knickpoints, incised channels and terraces, and the effect of aggradation-degradation cycles on avulsion and alluvial architecture. Channel-belt deposits have also been simulated using object-based stochastic models: however, existing models are very unrealistic, mainly because of poor definition of object shapes and the methods for positioning objects. Deterministic quantitative models that consider flow and sedimentary processes in channel belts are not very well developed, and it is only possible to predict sedimentary details for individual channel bar deposits. Further progress will be made as more information from modern channel belts becomes available. Recent work of relevance to modeling suggests that there are general relationships between the geometry of the different scales of bedforms (e.g., ripples, dunes, bars) in rivers and the geometry of their associated stratasets. - 12 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Limits to predictability arising from non-linear and chaotic behaviour: results from a numerical stratigraphic forward model of carbonate systems Burgess, P.1 & Emery, D.2 1 Shell International Exploration and Production B.V., Postbus 60, 2280 AB Rijswijk, The Netherlands, Peter.Burgess@shell.com 2 Dept. of Computing, University of Staffordshire Most conceptual and indeed numerical models of sedimentary systems assume simple linear behaviour, allowing straight forward prediction based on a relatively simple relationship between cause and effect. However, another possibility for sedimentary systems is non-linear behaviour leading to sensitive dependence and perhaps to chaotic behaviour. For even tiny differences in some starting condition, a system exhibiting sensitive dependence may produce quite different strata, a phenomenon known as divergence. Such systems are not easily predictable; it is not possible to predict the in detail the behaviour of such system over any time scale near or beyond the divergence timescale. To investigate this possiblity, a numerical stratigraphic forward model of a carbonate system with various non-linear components has been used to generate autocyclic meter-scale shallowing-upward parasequences and to investigate the consequences of sensitive dependence on initial conditions for parasequence development. Two model Cases with only a very small difference in initial starting topography have been run. Stratal patterns from the two runs are similar in that both cases produce stacked parasequences with a similar range of thicknesses and showing general shallowing upward trends. However, thickness and facies distributions differ significantly in detail as a consequence of divergence. Attempts to determine the Lyapunov exponent confirms that the model exhibits divergence but suggests that it is not truly chaotic These results suggest that we may be well advised to look for sensitive dependence and divergence in modern and ancient systems, and to take account of possibility of such behaviour in interpreting ancient strata; if real carbonate systems show similar divergence to these model results, predictions from conceptual and numerical forward models should be limited to general patterns unaffected by sensitive dependence. - 13 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 From QUINCUNX to Delta: A stochastic cellular automata approach to the creation of fans, volcanoes and deltas Burrough, P. University of Utrecht, Faculty Geosciences, Utrecht Centre for Environment & Landscape Dynamics, Heidelberglaan 2, Utrecht, The Netherlands, p.burrough@geog.uu.nl This presentation examines some useful stochastic models, including Francis Galton's Quincunx (or regression) board, for explaining the statistical properties of "sedimentary" structures like fans, deltas and volcanoes. This will be illustrated with a critical examination of a set of dynamic models created in a dynamic GIS environment. The paper will illustrate how well-known geological structures might develop from purely stochastic processes driven by gravity. It will show how minor variations and modifications of initial conditions can affect the detailed outcome of the model but at the same time the generation of large numbers of stochastic realisations demonstrates how easily variations on a generic theme can be arrived at, thereby enhancing our understanding of geological variation in the field. - 14 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling of the Middle to Late Pleistocene Rhine-Meuse system in the central Netherlands Busschers, F.S. & Van Balen, R.T. Vrije Universiteit Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands, freek.busscher@falw.vu.nl The Saalian to Weichselian (OIS 6 to 2) Rhine-Meuse system in the central Netherlands is a low-gradient fluvial system, located downstream of the hinge line of the southern part of the North Sea basin. As a result, the deposits of the Rhine-Meuse system (Kreftenheye Formation) consist of stacked coarse-grained packages of sand and gravel. Results of multidisciplinary analyses of new high resolution cores allow to construct a detailed model of the alluvial architecture of the Rhine-Meuse system. In addition, over 80 OSL dates place the deposits in a well constrained chronological framework. The cores are located along three large transects perpendicular to the main paleoflow direction. The easternmost transect is situated close to the hinge line, whereas the westernmost is close to the present-day coastline. A link to the offshore sediments is in progress (4th transect). Our study shows that the Rhine-Meuse deposits consist of 6 to 7, 10-25km wide stacked sediment bodies, formed during the last two glacial cycles. First results indicate that the formation of bounding surfaces is related to climate (sediment availability and discharge), glacio-eustacy and tectonic movements (glacio-isostacy). These parameters also seem to control the grainsize distributions of the separate sediment bodies. Modelling the alluvial architecture will allows us to infer the relative importance of these controlling factors. This in turn will enhance predictions of 3D grainsize distribution in this part of the Rhine-Meuse system, reducing the costs for the exploration for sand and gravel in central Netherlands. Modelling will first be done on a deltaic scale (100 km), focussing on the larger scale trends in architecture and grainsizes. Subsequently, a smaller study area (5 km) will be selected for high-resolution modelling. - 15 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Three-dimensional modelling of thrust-controlled foreland basin stratigraphy Clevis, Q. Oxford University, Mansfield Road, OX1 3TB, Oxford, United Kingdom, quintus.clevis@geog.ox.ac.uk In this study we present a 3D forward stratigraphic model, which aids in the interpretation of alluvial successions in foreland basins filled by axial-deltaic and alluvial fan systems, under conditions of variable tectonism and eustatic sea-level change. The first set of model experiments indicate that the onset of tectonic activity is reflected by rapid retrogradation of both depositional systems, widespread flooding and onlap of marine sediments. Syntectonic fluvial patterns on the axialdelta plain are dominated by bifurcating channels, swiftly relocating in response to the general rise of relative sea level induced by flexural subsidence. The resulting surface morphology of the axial delta is conical. Syntectonic eustatic sea-level fluctuations result in parasequence-scale packages of retrograding and prograding fan and deltaic sediments bounded by minor flooding surfaces and Type-2 unconformities. Incised channels are rare within the syntec tonic parasequences and are formed only during phases of tectonic quiescence when eustatic falls are not longer compensated by the subsidence component in the rise of relative sea level. Suites of amalgamating axial channels corresponding to multiple eustatic falls delineate the resulting Type-1 unconformities. Coarse-grained, incised channel fills are found in the zone between the alluvial fan fringes and the conical body of the axial delta, as the axial streams tend to migrate towards this zone of maximum accommodation. Similar suites of amalgamating axial channel belts are created when the foreland basin is detached from its substratum by a hinterland-dipping sole thrust and transformed into a thrust-sheet top basin. This is shown in the second set of experiments. Here, the competition between rates of regional flexural subsidence, and local detachment-induced uplift controls the accommodation space evolution and the stratigraphic patterns in the marine-influenced thrust-sheet top basin. Model experiments show that displacement over low-angle faults (2~6 degr.) with moderate rates (~5.0 m/kyr) results in a vertical uplift component sufficient to counteract the background flexural subsidence rate. Consequently, the basin-wide accommodation space is reduced, fluvio-deltaic systems carried by the thrust sheet prograde and part of the sediment supply is spilled over towards adjacent basins. The intensity of the forced regression and the interconnectedness of fluvial sheet sandstones increases with the dip angle of the detachment fault. - 16 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling the stratigraphy and preservation potential of meandering stream deposits Clevis, Q. & Tucker, G. Oxford University, Mansfield Road, OX1 3TB, Oxford, United Kingdom, quintus.clevis@geog.ox.ac.uk The fluvial processes of floodplain aggradation, lateral stream migration and the creation of cutt-offs have the potential to bury, obscure, expose, or even destroy portions of meandering stream deposits, resulting in a complex subsurface mosaic of sediments of a variety of ages and grainsizes. The aim of out study is to: 1)visualize the 3D spatial distribution of these meandering stream deposits and 2)quantify their preservation potential as a function of the natural process of meandering, climatic change and human influence, using the CHILD landscape evolution model. The landscape in the CHILD model is represented by an adaptable triangular mesh of nodes, especially designed for simulating the gradual shifting of meander bends. A new layer track routine is currently added in order to improve the resolution of the stratigraphic record accumulated by the model. - 17 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Numerical and analogue modelling of sedimentary basin formation Cloetingh, S., Garcia-Castellanos, D., Ter Voorde, M. & Sokoutis, D. Free University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands, sierd.cloetingh@falw.vu.nl Sedimentary basin systems provide an important source of information on the mechanical properties of the underlying lithosphere. At the same time, lithosphere rheology, appears to strongly affect the mode of basin formation and associated vertical motions. A close feed back exists between surface processes and deformation of the crustlithosphere lithosphere system at deeper levels. The combination of numerical and analogue modeling has generated new insight in the role of tectonic topography in source-sink relationships and drainage patterns. Case studies from different basinal settings, including the Pannonian – Carpathen system, the western Mediterranean and the Norwegian continental margin demonstrate the importance of lithosphere memory in basin fill. - 18 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Forward modeling of Meandering Channelized Reservoirs Cojan, I., Lopez, S. & Rivoirard, J. Ecole des Mines de Paris, 35, rue Saint Honoré, Fontainebleau, France, cojan@ensmp.fr Ecole des Mines de Paris has developed a strong experience in the characterization of heterogeneous reservoirs: Heresim pixel-based truncated and pluri-gaussian models, Boolean object-based models, and studies of field analogs in various settings. This expertise in geostatistics and in sedimentology is used to combine both stochastic and process-based approaches in the modeling of fluvial meandering channels at the oil reservoir scale. The model is built from the spatial evolution of the channel in time, and the deposition of the associated sedimentary bodies. The different elements have been implemented taking into account physical results and case study reports. The 3D evolution of the channel stems from equations developed in hydraulic studies and proved to generate realistic 2D shapes. Developments have been added to account for the longitudinal profile. A stochastic algorithm linked with physical parameters allows to simulate chute cutoff, levee breaching and avulsions. Where appropriate, the model makes it possible to generate the different deposits: point-bars, crevasse splays, overbank alluvium, mudplugs and lowland deposits. To be operational, the model is voluntarily controlled by a limited number of key parameters. These allow to build and to test different architectures, e.g. making the sinuosity and connectivity of sand bodies varying with avulsions frequency, or reproducing external forcing (aggradation or incision). These key parameters can be inferred from seismic or well data through the use of statistics, such as Vertical Proportion Curves giving the proportions of facies along the vertical. Finally, the resulting model is simple but robust and computationally fast. Depending on a limited number of key parameters, yet capable to represent various architectures, it can be used to produce one or multiple realizations of a reservoir, to build detailed 3D blocks in order to test hypotheses on architecture, or to extract training images, virtual wells. Conditioning lithofacies or granulometry, either regionally or at well data, is to be obtained by favouring migration or avulsion of the channel in preferential areas. - 19 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Paleozoic Basin Study: Structure, Stratigraphy and Quantitative Modelling in the Southern Cantabrian Mountains, NW-Spain Dietrich, B. University of Heidelberg, Im Neuenheimer Feld 234, D-69120, Heidelberg, Germany, birgit.dietrich@urz.uniheidelberg.de A genetic model for a multi-phase structurally deformed basin is created by the authors. The investigation area is located in the Southern Cantabrian Mountains of northwestern Spain which belongs to the external part of the Variscan fold-andthrust belt. Intense thin-skinned thrusting has taken place during the Variscan Orogeny. During the Alpine Orogeny, a second deformation event occurred which locally overprinted the Variscan deformation by fault reactivation and development of out-of-sequence thrusting. However, temperatures were low. Two transects positioned in the direction of thrusting are being studied. The stratigraphy to be modelled ranges from Cambrian to Carboniferous, focussing on the Carboniferous. The Lower Paleozoic succession consists mainly of shallow marine, siliciclastic sediments. Devonian mixed clastic-carbonate sedimentation displays rapid lateral and vertical facies changes due to the development of ramps, platforms and local basins. The Carboniferous strata start with condensed successions, followed by carbonates deposited in a progressively shallowing marine environment in front of the rising Variscan Orogen. Distal turbidites of the Barcaliente Formation are succeeded by the shallow carbonate ramp setting of the Valdeteja Formation. The upper part of the ramp deposits interfingers with rapidly lateral changing flysch deposits of the San Emiliano Formation. Allo- and biostratigraphy provide data to define time lines for quantitative modelling. Due to thrusting, the complete thickness of the Carboniferous succession is unknown. Hence temperature data derived from vitrinite reflection measurements aid in the interpretation. Backward modelling is used as a dynamic, deterministic analysis of the development of accommodation space. Subsequent to the backward modelling, dynamic, stochastic forward modelling simulates sedimentation and transportation processes. - 20 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Quantified Carbonate Platform Development: the Rosengarten Transect (Middle Triassic, Dolomites) Emmerich, A. Geol.-Pal. Institute, University Heidelberg, Im Neuenheimer Feld 234, D-69120, Heidelberg, Germany, emmerich@uni-hd.de A combination of basin and stratigraphic modelling was carried out on the Rosengarten platform (Schlern Formation) in order to quantify basin dynamics and platform evolution. The essential data sets for stratigraphic reverse/forward (PHIL) and basin modelling (PetroMod) were derived from existing studies and new detailed analyses on facies architecture and thermal maturity. We identified two major stages of platform evolution – the first one with high subsidence rates and aggradational sedimentary characteristics and the second one with low subsidence rates and strong progradational characteristics. The maximum of the total subsidence was reached after the first biozone of platform growth 1100m/Ma in proximal and 1000m/Ma in distal parts. Then the subsidence slowed down to values of about 50m/Ma to 150m/Ma. An explanation for the sudden stop in subsidence is the coeval development of a huge magmatic chamber in the area of Monzoni/Predazzo neighbouring the Rosengarten. Constant carbonate production rates between 750m/Ma (platform slope) and 850m/Ma (platform top) were determined by sequence stratigraphic forward modelling. These rates proved to be sufficient to keep platform growth up with sealevel rise during times of high subsidence and to produce fast progradation in the second stage of platform evolution. The mentioned modelling results are corroborated by the porosity evolution of the sediment package below the slope mimicing the N-S trend of slope progradation. Values of the regional vitrinite reflectance pattern in the underlying Permian lie around 0.55%VRr. The temperatures during maximum burial (presumably during Late Cretaceous times) most probably did not exceed 100°C. However, samples from the Rosengarten area neighbouring the Monzoni volcanic centre reveal values between 0.7 and 0.8%VRr indicating that the thermal influence of the Monzoni event was restricted to its nearest vicinity. The next step of our investigations will be pinpointing the timing of subsidence with the help of fission track analyses in apatites. - 21 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Incorporation of numerical and physical forward modelling into uncertainty analysis methods Feurer, J., & Van der Zwan, C.J. SIEP-SEPTAR, EPT-AEP, P.O.Box 60, 2280 AB Rijswijk, Jeurg.Feurer@SHELL.com NL. Kees.VanderZwan@SHELL.com, Hydrocarbon occurrences rely on the concurrent presence of a valid structure, hydrocarbon charge, a reservoir and a seal. The Oil Industry invests millions in the predicting of these, in technology developments to improve prediction, and many more millions in dry holes. Shell has a long tradition in Monte Carlo simulation to manage subsurface uncertainty (e.g. uncertainty of porosity, closed rock volume, etc), and is increasingly betting on the horse of forward modelling. Although lithological predictions integrate various well-established disciples (e.g. conventional and sequence stratigraphy, seismic stratigraphy, etc), the qualitative and quantitative forecasts remain variably accurate and, at times, simply fall short on consistency. Forward modelling is an effective means of improving both: the consistency by encouraging geoscientists to synthesize the (limited) data into one or more subsurface models, and the accuracy by simulating actual physical processes. To date forward-modelling tools have remained largely deterministic. This bears the inevitable consequence of undersampling the uncertainty space and deriving suboptimal statistics. The power of today’s computers is now sufficient to warrant the development of probabilistic forward-modelling tools capable of processing multiple alternative subsurface models. In this presentation an overview will be given of current probabilistic oil industry methodologies and how these models can be linked to best-fit and scenario-based forward reservoir models. - 22 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Role of fluvial transport during orogenesis: New numerical and analogue modelling techniques Daniel García-Castellanos, Katarina Persson, Dimitrios Sokoutis, Ivone Jimenez-Munt Free University Amsterdam Erosion is a main process controlling the tectonic evolution of orogens, effecting on their internal structure. Studies of the interplay between large-scale tectonics and surface transport usually assume 2D (cross section) settings, whereas river networks drive surface transport typically in the axial direction (parallel to the orogens). Do the spatial asymmetries introduced by fluvial transport influence on the tectonic deformation pattern during continental collision? The purpose of this work is to evaluate the 3D interaction between lithospheric deformation and fluvial transport during continental collision. For this purpose, we present results of two new modeling techniques coupling a fluvial transport numerical model (stream power approach) with a (left) numerical model of lithospheric viscous deformation (thermally coupled thin-sheet approach) and a (right) analogue sandbox model of upper-crustal deformation. The Numerical Model fully couples a thermo-mechanical model of lithospheric viscous deformation and fluvial erosion/deposition. The mutual feeback’s effects between these processes are accounted for. The hybrid Numerical-Analogue Model fully couples the dynamics of an analogue deformation model (sand-box) and the fluvial transport numerical model. Both components are linked via a scanner. The results show that surface erosion, transport and sedimentation have a significant effect on the distribution of lithospheric and crustal deformation during orogenesis. The axial transport along the river network can induce lateral changes on the tectonic evolution of orogen-basin systems, such as diachroneity. These effects are probably less important than those induced by inherited heterogeneities and lateral changes in the tectonic forcing. - 23 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 A strategy for automated inversion of process-based models Geel, C.R.& Weltje, G.J. Department of Applied Earth Sciences, Delft University of Technology, Mijnbouwstraat 120, 2628 RX Delft, The Netherlands, c.r.geel@ta.tudelft.nl Application of process-based models in production geology has been hampered by the difficulty of conditioning such models to well data. Several causes can be indicated, of which irreversibility is the most important. The occurrence of erosion surfaces prevents the application of a direct, simple inversion. Instead, a trial-anderror method or an automated goal-seeking algorithm must be employed. The fact that most process-based models are CPU-intensive prohibits extensive trial-and-error or automated goal-seeking approaches. Another fundamental problem is the choice of an appropriate objective function, i.e. a criterion for quantifying the discrepancy between a certain realisation and the well data. The ideal objective function should be insensitive to local minima, while its value should be interpretable in probabilistic terms. As these two aspects are difficult to reconcile, we have focused on the first requirement. In this paper we present a method that circumvents many of the above problems. We use a very fast forward model, which can produce hundreds of realisations in just a few minutes. At the same time, the model shows a sufficiently rich behaviour to mimic the first-order dynamics of a shallow-marine environment. The data to be matched are derived from permeability logs, as this quantity captures most features of a given grain-size distribution in a single number. Permeability logs are also readily available for oil wells, while quantities like grain-size distribution or (litho) facies are usually not available. We use a modified Karman-Cozeny equation to analytically determine permeability from grain-size distribution and porosity. The latter is derived empirically from measurements on a series of lognormally distributed coastal sands. Characteristic permeability-log shapes are converted into character strings, from which the objective function value can be calculated. The objective function itself is the Levenshtein distance widely used in spell checking, speech recognition, DNA analysis, and plagiarism detection. A relatively straightforward genetic algorithm is used to minimise this objective function. The method was tested on a several (synthetic) cases with complicated, nontrivial geometries. Up to seven parameters related to oscillations of sea level and sediment input could be inverted with a maximum error of 20% of their true value within 45 minutes using an ordinary desktop PC. A typical seven-parameter inversion requires about 50,000 iterations. - 24 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 3D simulation of sedimentary facies: application to Langhian reef buildups in Vallès-Penedès basin (Catalunya). Gratacos, O. Faculty of Geology, University of Barcelona, C/ Martí I Franqués s/n, 08028, Barcelona, Spain, gratacos@geo.ub.es Prediction of sedimentary facies and geometric architecture of sedimentary basin fills through process-oriented numerical models is becoming an important and useful tool in geological studies. Such information provides insight into spatial distribution of physical, chemical and hydraulic properties within sedimentary deposits. To date, various sedimentary modelling programs have been developed, most of them considering either clastic material or carbonate sediments, but very few models have attempted to consider interplaying, mixed carbonate-clastic sedimentation. Moreover, most programs are designed in two dimensions in space, which is a major restriction as geologic processes work in three dimensions creating a great variety of sedimentary facies types distributed in a complex 3D architecture of sedimentary bodies. In this contribution, we present a 3D forward mathematical simulation (SIMSAFADIM) that simulates clastic transport and sedimentation as well as processes of carbonate production, transport and sedimentation in three dimensions. We test the predictive capability of the model by simulating a marine transgression during Langhian time in the Vallés-Penedés basin (Catalunya, NE Spain) with a significant interplay between carbonate and terrigenous deposition. - 25 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Stratigraphic Forward Modelling of deep marine systems using Sedsim Griffiths, C. CSIRO, Australia, cedric.griffiths@csiro.au Sedsim has evolved to be able to model many of the features of submarine and lacustrine gravity-flow deposits including sandy basin-floor fans, channel-levee systems, ponded turbidites and debris flows. Examples are shown from modern and ancient systems, together with explanations of the modelling assumptions. - 26 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 The tectonic evolution of the northern North Sea from the Mid-Jurassic form 2D subsidence analysis Hanne, D. University of Cambridge, Wiedhof 11, D-31249, Harber, Germany, dtlef.hanne@yahoo.com Nine seismic lines have been investigated across the Viking Graben. From 2D subsidence analysis, Pseudo-3D images of strain rate distributions in the Graben in 10 Ma intervals and estimates of the amounts of stretching have been developed. A Mid-Jurassic and Late-Cretacous rift event (in the North of the Graben) were identified. - 27 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Fine-scale forward modelling - example from a Devonian reef of the Canning Basin Hasler, C.-A.¹ , Adams, E.W. 2, Wood, R.A.3, Dickson, J.A.D. University of Cambridge, Downing Street, CB2 3 EQ, Cambridge, United Kingdom, chas02@esc.cam.ac.uk MIT 3 Schlumberger Cambridge Research, Cambridge, United Kingdom 1 2 The spatial prediction of carbonate reef bodies within reservoirs - as well as that of carbonate geometries in general - has proved difficult despite many years of research by oil companies. Forward modelling at a fine scale may be a tool to better understand and predict the internal discontinuities within the reef bodies. The modelling is based on the data collected using a very accurate 3D differential GPS tools on several outcrops from the Devonian reef of the Canning Basin (Western Australia). The modelling processes reconstitute the environmental conditions such as the water-depth, the currents, the turbidity and the luminosity on the sea-floor in order to predict the lateral distribution of the major protagonists of the Devonian Reef Complex (stromatopororoids, stromatolites, tabulate corals, microbialites, coarse/fine sediments, micrite?) The vertical evolution is constrained by the eustatic evolution and by the ability of all the protagonists to grow and/or produce sediments. - 28 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling turbidity currents by CFD simulations Heimsund, S.1, 2, Hansen, E.W.M.3, Baas, J.H.4 & Nemec, W.1 1 Department of Earth Sciences, University of Bergen, 5007 Bergen, Norway snorre.heimsund@student.uib.no 3 Complex Flow Design AS, P.O. Box 1273, 7462 Trondheim, Norway 4 Department of Earth Sciences, University of Leeds, Leeds LS2 9JT, U.K. 2 E-mail: In this study, laboratory-scale turbidity currents have been simulated by a numerical approach known as computational fluid dynamics (CFD). Although relatively new in the field of sedimentological research, this method is widely applied in hydraulic engineering. CFD refers to the simplified numerical solution of the governing equations describing fluid flow, namely the continuity, momentum and energy equations. These are the mathematical formulations of the fundamental physical principles of fluid dynamics: the mass and energy are conserved, and the Newton second law is obeyed. For a viscous flow, they are known as the Navier-Stokes equations. The commercial CFD software package called FLOW-3D has been used in this study to perform numerical simulations of turbidity currents and to assess the validity of this deterministic method. FLOW-3D subdivides the flow region into a grid of variable-sized rectangular cells, with finite-difference (or finite-volume) approximations used to compute (explicitly or implicitly) and monitor the spatial and temporal changes of the basic flow values for each cell. The turbidity current is simulated using a drift-flux technique, a two-phase model that describes the relative flow of two miscible fluids with different densities. For viscosity evaluation (turbulence closure) a renormalized group theory (RNG) model is used; employed are also auxiliary models for gravity, pressure, density, shear stress, particle-fluid interaction and the advection, erosion, settling and deposition of sediment. A 3-D model has been constructed to imitate the laboratory ‘delta’ flume facility at the Utrecht University. A number of numerical experiments have been performed, focusing on the mean flow properties and flow structure of the whole current, including also the effects of topography, erosion and deposition. The relative sensitivity and importance of the individual flow parameters have been assessed with respect to variation in the initial conditions and seafloor configuration. A comparison of the numerical results with the Utrecht laboratory data shows their good correspondence, which indicates that the CFD models give realistic results and may be quite suitable for the simulation of turbidity currents. - 29 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Forward Modelling & Prediction of Aeolian Systems Using Fuzzy Logic, Constrained by Data from Recent and Ancient Analogues Hern, C.1, Nordlund, U.2, Zwan, C.J. van der1 & Ladipo, K. 3 1 Shell SEPTAR, P.O.Box 60, 2280AB Rijswijk, The Netherlands, caroline.hern@shell.com se-74020 Vänge, Sweden 3 PDC, Port Harcourt Nigeria 2 Herrboda, Aeolian reservoirs yield some of Shell‘s largest gas fields, such as NAM’s Groningen field. Although, aeolian reservoirs have been studied for many years, there is still room for improvement in the prediction of such reservoirs. With this intention, a project was initiated to evaluate the factors influencing aeolian systems, and deveolp a forward model using ‘fuzzy logic’. The key aims were to: predict the type, amount and distribution of major facies types in aeolian systems and, to compare resultant models with the regional-scale architecture Fuzzy rules and sets, which define the behaviour of aeolian systems, were constructed and used to modify the pre-existing fuzzy modelling software. The modelling procedure was validated by using input data appropriate to the Rotliegend climate and comparing the resulting models, in terms of thicknesses and spatial distribution of facies types, to well data from the Upper Rotliegend interval of the Lauwerzee Trough area, NE Netherlands. - 30 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Using sub-regional scale forward models to condition reservoir-scale stochastic scenarios Hern, C., Lammers, H., Burgess, P. & Nijman, M. Shell SEPTAR, P.O.Box 60, 2280AB Rijswijk, The Netherlands, caroline.hern@shell.com Reservoir scale 3D static models are constructed based on available well and seismic data, yet in many cases well data is sparse and seismic resolution is poor. This lack of well or seismic control in the area of interest may result in an unconstrained and unrealistic population of the 3D volume. In order to honour more truthfully the geology, one approach may be to condition the reservoir scale model to a regional scale numerical forward model. Advances in predictive numerical forward modelling have enabled construction of models constrained by available well data that reproduce and predict realistic geological architectures at a number of scales. We show how a sub-regional scale (37km long) forward model of the Grassy Member of the Blackhawk Formation (Utah), constructed in Dionisos, was used to condition a reservoir scale hybrid stochastic model. The forward model was clipped to extract a volume representative of a small reservoir (3 x 2km). The gross distribution of facies associations indicated in the forward model was used as a background in the stochastic model, into which shale objects were distributed. We compare the resultant hybrid model with detailed 3D models built and conditioned to synthetic wells derived from the outcrop panels. The methodology illustrates how the forward predictive capability of Dionisos models at a sub regional scale, can be harnessed and utilised in conjunction with reservoir modelling packages to fully describe facies scale variation at the reservoir scale. - 31 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Investigation of gravitational mass transport processes - using the Distinct Element Method as a modern tool in sediment transport modelling Huhn, K. & Kock, I. Research Center Ocean Margins, University Bremen, Am Fallturm 1, D-28334, Bremen, Germany, khuhn@uni-bremen.de High-production regions, i.e. the NW African continental slope offshore Mauritanian, are characterized by high sedimentation rates at the upper slope. These materials are episodically transported down-slope in form of slides of intact layered packages or by mixed and disordered debris flows and turbidites. The main purpose of our studies is to investigate the influence of different key parameters for (a) slope failure as well as (b) deformation and kinematic behaviour of gravitational mass flow processes. For our studies we use a numerical particle-based method - the Distinct Element Method (DEM). This technique enables a wide range of material parameters and model configurations. DEM simulations have been proved as an efficient tool to quantify the influence of material parameters, i.e. basal as well as internal friction, and geometrical settings for geodynamical processes like gravitational mass flows. This method enables detailed information about mechanics and kinematics of downslope processes, mass transfer patterns, as well as internal and morphological structures. Nevertheless, we have to investigate the influence of model configurations and assumptions in the face of transferability of model results into a natural system. Thus, we develop a 2d DEM model which is built up of spherical, elastic frictional particles. These spheres interact according to nonlinear-elastic Hertzian contact laws. The total number of particles in each experiment is constant to simulate identical sedimentation rates. First, we investigate the influence of deposition of sediments. Therefore, in experiment (a) all spheres settle down under gravity in a triangular shaped box on top of the upper slope. Removing the box boundaries releases a gravitational mass flow. In experiment (b), all particles are settled under gravity on top of a shelf surface and they are transported progressively by a bulldozer to the upper slope. In both experiments, we change iteratively the basal friction on top of the slope as well as internal friction of sediments. A decrease of basal friction as well as of internal friction of sediments causes always an increase of run out length of the flow while decreasing the thickness. In addition, an increase of internal friction of the sediment causes evolution of a steeper upper slope in experiment (b) and an increase in slope stability is observed. Thus, these experiment settings produce much more natural conditions. - 32 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling cycles of fluvial aggradation and degradation using a processbased alluvial stratigraphy model Karssenberg, D.1, Bridge, J.S.2, Stouthamer, E.1, Kleinhans, M.G.1 & Berendsen, H.J.A.1 1 The Netherlands Centre for Geo-ecological Research (ICG), Faculty of Geographical Sciences, Utrecht University, PO Box 80115, 3508 TC Utrecht, the Netherlands, d.karssenberg@geog.uu.nl 2 Department of Geological Sciences, Binghamton University, PO Box 6000, Binghamton, New York 139026000. We have developed a prototype process-based alluvial stratigraphy model with model run times representing time spans on the order of 103-105 years and a temporal resolution of 1 year. The model represents a river valley and coastal plain with a lateral spatial resolution on the order of 1-10 km2, defined by grid cells with a constant size, and a vertical resolution of less than 10-2 m. Sediment is transported by diffusional transport through a network of channels with a sediment flux boundary condition at the upstream point of entry and a base level boundary condition at the downstream edge of the model. Both boundary conditions can be varied in time. Each channel has a channel belt with a width that increases in time at an exponentially decreasing rate. The network of channels evolves as a result of channel bifurcation and channel abandonment. The timing and location of channel bifurcation is controlled stochastically as a function of the cross-valley slope of the floodplain adjacent to the channel belt relative the down valley slope. Channel abandonment, resulting in a bifurcation developing into an avulsion, occurs when the discharge of one of the distributaries falls below a threshold value. Floodplain aggradation occurs adjacent to actively aggrading channel-belts on floodplain surfaces that do not exceed the elevation of the active channel belt. Floodplain aggradation rate decreases exponentially with distance from the active channel belt. Channel-belt degradation results in floodplain incision and transport of eroded sediment from valley walls towards the incising channel belts. The erosion of valley walls is calculated as a function of stream power for each location on the valley wall. Compaction of channel-belt and floodplain deposits is modelled as a function of depth of burial. Tectonic movement can also be simulated by faulting and/or tilting of the floodplain in any direction. The model is programmed in a standard spatio-temporal modelling language that uses built-in functions and that allows easy addition of new model components. The model can be run in Monte Carlo simulation mode, such that many model runs are made to evaluate the effects of uncertainty in model input parameters. The model is still being developed and tested against real-world data. However, examples are given of the simulated effects of sea level change on rivers, such as: evolution of the channel network; the upstream migration of knick points; formation of incised channels and river terraces; and the effects of aggradation-degradation cycles on patterns of avulsion and alluvial architecture. - 33 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Channelised turbidity currents: new insights into flow structure and secondary circulation Keevil, G. School of Earth Sciences, Leeds University, LS2 9JT, Leeds, United Kingdom, g.keevil@earth.leeds.ac.uk Sinuous submarine channels are major morphological features which act as conduits for the transport of clastic sediment into the deep ocean. Such channels have long been considered to be analogous to meandering fluvial channels. These comparisons are commonly driven by similarities between the planform geometries of submarine and fluvial channels. Recent laboratory studies have revealed much about the velocity and density structure of unconfined turbidity currents. The velocity structure and pattern of secondary circulation (helical vortices with the long axis aligned parallel to flow) within meandering fluvial channels is also well understood. However, prior to this work, very little was known about the velocity and turbulence structure of channelised turbidity currents. Results from a series of new experiments will be presented where dense saline flows were pumped into submerged meandering channel sections. Ultrasonic velocity probes (UVP) provided high-resolution two dimensional flow fields. This has for the first time allowed the quantification of the velocity profile of a channelised turbidity current within a sinuous channel. Additionally these experiments have enabled the visualisation and quantification of secondary cells within these channels. The results clearly demonstrate a strong secondary flow cell which is strongest at the bend apex. The basal component of the flow cell moves from the inside to the outside of the bend, the reverse of what is expected within a fluvial channel. Such a finding has great significance when assessing the validity of the comparison between meandering fluvial channels, where the secondary flow cells influence the migration and evolution of bends, and sinuous submarine channels. It implies that different processes control the morphological evolution of sinuous submarine channels. - 34 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Predicting discharge and sediment flux of the Po River, Italy since the LGM Kettner, A.J. & Syvitski, J.P.M. Environmental Computation and Imaging Group, Institute of Arctic and Alpine Research (INSTAAR), University of Colorado, Boulder CO, 80309-0450, United States, Kettner@colorado.edu INTRODUCTION The EuroSTRATAFORM project aims at evaluating the influences of physical characteristics of sediment sources and their temporal variability due to climatic evolution and human impacts on continental-margin sediment flux. In the framework of the project a number of rivers were chosen to simulate; among others, the Po. A numerical climate-driven hydrological model, HydroTrend, is applied to the heavily anthropogenic influenced Po basin (drainage area 77,000 km2) in Northern Italy to predict sediment fluxes and water discharge to the Adriatic Sea since the Last Glacial Maximum (LGM). MODEL CONCEPTS OF HYDROTREND To simulate discharge and sediment discharge fluxes, HydroTrend incorporates drainage basin properties (river networks, hypsometry, relief, reservoirs) based on high-resolution digital elevation models (for example HYDRO1k DEM), along with other biophysical parameters (basin-wide temperature, precipitation, evapotranspiration, canopy, soil depth, hydraulic conductivity, glacier characteristics). A stochastic model (Morehead et al., 2003) is used to calculate the daily suspended sediment load fluxes: Qs Q Q Qs C (1) Wherein Qs is the daily suspended sediment discharge (kg/s), Q the daily discharge (m3/s), Q s the long-term average of Qs, Q the long-term average of Q, a lognormal random variable and c a normal random variable. The long-term average of Qs is defined as: Qs 3 A R e k T 4 (2) 5 Wherein A is the drainage basin area (km2), R the maximum relief (m), T the basinaverage temperature (C), 3, 4, 5 and k dimensionless coefficients which depend on climatic zone based on the geographical location of the drainage basins (Syvitski et al., in press b). studying the case of the Po river, climate changes from Last Glacial Maximum (LGM) until the present do not reach the threshold values to affect - 35 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 changes in ’s or k. The Long-term average suspended sediment will only be affected by changes in A, R and T . The daily bedload Qb (kg/s) is simulated using a modified Bagnold (1966) equation: s gQ seb Qb s g tan f when u ucr (3) Where s is sand density (kg/m3), is fluid density (kg/m3), Q is the daily discharge (m3/s), s is the slope of the riverbed (), eb is the bedload efficiency, is a dimensionless bedload rating term, f is the limiting angle of repose of sediment grains lying on the river bed (), u is stream velocity (m/s), and ucr is the critical velocity (m/s) needed to initiate bedload transport. The model with documentation, example files and references is available on the web: http://instaar.colorado.edu/deltaforce/models/hydrotrend.html. CALIBRATION TEST, PRESENT DATA VERSUS MODELED Present-day climate statistics including monthly mean temperature and precipitation and their standard deviations of the Po drainage basin are retrieved from daily data records (1977 - 1991) of 20 climate stations weighted in terms of influence by elevation. The yearly mean temperature, precipitation and their standard deviations also weighted in terms of influence by elevation are obtained from monthly data records (1760 – 1995) of 13 climate stations from the Global Historical Climatology Network of National Oceanic and Atmospheric Administration, (NOAA) (Vose et al., 1997) (http://ingrid.ldeo.columbia.edu/SOURCES/.NOAA/.NCDC/.GHCN/). HydroTrend verifies as an appropriate simulation model for the gauging station closest to the river mouth before being split into distributary channels. At this delta apex there is a high correlation between modeled discharges and time series of daily discharge data (1990-2001) and monthly discharge (1918-present). HydroTrend also generates similar discharge variability as compared to the 12-year daily measured discharges (fig. 1). Modeled data correlates significantly with the observed data (r2 of 0.72), as calculated from 100m3/s intervals. Figure 2 indicates that the monthly mean discharges are underpredicted during the winter months (November till March) and overpredicted during the summer (April till September). This deviation relates to the well known 'reservoir' effect common in river basins containing major reservoir lakes (Bobrovitskaya, 1996), and the discharge controlled management for hydro-power. Over a third of the discharge of the Po river discharge is affected by the hydro-power reservoirs (Camusso et al., 2001). - 36 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 measured 2500 1552 1532 1712 1500 741 1000 534 Number of events 2000 1912 modeled 0 2 11 00 0 3 3 90 00 - 10 00 0- 90 00 80 00 - 70 00 -8 0 00 60 00 -7 0 00 50 00 -6 0 00 40 00 -5 0 00 30 00 -4 0 00 20 00 -3 0 00 10 00 -2 0 00 0- 1 00 0 0 10 00 0 2 1 15 4 34 17 39 27 69 69 237 253 500 3 Discharge ranges (m /s) Fig. 1 Discharge distribution of12yr daily mean measured versus modeled discharge. Measured Modeled 3500 3000 Q (m3/s) 2500 2000 1500 1000 500 De cem be r No vem be r be r Oc to be r Se pte m Au gu st Ju ly Ju ne Ma y Ap ril Ma rc h Fe br u ar y Ja nu a ry 0 Time (Months) Fig. 2 Monthly mean discharge distribution based on 100yr monthly measured versus modeled data. The dashed lines indicate mean discharge values plus or minus their standard deviation. The arrows show the huge impact the reservoirs have on the monthly mean discharge distribution. - 37 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 The Po’s discharge is strongly affected by its 5 large reservoirs, sediment fluxes will likely be influenced even more. HydroTrend to some extent can be adjusted to incorporate reservoir effects on sedimentation. The model simulates trapping efficiency TE depending on the reservoir volume either by the Brown equation, for reservoirs smaller or equal than 0.5 km3, or the modified Brune equation by Vörösmarty (1997), for reservoirs larger than 0.5km3 (formula 4 and 5). TE 1 0.05 (4) Wherein Δτ is the approximated residence time and is estimated by: ni V i (5) 1 Qj Where Vi is operational volume of the reservoir i, Q is the discharge at mouth of each regulated subbasin j. Simulations show a TE of 20.4% of suspended sediment load for the entire Po drainage basin. At the delta apex, HydroTrend predicts an average discharge of 1,542 m3 s-1 and peak discharges comparable to the measured floods of 1951 and October 2000. Based on the formulas 1, 2 and 3, average suspended sediment load of 16 x 106 t yr-1 with peak years of 39 x 106 t yr-1 are predicted. Bedload contributes only ~2.5% of the total sediment output of the Po river system (table 1). There is a good correlation between bedload transport and high suspended sediment concentrations, figure 4. Table 1. Characteristics of the Po River, measured values against a 100yr simulation. Measured at Modeled apex (Pontelagoscuro) Long-term average discharge (m3 s-1) a 1,500 1,542 3 -1 Last century floods (m s ) 10,300 10,281 9,600 10,110 8,700 9,779 Average suspended sediment load (t 15 x 106 16 x 106 -1 b yr ) Peak suspended sediment load (t yr-1) c 35 x 106 39 x 106 Average bedload (t yr-1) --4.0 x 105 Sediment yield (t km-2 yr-1) b 201 207 a) Nelson, 1970 b) Cattaneo et al., 2003 c) Friend et al., 2002 - 38 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 4.5 4 ln Qb (kg/s) 3.5 3 2.5 2 1.5 1 0 2 4 6 8 10 12 14 ln Qs (kg/s) Fig 4. 100yr daily simulation of the Po, suspended sediment discharge (Qs) against bedload (Qb). LGM VERSUS PRESENT SIMULATIONS Adriatic sea level was about 120 meter below present during the LGM (Fairbanks, 1989). An estimate of the Po basin area during LGM was made based on present-day DEM and bathymetry data (Sandwell and Smith, 1999). The shoreline was about 350km SE of the present-day Po outlet, because of the shape of the relatively shallow Adriatic Sea. This implies that part of the Apennine rivers, for example the Metauro, Potenza and the Chienti, as well as present-day Northern Italian and Croatian rivers were contributaries to the Po basin which eventually drained into the modern foredeep basin of the Apennine chain. This shift in the coastline position increased the total area of the Po drainage basin by approximately a factor of 1.9 compared to its present estimated area (fig 5). The Alps were almost completely covered by the late-Würmian ice-sheet, but the ice cap did not add significant area to the Po basin as the glacier drainage divide roughly followed the basin divide (Baroni, 1996). For this 100 year LGM study we assume no changes in the late-Würmian ice-sheet area. The monthly climate statistics since the LGM, 21kyBP, are retrieved from the Community Climate Model1 (CCM1) (Kutzbach et al., 1998). Using the model is justified by Peyron et al. (1998) who pointed out that the cool steppe vegetations is coherent with simulations performed using the CCM1 model during the LGM. According to the model, mean annual temperature is 3 times lower (11.5°C vs. 3.5°C) and precipitation increases by 25%. A 100 year HydroTrend simulation indicate that the average discharge increases with a factor two, to 3045 m3 s-1. Bedload is closely related to the discharge (see formula 3) inducing an increase; present bedload of 4.0 x 105 t yr-1 versus 8.2 x 105 t yr-1 during the LGM (see fig. 6). The discharge distribution is more fluctuating during the LGM even though under - 39 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 modern climate the larger area would bring about less fluctuating discharge. The increase in bedload implies coarser total sediment flux during LGM compared to present sediment fluxes. Fig 5. Po drainage basin area at present (black, 77,000km2 with a river length of 670km), versus hypothetical drainage basin area at LGM (gray, 148,900km2 with a river length of 1250km). There appears to be no to little change in the average suspended sediment load (16.1 x 106 t yr-1 nowadays against 16.2 x 106 t yr-1 during LGM) though suspended sediment fluxes appear to be less fluctuating during LGM compared to present simulations (fig. 7). This is remarkable, because the nowadays decoupled small Apennine rivers (Metauro, Potenza and the Chienti) have sediment yields of ~1,400 t km-2 yr-1 and go hyperpycnal every 2 to 4 year. Qbedload present Q Present Qbedload LGM Q LGM 120 9.0E+05 110 8.0E+05 100 7.0E+05 90 6.0E+05 80 5.0E+05 70 4.0E+05 60 3.0E+05 50 2.0E+05 40 1.0E+05 30 0.0E+00 20 1 11 21 31 41 51 61 Number of modeled years - 40 - 71 81 91 Q (km3 /yr) Qb (t/yr) 1.0E+06 Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Fig. 6. Bedload – discharge comparison between present s and LGM simulation. Present conditions LGM conditions 4.50E+07 4.00E+07 3.50E+07 Qs (t/yr) 3.00E+07 2.50E+07 2.00E+07 1.50E+07 1.00E+07 5.00E+06 0.00E+00 0 10 20 30 40 50 60 70 80 90 Number of modeled years Fig. 7. A comparison between present yearly suspended sediment discharge versus LGM suspended sediment discharge. These simulations indicate a dynamic system under strongly changing boundary conditions. HydroTrend provides a method to quantify the effects of these dynamic boundary conditions and as such will generate input estimates for other modeling efforts. Future study on Po sediment fluxes will focus on a time continuous simulation since LGM. REFERENCES Asioli, A., Trincardi, F., Lowe, J.J., Ariztegui, D., Langone, L. and F. Oldfield, 2001. Sub-millennial scale climatic oscillations in the central Adriatic during the Lateglacial: palaeoceanographic implications. Quaternary Science Reviews 20, 1201-1221. Bagnold, R.A., 1966. An approach to the sediment transport problem from general physics; U.S. Geological Survey Professional Paper 422 (1), 1-37. Baroni, C.,1996. The Alpine “Iceman” and Holocene Climatic Change. Quaternary Research 46, 78-83. Bobrovitskaya, N.N., Zubkova, C. and R.H. Meade, 1996. Discharges and yields of suspended sediment in the Ob' and Yenisy Rivers of Siberia. In: Walling, D.E. and B.W. Webb (eds) Erosion and Sediment Yield: Global and Regional Perspectives. Intern. Association of Hydrological Sciences Publ. no. 236, 1996. Camusso, M., Balestrini, R. and A. Binelli, 2001. Use of zebra mussel (Dreissena polymorpha) to assess trace metal contamination in the largest Italian subalpine lakes. Chemosphere 44, 263-270. Cattaneo, A., Correggiari, A., Langone, L and F. Trincardi, 2003. The late-Holocene Gargano subaqueous delta, Adriatic shelf: Sediment pathways and supply fluctuations. Marine Geology 193, 61-91. Fairbanks, R.G., 1989. A 17,000 year glacio-eustatic sea level record: Influence of glacial melting rates on the Younger Dryas event and deep-ocean circulation. Nature 342, 637-642. - 41 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Friend, P.L., Amos, C.L., Panin, N. and F. Trincardi, 2002. Sediment supply and river discharge to the continental shelf – A synthesis of existing data for the Ebro, Rhone, Po and Danube river systems. Abstract of Joint European/North American EUROSTRATAFORM meeting, incorporating EURODELTA and PROMESS, September 2002. Kutzbach, J., Gallimore, R., Harrison, S., Behling, P., Selin, R., and F. Laarif, 1998. Climate and biome simulations for the past 21,000 years Quaternary Science Reviews 17, 473-506. Morehead, M.D., Syvitski, J.P., Hutton, E.W.H. and S.D. Peckham, in press 2003. Modeling the temporal variability in the flux of sediment from ungauged river basins. Global and planetary change 815, 1-17. Nelson, W.N., 1970. Hydrography, sediment dispersal, and recent historical development of the Po river delta, Italy. In: Morgan, J.P. (eds) Deltaic Sedimentation; Modern and Ancient Society of Economic Paleontologists and Mineralogists. Special Publication No. 15, November 1970. Tulsa, Oklahoma, USA. Peyron, A., Guiot, J., Cheddadi, R., Tarasov, P., Reille, M., Beaulieu, J.L. de, Bottema, S. and V. Andrieu, 1998. Clomatic Reconstruction in Europe for 18,000 YR B.P. from Pollen Data, Quaternary research 49, 183-196. Sandwell, D.T. and W.H.F. Smith, 1999. Bathymetric estimation. In: Satellite Altimetry and Earth Sciences, Academic Press, 1999. Syvitski, J.P.M., Kettner, A.J. and S.D. Peckham in press a. Predicting the Flux of Sediment to the Coastal zone: Application to the Lanyang watershed, northern Taiwan. Journal of Coastal Reserach. Syvitski, J.P.M., Peckham, S.D., Hilberman, R.D. and T. Mulder in press b. Predicting the terrestrial flux of sediment to the global ocean: A planetary perspective. Marine Geology. Vose, R.S., Schmoyer, R.L., Steurer, P.M., Peterson, T.C., Heim, R., Karl, T.R. and J. Eischeid, 1992. The Global Historical climatology Network: long-term monthly temperature, precipitation, sea level pressure, and station pressure data. ORNL/CDIAC-53, NDP-041. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee. Vörösmarty, C.J., Meybeck, M., Fekete, B. and Sharma, K., 1997. The potential impact of neoCastorization on sediment transport by the global network of rivers. Human Impact on Erosion and Sedimentation. IAHS Publication no. 245, 1997. - 42 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling bedding and petrophysical characteristics of tidal heterolithic reservoirs using a process-oriented approach Martinius, A.W. Statoil R&D,Arkitekt Ebbelsvei 10, Rotvoll, N-7005, Trondheim, Norway, awma@statoil.com This poster will present the results of five-year R&D programme to model the true 3D spatial architecture and petrophysical properties of tidal deltaic heterolithic deposits. Early work focssed on correct implementation of sedimentological processes, including empirical results from flume tank experiments, in a mathematical framework representing bi-directional deposition and erosion of a heterolithic sand-mud system in a tidal deltaic environment. The modelling tool, SBEDTM, has been successfully used to estimate effective flow properties via calibration of models against observed sub-surface well data. As well as giving a valuable insights into the nature of sedimentological processes (e.g. bedding architecture and bedding cyclicity) the modelling method has helped us to unravel the nature of flow properties in these heterolithic systems. This has resulted in an effective permeability prediction function incorporating effective flow and percolation theory. - 43 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Tectonic control on past circulation of the Mediterranean Sea Meijer, P.1, Slingerland, R.2 & Wortel,R.1 1 Faculty of Geosciences, Utrecht University, Budapestlaan 4, 3584 CD, The Netherlands, meijer@geo.uu.nl of Geosciences, Penn State University, USA 2Department We examine the effect of Late Miocene paleogeography on the circulation and water properties of the Mediterranean Sea by using an ocean general circulation model. Results obtained for the past are compared to a control experiment with the presentday geometry. To focus on paleogeography, atmospheric forcing is always based on the present-day climatology. We seek insight that allows us to test ideas based on observations and to formulate new working hypotheses. The Late Miocene represents an important stage in the evolution of the Mediterranean. The present-day model reproduces the main aspects of the surface to intermediate depth circulation and water properties. The model does not capture the deep circulation known to occur at present in both subbasins. When the subbasins are reconstructed to their Late Miocene shape (keeping intervening sills at present-day levels) the overall nature of the surface/intermediate depth circulation proves unaffected. The model, however, predicts intense deep circulation in the eastern Mediterranean – most likely due to the greater surface area of the reconstructed Adriatic Sea. Using the first paleoexperiment as a starting point several additional paleogeographical aspects are examined. - 44 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Numerical Modeling of the Sedimentary Systems of Large-scale Coastal System Tracts Niedoroda, A.W., Reed, C.W.1, Das, H.1, Donoghue, J.2, Ferraghazzi, S.2 , Wang, Z.B.3 & Stive, M.4 URS Corp., 3676 Hartsfield Road, Tallahassee, FL, 32303. USA. alan_niedoroda@urscorp.com Geology Department, Florida State University, Tallahassee, FL, 32306. USA Jdonoghu@mailer.fsu.edi 3 Marine and Coastal Management, WL/Delft Hydraulics, 2600 MH Delft. The Netherlands. Zheng.Wang@wldelft.nl 4 Delft University of Technology, Stevinweg 1, Room 3.69, 2600 GA Delft. The Netherlands. M.J.F.Stive@ct.tudelft.nl 1 2 The Coastal System Tract concept (Cowell et al., in press) expresses the importance of sediment-sharing processes between a variety of individual system elements in a given coastal region. The surf zone/beach system, the shoreface/shelf system and the inlet/estuary system are all examples of the individual system elements. There are many others examples dictated by the scale and nature of the specific problem undergoing analyses. The CST-Model has been developed to provide a quantitative tool that represents the complex interactions between arbituary combinations of the sediment-sharing elements. This is a large-scale behavior-oriented 2-DH model, based on representations of the physics of the macroscopic morphodynamic processes. The model operates on an inter-annual time scale so that it can resolve trends in system-wide morphology over periods of decades to millennia. It is capable of resolving various combinations of river sources, estuaries and lagoons, flood- and ebb-tide shoal complexes, and inlets in concert with a littoral surf zone, shoreface, continental shelf and upper continental slope. The shoreline translates according to the dictates of sediment transport gradients. Although the individual components are portrayed in different numerical schemes they are linked by a common time-step so that the nonlinear aspects of system integration are preserved. The CST-Model has been used to examine several different examples of coastal system tracts. In one case, the effects of millennial-scale Dansgaard-Oeschger late Quaternary climate and sea level cycles with amplitudes on the order of 0.5 m were examined using a system that contains a river, estuary, inlet, barrier island, shoreface and shelf complex. The system of clinoform development and coastal progradation predicted in the model are similar to those observed on the central Florida Panhandle coast. In another case, the development of a Holocene subaqueous shelf clinoform along the central Italian Adriatic coast was modeled. A number of different source and forcing parameter patterns were varied to demonstrate features that can be identified in the internal stratigraphy of the deposits. These compared favorably with patterns identified in the sub-bottom records. - 45 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 The CST-Model examples illustrate how large-scale processes can either be diagnosed with modeling techniques or simulated in order to make predictions. This model is publicly available to the research community. - 46 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 3-D Photorealistic models of geologic outcrops, a tool to collect quantitative data; methodology and histories examples in Utah and Wyoming Olariu, C. University of Texas at Dallas, 2601 North Floyd Road, 75080, Richardson, TX, United States, cornelo@utdallas.edu The study of outcrops remains a primary tool for understanding facies architecture, build conceptional model and for reservoir characterization studies. Despite the desire of geologists to study continuous, large outcrops, these are difficult to accurately quantify by conventional methods and cliff faces can be tens to hundreds of meters high and/ or kilometres long. Using digital cameras, fast laser rangefinders and GPS, 3-D photorealistic models of the outcrops can be built by: 1. scanning the terrain surfaces of the outcrops; 2. positioning a modest number of photo-control points; 3. digital image collection; 4. building terrain surfaces of the outcrops on the computer; 5. transferring image-coordinates into global coordinates; 6. transforming images on to terrain surfaces. Photorealistic models of the outcrop can be interpreted on a computer. The accuracy of the terrain surface is dependent on the laser used but can as good as millimeters. Also, sedimentary structures observed on the outcrop model depend on a combination of terrain detail and image resolution. Knowing the real 3-D positions of bed contacts on the photorealistic model of the outcrop, a continuous quantitative data such as bed thickness or inclination can be extracted. If the outcrop has a curved geometry or the same bed contact crop on the sides of a canyon, 3-D surfaces might be built by curve fitting and ultimately a solid volume model. The possibility of extracting continuous quantitative data from the outcrops supports the numerical modeling development of the deposits. Examples presented will be from delta front deposits of the Cretaceous Panther Tongue, Utah, and the Wall Creek, Wyoming. Panther Tongue deposits were interpreted using 3-D photorealistic methods together with ground penetrating radar (GPR). - 47 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Upscaling the time parameter in stratigraphic simulation models: efficient use of high-magnitude low-frequency events Overeem, I.1 & Storms, J.2 University of Colorado, Institute of arctic and alpine research, Campusbox 450, 80309-0450, Boulder, CO, United States, irina.overeem@colorado.edu 2 Delft University of Technology, Department of applied Earth Sciences, Mijnbouwstraat 120, 2628 RX Delft, The Netherlands, j.e.a.storms@citg.tudelft.nl 1 When developing a stratigraphic simulation model, much time and effort is generally put into upscaling sediment transport processes, rather than time. But even with present-day computing power, it is extremely difficult to simulate detailed (high time resolution) sedimentary development over long, geological periods of time. The efficiency of sediment transport processes is not constant in time but changes with the variable amount of energy which is available to the depositional system by means of e.g. water discharge (fluvial-deltaic system) or wave magnitude (coastal system). To speak with Derek Ager's (1980) words geologic history can be characterized as "long periods of boredom and brief periods of terror". We here present an upscaling approach that incorporates the high-magnitude low-frequency events that appear to be the important building blocks in stratigraphy, without loosing calculation efficiency in our models. This event-based approach has successfully been applied in two process-response models: SedFlux and BarSim. SedFlux is a coupled simulation model that simulates shelf architecture by including among others the following processes: bedload spreading, dispersal of a delta plume through either hypopycnal or hyperpycnal plumes, reworking of sea-floor sediment by storms and failure and subsequent transport as sediment gravity flows (turbidity currents or debris flows). All these processes are driven by discharge and sediment input. We select the river flood events that carry 90 % of the sediment from a record that has daily values, and only simulate those days in detail. The model averages over the intermittent days and deposition takes place as one single fair-weather layer. Depending on the erraticness of the river system under study this will save often an order of magnitude in calculation time. BarSim models evolution and stratigraphy of wave-dominated coastal systems. The model uses variable time-steps to simulate individual storm events and fair-weather periods. During storms net sediment transport is in offshore direction, whereas fairweather moves sediment back onshore. In addition, sediment supply from for example suspension or littoral drift only occurs during fair weather. Storm wave heights and recurrence interval are drawn from cumulative probability distributions to determine storm conditions and duration of the subsequent fair weather. We find that the event-based approach as compared to the traditional time-averaged approach has distinct influence on the modeled geometry and grain-size distributions. The longitudinal profiles, resulting from either model, show increased deposition distally. The river peak events in SedFlux shoot out delta plumes further - 48 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 offshore and high-magnitude events in BarSim are able to erode the shoreface and transport sediment beyond the wavebase, out of reach of the normal fair-weather processes. Simulated event beds can be interpreted in terms of reworking ratio, grain-size variability, and granulometric facies and can almost directly be compared to real event strata. Also, coupled processes that are threshold controlled are averaged out under time-averaged conditions. (e.g. turbidity current due to hyperpycnal plumes in SedFlux were absent under time-averaged conditions). To conclude, upscaling the time parameter to separate between events and non-events opens new options to better understand both small-scale stratigraphic and large-scale geometric variability. - 49 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Analogue modelling as a tool for investigating causality in alluvial stratigraphy George Postma Utrecht University, Faculty of Geosciences, Budapestlaan 4, 3584 CD Utrecht, The Netherlands,, gpostma@geo.uu.nl Analogue flume models that are properly scaled toward real world prototype riverdelta systems proof suitable to put constraints on existing models when and why rivers of any substantive size will or will not aggrade and preserve systems tracts and their bounding surfaces along their course. Flume modelling done so far have been proven to be adequate in ascribing variation in architectural style to rates of sealevel change (e.g. Van Heijst & Postma, 2001), to subsidence rates in growth fault settings (Van Heijst et al. 2002), and to rates of climate change (Van den Berg van Saparoea & Postma, 2003). Analogue flume modelling proved suitable for systematic exploration of sedimentary system’s sensitivity to allocyclic change. After a brief review the presentation shows the outcome of the first experiments on autocyclic behaviour of alluvial system in the new flume facility Eurotank at Utrecht University. Review The most important architectural styles and sequence stratigraphical models of fluvial systems are briefly reviewed. The large-scale architecture in fluvial-deltaicshelf settings (5th order bounding surface and higher) is defined by the amount of sediment transport that is induced by the rate of sea level change and the ratio of discharge (climate signal) and available bed load (the latter depending on denudation rate and temporarily-stored sediment). Since the amplitudinal variation in sea-level change is much different for fourth order sea-level changes during greenhouse and icehouse periods we think that the present is not the key to the past for understanding fluvial architecture and that present sequence stratigraphical models need to accommodate for these differences. Autocyclic behaviour The first experiments that have been obtained from Eurotank focus on the autocyclic behaviour of the alluvial system. With the term autocyclic is meant the development of an the alluvial system without change in the total energy and material input into a sedimentary system. Keeping all input variables constant, the development of the alluvial system between valley outlet and shelf margin follows a three-phase development. In the first phase, water jets out of the valley causing a cone-shaped, channel-levee depositional feature on the flat coastal plain. During the second phase, water is spreading radially over the forming cone reshaping it into a radial-braiding alluvial fan system. The aggradation rate is high at this stage. In the third phase individual channels start to form at the toes of the fan. These channels develop down to the shoreline, while the water flow is still unconfined. Importantly, it is observed that the - 50 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 carrying capacity of the water flow is much reduced after passing the fan toe. After a while, a multiple channel network start to form by avulsion at the fan apex producing multiple point sources at the shelf margin. The three phase development is similar to that found by Van Wagoner et al. 2003 and Dunn et al. 2003. Discussion Although a systematic experimental data base is still lacking at time of presentation, it is clear that the radius of the braided system is increasingly smaller for lower discharges. Also the time period needed to come to channel formation is shorter for lower discharges. The observed transition from a radial braided system to a channel belt system is marked by a slope break being at the boundary of a strongly depositional landscape producing sheet sandstones and a slightly depositional to erosive landscape with distinct channel belts. Blair & McPherson (1994) noted a similar gradient difference of modern alluvial fans and rivers. The three stage evolution of alluvial systems is also seen in the evolution of a crevasse (Smith et al. 1989; Pérez_Arlucea & Smith, 1999, Farrell, 2001) showing similar development from sand sheet to channels. Murray and Paola (1994) called a braided system a fundamental landform, where the only factors essential for braiding are bed load sediment transport and laterally unconstrained free-surface flow. However, our experiments show that where the channel belts come into existence the flow is still unconfined. In our opinion, the carrying capacity, governed by the discharge per unit area seems much more important than the confinement of the flow. Preliminary data from analogue experiments shows, indeed, that the position of the break between these two fundamental landforms depends on discharge and thus will shift back and forth with climate change. Further research shall contribute to establishment of boundary conditions where fundamental landforms shall exist. These conditions are to be estimated from physical modelling and to be verified along natural examples. References Blair & McPherson (1994) J. Sediment Research 64, 450-489. Dunn et al. (2003) AAPG/SEPM Abstracts Farrel, K.M. (2001) Sedimentary Geology 139, 93-150. Murray and Paola (1994) Nature, 371, 54-57 Pérez_Arlucea & Smith (1999) J. Sediment Research, 69, 62-73. Smith et al. (1989) Sedimentology, 36, 1-23 Van Wagoner et al. (2003), Abstract AAPG/SEPM 2003, Salt Lake City Van den Berg van Saparoea & Postma (2003) Abstract AGU-EUG Mtg. 2003 Nice. Van Heijst & Postma (2001) Basin Research, 13, 269-292. Van Heijst et al. (2002) AAPG Bull, 46, 1335-1366. - 51 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 2D Subsidence Modelling of Middle Triassic Carbonate Platform Development in the Lombardian Alps (Italy) Seeling, M. University of Heidelberg, Institute of Geology and Paleontology, Im Neuenheimer Feld 234, D-69120, Heidelberg, Germany, Michael.seeling@urz.uni-heidelberg.de An Upper Permian to Upper Triassic succession with a thickness of more than 2000 m is well exposed in the Camonica Valley of the Lombardian Alps (Italy). The clastic and carbonate sediments were formed on a passive continental margin, which was occasionally influenced by strong regional transtension. Several carbonate build-ups developed in the Middle Triassic, e.g. the Pora platform and the Concarena platform (Esino Fm.) in the late Ladinian. The rimmed Pora platform has a thickness of up to 400 m, a lateral extension of approximately 10 km and shows strong progradation of megabreccia clinoforms. It covers a time interval between the Archelaus zone and the Ladinian/Carnian boundary (top of Regoledanus zone). The Concarena platform includes a succession of approximately 1500 m thickness. It measures only 4 km in diameter and shows less Progradation than the Pora platform. All typical facies belts of a carbonate platform are present. Large parts of the reef crest still exist and flattening out clinoforms flank this isolated platform. Parts of the existing lagoon consist of hundreds of shallowing upward meter-scale cycles. Detailed basin analysis with reverse modelling shows a complex pattern of differential subsidence between the two platforms in the Camonica Valley. The platform geometries were controlled by an asynchronous development of accommodation space at the beginning of platform growth, induced by tectonic and volcanic activity in Ladinian time. Decompaction calculations of the Ladinian sediments and stratigraphic/geometric restoration of overlying Carnian succession shows a subordinate influence of compaction-induced subsidence on development of accommodation space. Total subsidence rates for the Pora platform are between 80 m/Ma (platform core) and 170 m/Ma (basin) and were strongly outpaced by sedimentation rates. At the end of the Longobardian, the basin had been completely filled and the initial bathymetry between platform and adjacent basin was nearly levelled. Karst features (Calcare Rosso) on top of the entire platform indicate subaerial exposure at the end of platform growth, induced by a global sea-level fall. Calculations of total subsidence rates for the Concarena platform yield values between 320 m/Ma and 360 m/Ma. Sedimentation rates were higher than subsidence rates, but the basin could not bee filled: on the end of platform growth the basin still existed and fine-grained, still water carbonates were accumulated. - 52 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 The Community Surface-Dynamics Modeling System: An Environment for Developing, Sharing, and Using Sediment Models Slingerland, R. Department of Geosciences, The Pennsylvania State University, 16802, University Park, PA, United States, sling@geosc.psu.edu On behalf of sedimentary geologists worldwide, the US National Science Foundation in collaboration with other science-funding agencies is poised to establish an initiative called the Community Surface-Dynamics Modeling System (CSDMS). CSDMS is a modeling environment containing a community-built and freely available suite of integrated, ever-improving software modules predicting the transport and accumulation of sediment and solutes in landscapes and sedimentary basins over a broad range of time and space scales. The scientific infrastructure for CSDMS will be coordinated and funded by government agencies and industry and structured to allow sedimentary modelers from the geological, oceanographic, and engineering communities to determine the optimum algorithms, input parameters, feedback loops, and observations to better predict sedimentary processes and their products. The system is being designed to address both basic science and applied problems such as improved assessment of risk from natural hazards, improved predictive capability of stratal architecture, an understanding of the manner in which sedimentary basins and erosional landscapes participate in biogeochemical cycles, and better interpretation of the record of global and regional climate change. The modeling system consists of three major components coordinated by a system supervisor in the form of GUI, which provides an interactive environment for user access. The Standard Utilities component maintains and stores all data and variable arrays in a compact and quickly retrievable format and contains the Module Connector, an application that allows users to easily link together process-modules from the module library to build a model, thus providing graphical, icon-based model construction. The Module component contains a variety of communitysupplied, compatible computer programs simulating sedimentary processes. Several modules for a given process may be present, each representing an alternative conceptualization or approach to simulating that process. Conceptualizations in a module may be of the traditional PDE form, cellular, or rule-based. The Toolkit Component contains the basic information technology for generating grids and solving equation sets, such as high-order PDE and hybrid PDE/cellular solvers, adaptive meshes, automated mesh and algorithm selection, time-stepping, and domain coupling procedures. Automatic code generators will construct spatial simulations and enable distributed processing over a network of parallel and serial computers, allowing the user transparent access to a network of computing facilities. Nothing approaching the CSDM project in scope or level of cross-disciplinary integration has ever been attempted in the sedimentary geology and surface - 53 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 processes communities, but some examples of current multi-system modeling will be presented to illustrate its potential. - 54 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Avulsion, autogenic or allogenic controlled? Stouthamer, E. & Berendsen, H.J.A. The Netherlands Centre for Geo-Ecological Research (ICG), Department of Physical Geography, Faculty of Geographical Sciences, Utrecht University, E.Stouthamer@geog.uu.nl Because avulsion is an intrabasinal process it is generally considered to be autogenic. However, the avulsion process is not only controlled by autogenic but also by allogenic processes. The Rhine-Meuse delta is without doubt the most extensively studied delta regarding avulsion processes on the timescale of the Holocene. Data suggesting an autogenic nature for the timing of avulsions are: the interavulsion period in the Rhine-Meuse delta on average appears to be constant (~ 945 cal yr); the natural levees of all avulsed channels have a constant elevation relative to groundwater levels, suggesting that the critical superelevation for avulsion remains constant through time; a weak 500 14C yr cycle seems to be present in the avulsion frequency. However, the avulsion locations are not randomly distributed over the delta. Avulsion locations are related to allogenic factors, interacting over time and space. Eustatic sea level rise and glacio-hydro-isostacy played a dominant role from 75004500 14C yr BP, (local) tectonics from 4500-2800 14C yr BP and discharge/sediment load changes from 2800-1000 14C yr BP. After 1000 14C yr BP avulsions are controlled by humans. In conclusion, it seems that avulsion is controlled by both autogenic and allogenic processes. The relative importance of these processes varies over time and space and leads to different alluvial architectures. - 55 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Variation in dip of lateral accretion surfaces in subrecent fluvial deposits, Pannonian Basin, Hungary: a reflection of climatic fluctuations or just meandering excursions? Sztanó, O. 1 & Mészáros F. 2 1 Department 2 of Geology, Eötvös University, H-1117, Budapest, Pázmány P. sétány 1/C, sztano@ludens.elte.hu The Relief Laboratory, H-8442. Hárskút, P.O.Box 1, ferenc@relief.hu The study area and object Ultra high-resolution single-channel seismic profiling provided a detailed picture about the geometry of Pleistocene - early Holocene alluvial deposits underlying the recent bed of river Tisza. The alluvial architecture of 10-20 m thick (10-30 ms twt) sediments between the riverbed and its first seismic multiple has been imaged with a resolution of 0.5 m horizontally and 0.1 m vertically (Tóth et al, 1997). The ultra high-resolution seismic sections were acquisited along the temperate climate, high sinuousity meandering Tisza river, which is the largest river within the Pannonian Basin. From the Pleistocene onward due to moderate rate of subsidence the ancestors of the Tisza and its distributaries have been filling up the basin in an overall aggradational style, with minor and local events of incision only. Seismic interpretation led to the identification of several distinct seismic facies with characteristic bounding surfaces which were explained in terms of architectural elements (Sztanó et al, 2002). The most important elements are the followings: - Bundles of high amplitude oblique, tangential or sigmoidal parallel clinoforms were interpreted as macroscale inclined strata sets (sensu Bridge 1993). Having a strong, relatively flat basal reflector these were regarded as the result of lateral accretion. The series of inclined reflections form several storeys, which are corresponding to 5-7 m thick sediments, horizontally covering a distance of 250 m up to 3500 m (!). - Trough-shaped high to moderate amplitude parallel or converging reflections were frequently found next to series of inclined reflectors. The troughs are 150-200 m wide and 5-8 m deep. This geometry was interpreted as channel fill after natural cut offs. - Inclined or trough-shaped reflectors often cut sharply into horizontal, weak to good continuity parallel reflections of significant length. These were interpreted as vertically accreted fine grained overbank or floodplain deposits. Both the lateral accretion deposits and the abandoned channel-fills indicate the development of a relatively large river during Late Pleistocene- Early Holocene times similar in size to the recent meandering Tisza. In series of lateral accretion deposits longer than a few hundreds of metres a variation in dip angle of the inclined strata sets is obvious. The aim of modelling would be to reveal the factors that influence these geometrical variations, and to distinguish climatically induced palaeohydrological effects from random meander excursions. Former modelling of meander evolution - 56 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Simulations by Bridge (1975) focused on changes in plan form and cross section with respect to meander movement and increase of amplitude and sinuosity. Scouring and filling produced the relief of the basal surface related to the slope of lateral accretion surfaces, but angle variations were not adressed yet. Willis (1989) recognized that point bar deposits thicken and inclined bedset surfaces become steeper towards the bend apex and with increasing sinousity. Out of these effects the change in sinuosity seems to be the more important. The role of direction and position of cross-sections (outcrops) were also displayed. A more detailed modelling and comparison with real outcrop situations was presented by Willis (1993). It is shown that with increasing bend wavelength lateral accretion deposit thickness and dip of bedsets decreases. The gradual reduction of discharges has the same effect on geometry. Variations in dip angle of inclined bedsets produce two types of disconformities within inclined bedsets: sets with low angle dips truncated by more steeply dipping ones are intrepreted as the result of sudden increase in discharge; while sets having steeper dips below the disconformity overlain by relatively less dipping surfaces are resulted from a drop of sinuosity during chute cutoff. The above modells also take variations in vertical grain size and transport direction into consideration. Variation in dip angle of inclined bedsets One of the longest series of inclined bedsets forms an about 3 km long, 7 m thick storey (Fig.1). Its basal surface runs 10 m below the recent river bed. The true dip of the bedsets is very low (about 2-3o), but due to vertical exageration the subtle variations in the dip angle are clear. The seismic section is rather straight in plan view; the lateral-accretion bedset surfaces most likely represent an oblique section of the ancient meander bend. These surfaces are often truncated by steeper or offlapped by flatter ones, followed by reactivations. The initial downcutting is relatively steep and is overlain by a series of progressively steepening oblique reflectors. This 500 m long part of the section is divided into bundles (1-5) by discontinuities, which having a larger dip truncates the underlying strata, and follow each other in 100 m distances. On the following 1000 m there is only one single erosional surface (7). However, another kind of discontinuity, marked by onlaps of more flat lying reflectors is frequent (6, 8, 9, 10). From bundle 5 to 6a a gradual decrease of dip angles, from 6b to 7 an increase and from 8 to 11 a decrease occurred. Parallel with this flattening a thinning of the unit took place. On the following 600 m erosional truncations became frequent again, accompanied with progressive increase of dip angles (11-15) and a thickening of the storey. In bundle 15 the dip angle decreases, in 16-17 it remains constant, but from 18 to 21 a gradual decrease is evident, however the thickness of the unit does not change. The upper portion of surface 21 is erosional and even shows a ridge and runnel topography. Interpretation and discussion According to modelling results the left side of the section (bundles 1-5) reveals the birth of a bend with increasing sinuosity parallel with a progressive increase of peak discharges in regular timespans. From 5 to 6a the growth of the bend wavelength can - 57 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 be supposed followed by a renewed phase of increasing sinuosity (6b-7). With this expansion the bend most likely reached its largest size, which also indicates the greatest discharges ever reached. The frequent occurrence of erosional floods during the beginning points to a variability of climate which later became balanced. From unit 8 to 11 a considerable shallowing of the channel took place indicating a gradual decrease of discharge, again with fluctuating climatical background. This was followed by an increase of sinuosity, discharge and frequency of large, erosive floods (11-15), coupled with the deepening of the channel. For a short time growth of bend wavelength can be supposed (15) which was followed by a period of constant uniform translation during stable conditions (16). After the deep incision (17) a gradual widening of the channel without changing depth can be suspected, which might have been associated with a slightly increase in discharge. Surface 21 can be interpreted as chute cut-off. Unfortunately from a seismic section paleo-transport directions cannot be read in order to follow channel orientation and surface shape. Thus an alternative interpretation for bundles 8-11 might be that the direction of meander migration changes and the section goes from bend apex to inflection point. In that case it has nothing to do with disharge variations. Having an excellent 3-D controll on exposed lateral accretion surfaces and scroll bar morphology Diaz-Molina (1993) suggested that all erosional surfaces between conformable lateral accretion units demonstrate changes in channel displacement direction. Although this cannot be excluded the observed consistent and repeated variations in the dip angle of the lateral-accretion surfaces were not likely to be produced by random meandering. If the above assumed variation of discharges is interpreted as the reflection of climatic changes a geological time frame is needed. In lack of direct data the recent rate of meander migration can be considered as the best analogue available. That was between 0.5-1.5 m/year averaged from the last 120 years on this part of Tisza river, thus deposition of the whole succession may took place during about 3000 years. The formation of the river (1-5) as well as the growth of the meander bend (6-7) could have happened at about the Pleistocene/Holocene boundary when the rise of the temperature was associated with an amplified amount of precipitation and runoff (climatical variability first followed by more stable conditions). Later on precipitation decreased resulting in the decrease of discharge, reflected by shallowing of the river (8-11) in the first part of the Boreal. Calculating with the same average rate of meander migration later the increasing discharge (11-15) indicates the increase of precipitation again, followed by constant high values (15-17) during at the onset of the Atlantic. The end of the section may point to the maximum discharge attained during the Early Atlantic. The above deduced climatic fluctuations are in accordance with those observed by several means in the Pannonian Basin (Gábris, 1995). Conclusions Internal erosional surfaces indicate major floods in a recurrent period of about 100 years during periods of climate change. The variations in dip angle of lateral accretion surfaces through changes of morphometric parameters (sinuosity, wavelength, width/depth ratio) reveal periodic increases and decreases in discharge - 58 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 due to climatic effects. The seismically studied fluvial deposits nicely reflect those variations in precipitation which took place during the Late Pleistocene –Early Holocene. Acknowledgement The seismic survey was carried out by the Geomega Ltd., Budapest. The study was supported by the Hungarian National Science Found OTKA no. T.037724, T.032956 and T.034979. References Bridge J.S. (1975) Computer simulation of sedimentation in meandering streams. Sedimentology 22, 343 Bridge J.S. (1993) Description and interpretation of fluvial deposits: a critical perspective. Sedimentology 40/4, 801-810 Diaz-Molina M. (1993) Geometry and lateral accretion patterns in meander loops. In: Marzo M. & Puigdefabregas C. (eds.) Alluvial Sedimentation, IAS Spec. Publ. 17. pp.115-131, Blackwell, Oxford Gábris Gy. (1995) River activity as a function of changing palaeoenvironmental conditions during the Late Glacial – Holocene period in Hungary. In: Frenzel B. (ed). European river activity and climatic change during the Late Glacial and Early Holocene, Fisher Verlag, 205-212. Sztanó O., Tóth T., Magyari O., Magyari Á. & Horváth F. (2002) :Alluvial architecture from ultra highresoltution single-channel seismic survey of the meandering Tisza river, Pannonian Basin, Hungary , 16th International Sedimentological Congress Abstract Volume 357-359 Tóth T., Simpkin P., Vida R. & Horváth F. (1997) Shallow water single and multichannel seismic profiling in a riverine environment, The Leading Edge, 16/11, pp.1691-1695 Willis B.J. (1989) Palaeochannel reconstructions from point-bar deposits: a three-dimensional perspective. Sedimentology 36, 757-766 Willis B.J. (1993) Interpretation of bedding geometry within ancient point-bar deposits In: Marzo M. & Puigdefabregas C. (eds.) Alluvial Sedimentation, IAS Spec. Publ. 17. pp.101-114, Blackwell, Oxford Fig.1. Variation in dip of inclined bedsets drawn from ultra-high resolution seismic images were interpreted as lateral accretion surfaces (LA). Numbers mark bundles. see next page for enlargment - 59 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 - 60 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 The Role of Numerical Sedimentary Process Models in Hydrocarbon Exploration and Reservoir Characterization Tetzlaff, D. Schlumberger-Doll Research, 36 Old Quarry Road, 06877, Rifgefield, Connecticut, United States, dtetzlaff@slb.com Forward numerical models of sedimentary processes were initially developed as a research tool in sedimentation. Their representation of physical sediment-transport phenomena has increased their appeal in the oil industry as a complement to more descriptive geostatistical methods. Here we report some of the successes and difficulties we have encountered in the development and use of a siliciclastic sedimentary process model for hydrocarbon exploration and reservoir characterization. A useful sedimentation model must represent natural processes in a practical amount of computer runtime. This need is exacerbated by the fact that forward models must often be run repeatedly in the search for the initial and boundary conditions that cause the output to conform to observations. Efficient representation of the processes of erosion, transport, and deposition appears to require the use of more than one numerical method, depending on the depositional environment represented. Steady flow situations (rivers at normal stage, long shore currents, etc.) appear to be adequately handled by finite-difference schemes, whereas highly unsteady flow involving sediment transport (floods and turbidites) is more efficiently represented by “particle” numerical techniques. Users of a sedimentation model’s output (seismic interpreters and log analysts among others) require that the information produced by the model (sedimentary deposit geometry, lithology, and rock properties) be converted to a form that is directly comparable to their data (seismic data and well logs). Therefore, we have implemented a procedure to obtain petrophysical properties from a sedimentary model’s output and developed a 3-D hybrid seismic modeling technique based on a ray-plane-wave approximation. The resulting seismic sections outline lateral extent and dip of the simulated deposits and exhibit amplitude variations with offset (AVO) that are directly comparable with seismic data. We show an application example in which we apply the methods outlined above to a deep-marine reservoir analogue from the Tanqua Karoo Basin, South Africa. The example suggests that the combination of these techniques forms a workflow that constitutes a promising basis for expanding the application of sedimentation models in the oil industry. - 61 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Models that talk back Tipper, J.C. Geologisches Institut, Albert-Ludwigs-Universität, Albertstrasse 23B, D-79104 Freiburg, Germany, john.tipper@geologie.uni-freiburg.de 1) Each model that a scientist makes of a system is an expression of a particular set of ideas. These ideas - they can be dignified by the terms ‘hypothesis’ or ‘theory’ depending on the amount of evidence that exists to support them - are expressed in the model in a form in which they can readily be worked with; this form may be physical, or mathematical, or graphical, or even verbal. There are two reasons for expressing ideas in this way. Either it may be hoped that experimentation with the model will help in understanding the structure and function of the parent system, or it may be believed that the model can predict how that system will behave under specified input conditions. Understanding and prediction can aptly be described as the twin goals of scientific modelling. 2) Understanding clearly must precede prediction, but how can a system ever be known to be understood - or at least be known to be well enough understood to allow a model of it to be used for prediction? There is a range of answers to this question, from the principled to the pragmatic. The principled answer is that a system should be taken to be understood only when it can be fully represented in terms of fundamental physical laws. The pragmatic answer is that a system can be taken to be understood as soon as any model has been made of it that can be shown to predict how it will behave for a wide range of specified input conditions. The principled position allows modelling only for the purpose of prediction; the pragmatic position allows modelling both for prediction and for understanding. 3) Modelling for prediction is carried out when a system is judged to be well enough understood that a model of it can be used for prediction for the particular set of input conditions that is of interest - this judgement is of course often entirely selfserving. Modelling for understanding is carried out when a system is judged not to be well enough understood. Modelling for understanding takes two forms: the testing of the ideas expressed in a model, and the exploration of the consequences of varying a model’s assumptions or parameters. Modelling that involves testing is essentially deductive in its nature - as is modelling for the purpose of prediction; modelling that is exploratory can be inductive or deductive, depending on the conditions under which it is carried out. 4) The standard strategy employed when models are used for testing involves running the model and its parent system under identical input conditions; the behaviour of the model is then compared to the behaviour of the system. If the behaviour of the model is identical to that of the system, the confidence that the modeller has in the correctness of the ideas expressed in the model is increased; if the behaviour of the model differs from that of the system, the confidence in the - 62 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 correctness of those ideas is decreased. This strategy is attractively simple and is therefore widely used. Its simplicity is deceptive, however, and problems inevitably arise whenever it is implemented in practice. The most substantial of these are: (1) the problem of adequate access to a model’s parent system, (2) the problem of judging the significance of differences in behaviour, and (3) the problem of allowing for nonuniqueness. These problems are particularly acute for the types of models with which earth scientists commonly try to work – i.e. for field-scale models of natural systems operating over long periods of time. 5) It is helpful to characterise models in terms of the degree to which they are logically identical to their parent systems. A model that is logically identical to its parent system is one for which the system is fully understood and for which the input state is always known. A model that is not logically identically to its parent system has free parameters or auxiliary hypotheses; the free parameters might be being used to make up for deficiencies in the basic understanding of how the system works, and the auxiliary hypotheses might be being used to make up for lack of knowledge about the system’s input state. A model that is logically identical to its parent system does not require testing; it can be used immediately for prediction, for all of the allowable input states. A model that is not logically identical to its parent system needs at least some testing; it can nevertheless also be used for prediction, for some input states. The amount of testing that a model needs is measured by the proportion of input states for which the model cannot be trusted to show the same behaviour as the system itself; the amount of prediction possible from that model is measured by the proportion of input states for which the model can be trusted. The relationship between the amount of testing needed and the amount of prediction possible is shown schematically in Figure 1, as a function of the extent to which a model and its parent system are logically identical. 6) The testing-prediction relationship curve defines two distinct styles of modelling. One style (below the curve) is modelling that is inherently incapable of being critical - it is what Medawar has termed ‘academic play’. It tests models less frequently than is necessary, it uses them for prediction in circumstances where they cannot be relied on to behave in the same way as their parent systems, and it includes exploratory investigations that are essentially inductive (Baconian experimentation). The other style of modelling (above the curve) is modelling designed always to err on the side of safety. It tests models more frequently than is really necessary, it uses them for prediction less frequently than would be possible, and it includes exploratory investigations that are strictly deductive (Kantian experimentation). The first style of modelling is prevalent whenever models are made that have unconstrained auxiliary hypotheses and large numbers of free parameters; such modelling says very little of value about the system concerned, because models such as these do little more than obey the modeller who made them. The second style of modelling is effectively possible only for models that are close to being logically identical to their parent systems - these are models that talk back. - 63 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 7) An example will be given of each of the two styles of modelling, both to illustrate the essential differences between them and to point out their strengths and weaknesses. A model of landscape development in southeastern Australia provides the example of ‘academic play’, and a model of sedimentary cyclicity provides the example of a model that talks back. Figure 1. The testing-prediction relationship curve separates the two styles of scientific modelling. - 64 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling the impacts of climate change on erosion, sediment production, and landscape evolution Tucker, G. School of Geography & the Environment, Oxford University, Mansfield Road, OX1 3TB, Oxford, United Kindom, gtucker@nimbus.geog.ox.ac.uk Sediment supply is one of the primary controls on deposition rate and stratal architecture. In this talk I review analytical and numerical models of landscape denudation, sediment production, and coupled subaerial erosion-deposition systems. The present state of the art consists of a variety of different models. Most of these represent fluvial and hillslope processes, though some attempts have been made to capture glacial dynamics as well. All drainage basin evolution models are able to reproduce branching drainage networks and ridge-valley topography, but these models make different (and generally testable) predictions about transient responses to tectonic, climatic, or eustatic change. Efforts at testing alternative models are ongoing and involve a range of approaches, including tests based on case studies with known initial and boundary conditions and known forcing. Several lessons have been learned recently about the sediment-flux responses to external (climatic, eustatic, and tectonic) forcing. Models predict, for example, that when the forcing timescale is comparable to the geomorphic response timescale, the nature of sediment-flux curves differs fundamentally between scenarios of tectonic versus climatic forcing – the former producing a delayed but sustained response, the latter producing an immediate but transient response. These response timescales, and their sensitivity to factors such as source terrain size, can be derived from analytical or numerical models, but their predictions are mostly untested. In terms of climate forcing, recent work has unpacked the overly simplistic ‘wet versus dry’ thinking of the past and begun to reveal how particular aspects of climate change can influence denudation rates and sediment supply. For example, theoretical work implies that short-term climate variability can have a dramatic impact on sediment production, which supports the hypothesis that climate change is responsible for worldwide accelerated sedimentation in the late Cenozoic. Although the mechanisms leading to geomorphic sensitivity vary, nonlinearity and sensitivity to highfrequency climate change seem to be general properties of many different geomorphic systems. - 65 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Scaled models of syntectonic sedimentation – review of principles and first results Urai, J.L. & Kukla, P.A. RWTH Aachen University Sedimentary systems which are active under conditions of active tectonics show a range of different types of non linear responses which are incompletely understood. Scaled models are useful tools to understand the interactions in both sedimentary and tectonic systems. However, experiments which might combine these two have not been attempted yet. In this paper we review the basics of designing scaled model experiments of both tectonic and sedimentary systems, and discuss the limitations of current techniques. Some of the limitations of current experiments have been resolved by a new sandbox now in operation at Aachen University which operates submerged in water. This way, models containing water-saturated layers of sand and clay can be deformed. Recent developments in image analysis technology allow real-time high resolution analysis of the displacement field at length scales less than a grain’s diameter, and resolution of an unprecedented level of detail in the organization of deformation patterns. This system is suitable for experiments where a simple sedimentary system is simulated during movement on an active fault. We present first results of such experiments and discuss the scope for larger scale and more complex simulations involving growth faults. - 66 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Outcrop and seismic constraints for 3D numerical stratigraphic forward modeling Van Buchem, F., Granjeon, D. & Eschard, R. Institut Français du Pétrole, Rueil-malmaison, France Stratigraphic forward modeling allows to quantify the interaction between the three main “processes” that control sedimentation: accommodation (tectonism, eustasy and compaction), sediment supply (clastic supply and carbonate production) and sediment transport. A key problem, however, is to determine the relative contribution and influence of each one of these parameters. The only way to address this problem is to work with the best documented and understood case studies available. In addition, the insight gained from them, feeds the memory of the model, and can be used as constraints in less well documented studies. In this presentation we will give two examples of carbonate systems, one mixed carbonate/siliciclastic system, and two siliciclastic systems. In most of the examples high resolution sequence stratigraphy has been applied to unravel the fine scale architecture of these systems. - - - - The first carbonate example is the Natih Formation of Cenomanian age (Upper Cretaceous) that outcrops in northern Oman (van Buchem et al., 2002). Our work demonstrated that in this formation twice an organic-rich intrashelf basin was formed mainly controlled by the response of the sedimentary system to the rate of sea level rise. Tectonism only played a minor role in the initiation of the intrashelf basins. The stratigraphic modeling effectively produced these basins, based on differential sedimentation rates of the carbonates in response to the sea level fluctuations. The second carbonate example is from the Upper Miocene in Mallorca. The work by Pomar (2001) showed that a change in ecosystem caused a change from a low angle ramp to a rimmed platform. Using the ecological parameters of this study in the stratigraphic modeling (depth of carbonate production, type of producers), allowed to reproduce this transition in stratigraphic organisation. In the Lower Miocene of south central Turkey a mixed carbonate siliciclastic system has been studied in seismic scale outcrops (Bassant, 1999). Here the interplay between carbonates and siliciclastics was the main challenge, with siliciclastics inhibiting carbonate sedimentation during highstand and fall of sea level (around the point source), while during sea level rise an ecological ‘window’ was created for carbonate sedimentation. Introducing a sensitivity to siliciclastic input for the carbonate producers permitted to model this complex interfingering pattern. The Eocene of offshore Brasil has been studied using wells and high resolution 3D seismic (Pinheiro-Moreira, 2000). This database allowed to determine the 3D organisation of the turbidite system. Stratigraphic modeling led to a validation of the seismic analysis in this deep offshore environment, and to a better - 67 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 understanding of the timing, geometry and facies of turbidites in two different basin margin settings (ramp and shelfbreak; Granjeon et al., 2001). - The last example concerns the Colorado basin in Argentina. This basin was initiated during an early Mesozoic rifting period, followed by a Cretaceous to Miocene sag period, to ultimately reach the present day passive margin configuration. Only the upper third (Upper Cretaceous to Quaternary) of the sedimentary column is covered by wells, located at the margins of the basin. A stratigraphic simulation was constraint by excellent seismic data and these few wells, to quantify the geological parameters and evaluate the facies distribution in the non-drilled areas (Mancilla et al., 2002). Basin modeling was then used to asses the hydrocarbon potential of this basin (Haring et al., 2002). Though stratigraphic modeling is in many ways a powerful tool to test the coherency and imagination of the geologist, the ultimate test of the modeling results is the comparison with the hard field and subsurface data. The case is made here that the real advances in stratigraphic modeling will come from top quality geological studies, allowing to quantify the parameters of dynamic sedimentological models in a high resolution time framework. Acknowledgements: Dionisos is a 3D stratigraphic forward model, developed by IFP and a consortium of companies including Total, Shell, Repsol-YPF, Chevron-Texaco, Petrobras, Agip, IMP, and Pemex. - 68 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Impact of discharge and sediment flux on Basin Margin Architecture: An experimental approach Van den Berg van Saparoea, A.P., Postma, G. & Dalman, R. Universiteit Utrecht, Faculteit Aardwetenschappen, Budapestlaan 4, 3584 CD, Utrecht, The Netherlands, apvdbvs@geo.uu.nl In most cases it is very difficult to evaluate the stratigraphic record in terms of formational processes. Important parameters in fluvial-deltaic settings are relative sea-level and the ratio of discharge (Q) and solid load (Qs). We use analog modeling for unraveling the role of each of these formational processes. The changes in the modeled system are monitored by means of digital elevation maps (DEM's) of the surface, and are calibrated against the last glacial cycle stratigraphic record of the Colorado (Texas, USA) fluvial-deltaic system. The influence of relative sea-level at constant Q/Qs ratio has been investigated in a previous series of experiments (Van Heijst et al. 2001, Basin Research). In the experiments described here Q/Qs ratio is varied, while the rate of change in relative sea-level is kept constant. Our preliminary results indicate that the efficiency of sediment transport increases with higher Q/Qs ratios. A system with high Q/Qs ratio is predominantly progradational, while a system with a low Q/Q s ratio shows progradation and aggradation over the entire shelf. Furthermore, a high Q/Q s ratio leads to increased headward erosion rates in the fluvial valleys incised in the shelf. Therefore, connection of the shelf valleys with the trunk river occurs earlier than in case of a lower Q/Qs ratio, which means that even though the system is fluvial dominated (i.e. strongly progradational), the influence of sea-level fluctuation on the fluvial domain is strong. A sudden increase in discharge (and thus increase in Q/Qs ratio), as is postulated to have occurred in the Colorado system, results in the model in instantaneous adjustment (lowering) of the equilibrium slope in the fluvial valley, and increased sediment yield at the river valley mouth. The climate shift also causes increased avulsion frequency and aggradation over almost the entire shelf. If the pulse coincides with the late sea-level fall (past the fall inflection point), it will delay the onset and magnitude of headward erosion on the shelf and so decreases the influence of sea-level on the river system. A shift in climate can apparently have a strong influence on the development of the system, modifying importantly the impact of glacio-eustatic sea-level changes on the system. In conclusion: - High Q/Qs ratios produce strongly progradational systems, high knickpoint migration rates in the valleys and relatively early connections of shelf and fluvial valleys; - 69 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 - Low Q/Qs ratios produce predominantly aggradational systems, low knickpoint migration rates and relative late connections of shelf and fluvial valleys; - Climate shift towards wetter conditions causes a strong sediment pulse at the river mouth, increases avulsion activity and so enhances wide distribution of sediment over the shelf. - 70 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 High resolution 3D Forward Stratigraphic modelling of Cretaceous carbonate platform systems VAN DER ZWAN, C.J.1, MASSE, J.-P.2, BORGOMANO, J.1, 2, LAMMERS, H.1 & FENERCI-MASSE, M.2 SIEP-SEPTAR, P.O.Box 60, 2280 AB, Rijswijk, The Netherlands, Kees.vanderZwan@shell.com University of the Provence, Labo Sedimentologie, 3, place Victor Hugo, 13331, Marseille, France, jborgo@up.uni-mrs.fr 1) 2) 3D Forward stratigraphic modeling is applied in the petroleum industry to predict the presence and quality of reservoirs, source rocks and seals both on exploration and production scales. These models consist of 3-dimensional representations of facies attributes that can be related to physical properties of reservoirs, seals and source rocks. The forward modeled facies and geobodies can also be used to constrain production scale reservoir architectures. The main input parameters to the 3D stratigraphic modeling in any depositional settings are: (1) tectonic subsidence, (2) initial bathymetry, (3) eustatic sealevel and (4) sediment supply. In a clastic setting, the latter parameter reflects the amount of sediment entering the system. In a carbonate setting this reflects in situ production and re-distribution of carbonate material. The first three parameters can be obtained or reconstructed from standard seismic and deterministic geological interpretation or are directly availa ble from regional sources. The fourth parameters, carbonate production is the subject of ongoing research, is dependent of many other ecological and environmental parameters and requires sensitivity analysis Recent high-resolution outcrop studies of Lower Cretaceous systems in Southern France improved our understanding of the development of Cretaceous carbonate platforms. This data set has shown the interrelationships between water depth, energy levels and the various facies associations in a 'Rudist' platform setting. These data have been used to calibrate the 3D forward stratigraphic modelling of similar Cretaceous carbonate platform systems in the subsurface at the scale of producing oil fields. The software package DIONISOS accurately modelled the stratigraphy and facies distribution of a Cretaceous carbonate platform in the Cassis and Mont Puget outcrops near Marseille (SE France) and in a Cretaceous producing field. In Cassis and Mont Puget, inner and outer platform settings were modelled, respectively, whereas in the Cretaceous Field various scenarios were used to test the sensitivity of the controlling parameters. Evaluation of the resulting models shows that the most important parameter in stratigraphic forward modelling of such carbonate platforms is changes in "accommodation space". This parameter determines the space available to be filled by carbonate growth. The changes in carbonate growth potential have less impact, provided carbonate growth is sufficient to fill the space created. Forward models have been built for inner and outer platform settings both in - 71 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 outcrops and in producing fields. Independent calibration of these forward models indicates that they compare closely, with a vertical resolution of one meter, with stratigraphic architecture of outcrops and fields. The ability of stratigraphic forward modelling to build such accurate and highly detailed models demonstrates that stratigraphic forward modelling is a valuable new technique not only to improve our understanding of the subsurface, but also for the prediction of reservoir architecture both on an exploration and production scale. - 72 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Combination of Structural Balancing, Reverse Basin and Forward Stratigraphic Modelling: Southern Cantabrian Basin (NW-Spain) Vesolovsky, Z. Geologisch-Paläontologisches Institut, University of Heidelberg, Im Neuenheimer Feld 234, D-69120, Heidelberg, Germany, zbynek.veselovsky@urz.uni-hd.de The processed transect (24km) situated at the southern margin of the Cantabrian Mountains (NW-Spain) comprises the whole basin infill between top of basement (560Ma) and time of maximal burial (34Ma). This part of the Variscan orogen is represented by a foreland thrust and fold belt, further shortened during Alpidic orogenesis, offering magnificent outcrops in each individual thrust sheet. In order to carry out basin modelling at this tectonically affected region, a multidisciplinary approach of detailed field mapping, structural balancing and stratigraphic modelling is required. Total tectonical shortening of the basin, derived from 2D structural balancing of the deformed basin infill, amounted to 54% at minimum. 2D reverse basin modelling investigated the quantitative development of the basin architecture and long-term evolution of accommodation space in time. Stratigraphic forward modelling quantified sedimentary processes (erosion and sedimentation rates), refined the sequence stratigraphic model, and offered geometrical minimum/maximum models of the sedimentary patterns. Three major subsidence trends obtained from 2D reverse basin modelling reflect different plate-tectonic development stages of the basin. Precambrian to Ordovician times recorded uniform tectonic subsidence rates (-11 to 19 m/Ma). Differential subsidence triggered by local tectonics (-28 to 66m/Ma) generates marked changes of depositional environments (Silurian until Late Devonian). Our study indicates major influence of total subsidence and sediment supply liable for the remarkable alternating siliciclastic and carbonate deposits during the Devonian. Eustatic sealevel fluctuations were only of subordinate importance. In the Upper Carboniferous, eastward migrating depocentres coupled with high rates of tectonic subsidence (-177 to 300 m/Ma) reflect movements of the Variscan front. - 73 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Non-unique sequence stratigraphic architectures Waltham, D., Udofia, M. & Nichols, G. Department of Geology, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK, d.waltham@gl.rhul.ac.uk Sequence stratigraphic architectures are controlled by subsidence, eustatic sealevel and sediment supply. A key problem, both for real-world architectures and for computer models, is that there may be more than one combination of these controls which produces the same geometry. In this paper we investigate the general conditions under which this non-uniqueness occurs and we illustrate this using computer generated cross-sections which, indeed, show identical architectures resulting from differing control-functions. The simplest case is that of a non-erosive model with a constant coarse/fine ratio in the input sediment. Such models produce identical results for all runs which have the same sealevel versus sediment supply curve. More general cases with timevarying coarse/fine ratios and time-varying tectonics can be handled by requiring these curves to also be fixed when plotted against sediment supply. All of the preceding examples are special cases of models in which the incremental change in stratigraphy is controlled by the amount of accommodation space but not by its rate of creation or destruction. This requirement is violated if an erosion rate is introduced to the model. Surprisingly, however, multiple near-identical geometries can still be produced for erosive models provided the sealevel curves are suitably adjusted. Finally, the methodology derived from the above conclusions is applied to a 2D seismic line from the Baltimore Canyon. - 74 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Insights into Carbonate Platform Drowning Mechanisms using Stratigraphic Forward Modeling Warrlich, G. & Burgess, P.M. Shell International Exploration and Production B.V., P.O.Box 60, 2280 AB Rijswijk, The Netherlands Mechanisms of carbonate platform drowning remain poorly understood, since modern production rates imply that even the most rapid rates of relative sea-level rise are insufficient to outpace production. Other proposed drowning mechanisms include water temperature changes, changing nutrient levels, plate movement into higher latitudes, erosion or stripping of sediment from platform tops by transport, and rapid water depth increase during lag periods (deep flooding). We have used two different stratigraphic forward models to investigate the operation of the latter two mechanisms. One model concentrates on sediment transport on the platform top, and the other that uses cellular automata to control carbonate productivity and hence simulates a lag process. Model runs combine long term subsidence with a 100 ky eustatic signal and typical shallow water carbonate production rate to determine under what combination of these parameters drowning occurs. - 75 - Analogue and numerical forward modelling of sedimentary systems; from understanding to prediction Utrecht, 9 – 11 October 2003 Modelling Source Rock Distribution and Quality Variations: The New OF-Mod 3D Technology Zweigel, J. SINTEF Petroleum Research, S.P. Andersens veg 15b, 7465 Trondheim, Norway, Janine.zweigel@iku.sintef.no In exploration the ability to predict hydrocarbon occurrence and quality variations within a prospect - prior to drilling - is of large importance. Particularly, recently developed 3D modeling techniques are gaining significance with respect to volumetric hydrocarbon predictions. The source rock is the basis of every petroleum system and the first pre-requisite for a hydrocarbon accumulation to occur. But source rock distribution, type and quality variations are still among the least constrained parameters during basin modeling studies. The OF-Mod program is a process-based modeling tool, which mimics the development and the variation of source rock facies. Consequently, it allows quantitative prediction of source rock potential away from well control and results in a significantly improved input for kinetics, expulsion and migration studies. The concept was designed in early 98 and in meantime the 2D version of the software has been applied in many national and international studies and has gathered scientific approval. The further development of the software into a 3D version is a consequence of the technological evolution during the last years and will lead to a significantly improved picture of the complex interaction of processes effecting source rock deposition. In a theoretical example we will illustrate the possibilities, differences and consequences of 2D and 3D source rock models with respect to quantitative estimates. Additionally, we will show the capabilities of the new 3D tool with special attention to integrated, high-resolution basin modeling studies and, in contrast, application in exploration frontier areas, where only little information is available. - 76 -