Numerical Modelling of Carbonate Platforms and Reefs: Approaches and Opportunities

This paper compares different computing procedures that have been utilized in simulating shallow-water carbonate platform development. Based on our geological knowledge we can usually give a rather accurate qualitative description of the mechanisms controlling geological phenomena. Further description requires the use of computer stratigraphic simulation models that allow quantitative evaluation and understanding of the complex interactions of sedimentary depositional carbonate systems. The roles of modelling include: (1) encouraging accuracy and precision in data collection and process interpretation (Watney et al., 1999); (2) providing a means to quantitatively test interpretations concerning the control of various mechanisms on producing sedimentary packages; (3) predicting or extrapolating results into areas of limited control; (4) gaining new insights regarding the interaction of parameters; (5) helping focus on future studies to resolve specific problems. This paper addresses two main questions, namely: (1) What are the advantages and disadvantages of various types of models? (2) How well do models perform? In this paper we compare and discuss the application of five numerical models: CARBONATE (Bosence and Waltham, 1990), FUZZIM (Nordlund, 1999), CARBPLAT (Bosscher, 1992), DYNACARB (Li et al., 1993), PHIL (Bowman, 1997) and SEDPAK (Kendall et al., 1991). The comparison, testing and evaluation of these models allow one to gain a better knowledge and understanding of controlling parameters of carbonate platform development, which are necessary for modelling. Evaluating numerical models, critically comparing results from models using different approaches, and pushing experimental tests to their limits, provide an effective vehicle to improve and develop new numerical models. A main feature of this paper is to closely compare the performance between two numerical models: a forward model (CARBONATE) and a fuzzy logic model (FUZZIM). These two models use common data sets, thereby permitting one to obtain similar results (Norlund, 1999). The geological model we use to test the validity of these two numerical models comes from a Holocene coral reef located at the island of Mauritius, Indian Ocean. A detailed description of the stratigraphy and general geological setting is given by Montaggioni and Faure (1997). The general stacking pattem and facies distribution obtained are similar to the Holocene reef section used as an example (Dalmasso, 2001). The results here include: (1) an enhanced understanding of similarities and differences between models and modelling philosophies; (2) critical evaluation of applications and assessment of how models have been utilized; and (3) improvements and refinements in techniques for generating and describing model inputs and outputs. The models have various drawbacks, most of which are due to the lack of knowledge about the systems we are trying to model, but also due to the incapability to make effective use of available knowledge. There is no perfect system for modelling carbonate platforms. Numerical modelling can help with both of these problems because it provides effective methods for making qualitative data available for numerical modelling and allows effective exploration of the behaviour of systems through experimentation, thereby increasing our understanding.