Time-series analysis of a heterolithic, ripple-laminated deposit (Early Jurassic, Tilje Formation) and implications for reservoir modelling

A strongly heterolithic lithofacies from the lower part of the Tilje Formation is analysed using time-series analysis. We use, in addition to classical spectral analysis, a wavelet method that detects where in the time-series the periodic components are present. This allows us to evaluate the relative influence of various depositional processes as a function of time or depth. Based on the sedimentary structures, which are interpreted to be combined-flow ripple cross-lamination overlain by either graded rhythmites or fluid muds, and the time-series analysis, we attribute the detected periodic components to several (self-) oscillating processes. The transformation from depth-related data to time-related data is necessary to proceed from an observed geometric cycle to an interpreted time-cycle that can be related to an oscillating process. The periodic signal detected is related to a periodic increase in deposition of sand and mud dominantly supplied by a river system. At the shortest time-scale, a semi-annual increase in river sediment supply is related to a tidal signal that influence the sediment distribution. Other self-oscillating processes, like seasonal variations in river discharge and delta-lobe switching, are also interpreted to have influenced the sedimentary record. However, we realize that an exact reconstruction of the depositional history, based on the available data, is challenging both because several processes have operated simultaneously but also because it is difficult to assign a time-unit to each event. rtical variation in sand fraction (which is the amount of the lithological clean sandstone component relative to the total rock volume) is also quantified by time-series analysis. We show, using published results on 3D bedform modelling, that the observed periodic variation in sand fraction will have a large influence on the permeability of these rocks. This shows that the incorporation of quantified geological data as predictable vertical variation in mud fraction will reduce the uncertainty of the reservoir model.