USING SYNTHETIC FORWARD SEISMIC MODELS OF CHANNELIZED DEEP-WATER SLOPE DEPOSITS TO INFORM STRATIGRAPHIC INTERPRETATION, TRES PASOS FORMATION, MAGALLANES BASIN, CHILE

Subsurface interpretation of deep-water channels from seismic-reflection profiles is inherently challenging due to the sub-seismic scale of key stratigraphic surfaces and facies transitions. Synthetic seismic modeling can provide valuable insight into multi-scale (bed to composite channel complexes) stratigraphic architecture and are ideal for: 1) understanding uncertainties inherent to seismic interpretation, and 2) assessing the interpretation error from seismic reflectivity profiles at different frequencies. Cretaceous slope channel-fill deposits in the Magallanes Basin of southern Chile were used as the basis for a bed-scale model (0.25 m vertical resolution) of a single 14 m thick by 300 m wide channel element. This channel element was used to generate forward seismic models using acoustic rock properties from analogous deep-water Gulf of Mexico deposits. The results were analyzed to quantify the error in predicting channel thickness and seperability at different frequencies and to quantify how the error scales with true channel thickness, internal channel architecture, and wavelet frequency. Two channel elements were then stacked to elucidate the predictability of stacking patterns between two channel elements. Finally, the analysis was expanded to a series of channel elements that collectively comprise a composite channel complex (~100 m thick). The results of this modeling effort are applied directly to understand interpretation error in subsurface seismic reflection profiles. Single channel elements are tuned at 60 Hz and below, upon which a predictive error model was generated relating true channel thickness to interpreted thickness as a function of frequency. Vertically stacked channels and laterally offset channels are difficult to differentiate in both two- and multi-channel models. While the combination of aggradation and lateral offset offers more insight to the number of channels and predictive morphologies that aid in channel seperability and recognition of channel stacking patterns, identification of individual channel elements is obscured. Furthermore, amplitude analysis shows a positive correlation between increasing frequency and amplitude helping to improve channel seperability and interpretation.