**Estimation of Hydraulic Conditions of Turbidity Currents by Inverse Analysis; Application on Cretaceous Panther Tongue Deposits, Utah, USA**

**Hajime Naruse ^{1} and Cornel Olariu^{2}**

^{1}Faculty of Science, Chiba University, Chiba, Japan^{2}Jackson School of Geosciences, University of Texas at Austin, Austin, Texas, USASediment-laden density flows called turbidity currents intermittently occur in delta foresets or submarine canyons, transporting sediments form shallow to deeper regions and forming turbidite deposits. Actual hydraulic conditions of turbidity currents such as flow velocity or sediment concentration has been hardly clarified because the flows are subaqueous so that it is quite difficult to observe directly.

Here, we attempted to establish the method to estimate hydraulic conditions of turbidity currents by inverse analysis of ancient turbidite deposits. Numerical simulation can predict longitudinal thickness profile of turbidite beds with prescribed initial conditions such as slope inclination, flow velocity, flow thickness, sediment concentration, and so on. We calculated turbidite bed profiles with arbitrary initial conditions x, and then quantify the degree of difference of thickness profiles between the calculated bed thickness and the actual turbidite bed thickness, which is subject to analyze, as a model evaluation function W(x). We find the combination of initial conditions x that minimize the function W(x), and we consider the discovered parameters x as a condition that actualize deposition of the natural turbidite deposits. In other word, our inverse analysis is a non-linear programming problem that search the set of the hydraulic condition under which the actual turbidite beds deposited.

To test our method, we examined the deposit of river-generated turbidity (hyperpycnal) currents in the Panther Tongue Sandstone, Utah, USA. The investigated hyperpycnal beds were inferred to have deposited on the foreset of the delta. The hyperpycnal facies links with deposits of the channel-moth bar complex at their upstream end, and pinch out basinward at the base of the distal delta front. River-generated turbidity (hyperpycnal) currents are generally thought to be quasi-steady flow so that we can approximate the flow behavior as the steady flow in our calculation. We employed the 4 equation one-dimensional model of turbidity currents as a forward model, and assumed that sediments are composed of fine sands and wash-loads. The square mean of difference between calculated and observed hyperpycnal beds thickness were used as a model evaluation function. This evaluation function is generally multi-modal. Therefore we firstly searched the parameters using the genetic algorithm, and then obtained the optimized solution (initial conditions) by conjugate gradient algorithm.

Using of inverse analysis, we estimated that the river-generated turbidity current that formed a hyperpycnal bed in the Panther Tongue Sandstone had around 2 vol. % in concentration and 1 to 4 m/s in flow velocity. Calculated flow conditions seem to be reasonable, although the duration of the deposition from the flow was 30 minutes to 2 hours, which is relatively short compared with modern flood events. The obtained parameter set of initial conditions could not be convergent to a single value and numerous sets of initial conditions for the model produced the equivalently optimized beds. Thus, it is necessary to evaluate the range of values of our estimations against modern systems, and progress both of the forward model and the evaluation function in the future development of the method.

AAPG Search and Discovery Article #90079©2008 AAPG Hedberg Conference, Ushuaia-Patagonia, Argentina