PHYSICALLY-BASED SIMULATION OF PALEOTOPOGRAPHY FROM LIMITED OBSERVATIONS

Lithologic boundaries can be produced in a number of ways with each way imprinting a geometrical signature of the formative process. In the case of preserved paleotopographies, one expects topographic attributes such as drainage networks, hillslopes, and valleys to be present in the boundary. These features can be important to groundwater flow and contaminant transport. The elevation of a boundary throughout a region is usually inferred by interpolating between known elevations at a limited number of locations. Unfortunately, standard interpolation methods such as kriging and splines do not reproduce draining valley networks. In this work, we present a physically-based method to simulate a paleotopography based on known elevations at wells and observed attributes at exposed contacts. The method is based on an available topographic interpolation technique that calibrates a simple landscape evolution model to reproduce a collection of known elevations while preserving a dendritic drainage pattern. Here, the method is extended to allow inflow from outside the domain of interest and is used to simulate a largely-buried paleotopography. In particular, the method is applied to the base of the Tshirege Member of the Bandelier Tuff at the Los Alamos National Lab in New Mexico. This boundary is thought to be a paleotopography formed on material from an ash flow and preserved by a subsequent ash flow. From contacts that are exposed in modern arroyos, the boundary is known to exhibit paleochannels and approximately parabolic hillslope profiles. The elevation of this surface is simulated based on 49 wells over a 25 square mile region. The model parameter controlling the hillslope scale is estimated from the exposed contacts, while the parameter controlling the concavity of the long-profiles of channels must be estimated from the current topography. The latter parameter has a relatively weak influence on the results. The method is applied with varying initial conditions to produce a set of viable simulated surfaces, which are used to probabilistically estimate the direction of flow on this surface throughout the simulation domain.