Andrea Zerilli

3D inversion of total field mCSEM data: The Santos Basin case study

We present a marine Controlled Source Electromagnetic (mCSEM) 3D interpretation workflow based on anisotropic inversion applied to a dataset acquired in the Santos Basin offshore Brazil as part of a co-operation project between Petrobras and Schlumberger to evaluate the integration of deep reading Electromagnetic (EM) technologies into the full cycle of oil field exploration and development. The mCSEM dataset was acquired to demonstrate the improved detection and delineation of challenging EMtargets such as “smaller and deeper” hydrocarbon filled reservoir zones in complex background using 3D total field data. The project area hosting a proven reservoir was covered by a receiver grid and an orthogonal source lines grid extending beyond the receivers in the in line and cross line directions with all receivers active throughout the acquisition of both the orthogonal sets of source lines. We show that 3D inversion of the mCSEM total field data embedded in an advanced integrated workflow improves our ability to delineate hydrocarbon, their position and thickness resolution and increases our confidence about the resistivity at the reservoir(s) level. This increased resolution can provide in subsequent integrated interpretation workflows detailed information about reservoir volume distributions.

Bringing complex salt structures into focus — A novel integrated approach

Advances in seismic imaging have changed the way we view salt bodies. Once seen as impenetrable barriers to geophysical probing, many salt structures are now proving to be thin blankets or complex geometries shielding rich reserves. Prestack depth imaging and interpretative processes have advanced to a point, where, in many cases, subsalt horizons are imaged as clearly as the areas outside the salt. Subsalt structural as well as stratigraphic interpretations are possible using this technology that is being rapidly adopted by oil companies as a key to reduce risk and improve data accuracy. Currently, its full potential is far from being completely realized. Issues limiting the success of PSDM have included poor data, incorrect geologic models and associated velocity fields, inaccurate processing algorithms, anisotropy, near-surface and topographic effects, lack of a true amplitude solution, computer power and overall cost. In this paper we discuss how alternative geophysical data can be fully integrated in the cycle of iterative depth migration anisotropic model updating to develop a higher resolution earth model, that can be used for improved seismic imaging. We illustrate in a case history of a salt structure in densely populated Northern Germany how additional geophysical measurements that focus on density and resistivity contrasts can significantly improve the seismic interpretation. The integrated modeling of high resolution gravity and magnetotelluric data leads to a new and more reliable model.

Lineaments analysis for potential‐fields data using neural networks

ture to be analysed (structural characteristies) to determine a consistent set of If, for some p,I pI Y (2) specifying the connections and thresholds of a symmetrical Neural Network of binary threshold units. The Network evolves, starwemarkthe centralpixelof the neighbourhood ting from an initial state, to a system as an edge pixel. configuration with higher values of the We can avoid the problem of fixing the consensus function. Network evolution can be parameter Y, following the classification deterministic or stochastic. method proposed by Ulupinar and Medioni Tests have been conducted on both synthetic (1988). In this case we replace (2) by: and real data with very good results.

Improving Magnetotelluric Data Degraded By Correlated Noise With Robust Regression Analysis And Recurrent Neural Networks

Over the past few years Magnetotelluric (MT) has seen extensive use in Italy for operational hydrocarbon exploration. MT is been applied . successfully for both reconnaissance and deta i led permit evaluation in a variety of difficult environments such as complex geology, rugged topography and high cultural noise. One important aspect of technical concern has been data processing. Results obtained from conventional methods are inadeguate when data quality is poor. One main question has been how to reduce noise in the data as much as possible. The remote reference processing has turned out to be very effective, although not definite in improving data quality. Various sort of robust data adaptive weighting scheme (Stodt, 1986, Sutarno and Vozoff, 1991, Larsen et al., 1996) have been applied and have improved the situation greatly. However, all to often there are still critical areas and/or frequency ranges where useful data are difficult to obtain. Typically, these difficulties are most severe if there are outliers or strong correlated noise signals due to electric train signals and active cathodic protection of pipelines in most data sections. Here we report on our efforts to develop an improved method to obtain interpretable data in these difficult areas. Our approach is a combination of use of a very far fixed remote reference tested to be free of regional correlated noise sources and of a novel processing approach based on: selection of subsets of data (based on coherence) followed by filtering of these subsets using a reweighted least median of squares for the E M 1 . 8