Abstract: Neural Network Processing of High Sensitivity Aeromagnetic Data and Integration with Subsurface Geology, Paleozoic Play, Nebraska Panhandle, USA
High sensitivity aeromagnetic data, processed by neural network techniques, and displayed in shadow-graph format, have revealed textures and fabric of Precambrian basement that have influenced sedimentation, diagenesis, and petroleum accumulation in Paleozoic carbonates of the Nebraska panhandle.
A magnetic survey was flown at 300^prime ground clearance on a 0.5 by 1 mile grid spacing. An optically pumped cesium vapor magnetometer sampled continuously at 114 second intervals. Magnetic signals due to cultural features were removed from the data by filtering operators designed to eliminate "spikes" from the dab Flight line photos were used to cross-check cultural filtering.
Neural network processing involves training a computer program to "recognize" a characteristic magnetic signature by successive iterations and comparison to that of a known structural model. Error between the calculated and desired output is determined and back-propagated into the neural network, allowing closer approximations during successive iterations. In this manner, the program "learns to recognize" a specific pattern of magnetic signals that have structural significance. The neural network program was trained on a series of magnetic signatures of hypothetical faulted basement blocks of similar size and displacement to those underlying known producing fields. Following training of the neural network, the aeromagnetic data were analyzed by the program, and a series of possible st uctural anomalies was identified.
Shadow graphs were constructed from the magnetic data by "shining" an imaginary sun on the second vertical derivative, reduced to pole data. As the orientation of the "sun" was varied, shadows were created that highlighted both major and minor variations in magnetic intensity. Proximal edges of shadows indicate changes in slope of the magnetic field. These slope changes represent both compositional and structural features of the Precambrian basement. Most larger features are due to large-scale compositional variations within the Precambrian basement. These are represented by relatively continuous shadows, and have a northeasterly trend. Smaller features are most likely due to small-scale structural displacements of basement blocks, and are represented by relatively discontinuous shado s that trend northeast, northwest, and north. These smaller features are similar in scale to block-faulted basement structures that have been identified by means of seismic data and have been successfully explored for hydrocarbons in the Nebraska Panhandle.
AAPG Search and Discovery Article #90959©1995 AAPG Rocky Mountain Section Meeting, Reno, Nevada