Multi-source data fusion combines gravity, magnetics and other data sets to maximize information from these multiple sources.
Gravity, magnetism, seismic reflection, radiometric, and drilling data are commonly used to infer specific properties of rocks such as temperature conductivity, permeability, porosity, density and fracture levels. Interactive multi-image fusion is a tool for facilitating the interpretation of complex information from multiple data sources.
Multi-source data fusion of gravity, magnetic, and remote sensing can add significant value to the structural interpretation because new processing, with resolution superior to traditional methods, can be used where seismic is of poor quality or not available.
We have adapted a range of fusion techniques supporting the fusion of structural, gravity and magnetics data that can provide realistic crustal structural trap architecture.
Our methodology is to decompose the gravity and magnetic images into multi-scale ensembles, fusing the decomposed images based on specific data fusion rules that decide how the sources will be combined, followed by re-assembling the fused image enabling the final gravity and magnetics interpretation.
Quantitative Data Fusion
From a statistical perspective, we determine the degree of dependence between each of the data sources to fuse them in a single estimation for all the rock properties. This ensures the predictive uncertainty, which leads to risk minimization in drilling and better decision-making.
We have developed a statistical data fusion approach using Bayesian methods that are flexible and robust approaches to getting to grips with uncertainty in geological problems, allowing the incorporation of information from various sensor modalities and different types of prior knowledge. These techniques automatically determine statistical dependencies between gravitational and magnetic fields.
Qualitative Data Fusion
We have adopted a new technique (based on Kovesi et al, 2014) that compresses the dynamic range of an aeromagnetic or gravity image while preserving and enhancing local features, extracting local phase and amplitude values of the image, and then reducing the dynamic range, for example by taking the logarithm, and then reconstructing the image using the original phase valuesWe have adapted a range of techniques that compress the dynamic range of an aeromagnetic or gravity image while preserving and enhancing local features.