Connectivity-informed Drainage Network Generation using Deep Convolution Generative Adversarial Networks
This work addresses computational bottlenecks in stochastic network modeling for earth and material sciences, offering an incremental improvement in efficiency for generating high-contrast features like drainage networks.
The study tackled the high computational cost of generating drainage networks for statistical evaluation by applying Deep Convolutional Generative Adversarial Networks (DCGANs) to quickly reproduce networks from existing samples, with a novel connectivity-informed method outperforming other approaches by training DCGANs more effectively and better reproducing accurate networks.
Stochastic network modeling is often limited by high computational costs to generate a large number of networks enough for meaningful statistical evaluation. In this study, Deep Convolutional Generative Adversarial Networks (DCGANs) were applied to quickly reproduce drainage networks from the already generated network samples without repetitive long modeling of the stochastic network model, Gibb's model. In particular, we developed a novel connectivity-informed method that converts the drainage network images to the directional information of flow on each node of the drainage network, and then transform it into multiple binary layers where the connectivity constraints between nodes in the drainage network are stored. DCGANs trained with three different types of training samples were compared; 1) original drainage network images, 2) their corresponding directional information only, and 3) the connectivity-informed directional information. Comparison of generated images demonstrated that the novel connectivity-informed method outperformed the other two methods by training DCGANs more effectively and better reproducing accurate drainage networks due to its compact representation of the network complexity and connectivity. This work highlights that DCGANs can be applicable for high contrast images common in earth and material sciences where the network, fractures, and other high contrast features are important.