Zhendan Cao | Searcharxiv

arXiv2021

Training Image Selection using Recurrent Neural Networks: An Application in Hydrogeology

Multiple-point geostatistics plays an important role in characterizing complex subsurface aquifer systems such as channelized structures. However, only a few studies have paid attention to how to choose an applicable training image. In this paper, a TI selection method based on Recurrent Neural Networks is proposed. A synthetic case is tested using two channelized training images given the hydraulic head time series. Three different RNNs architectures are tested for the selection performance. Various scenarios of the model input are also tested including the number of observation wells, the observation time steps, the influence of observation noise, and the training dataset size. In this TI selection task, the GRU has the best performance among all three architectures and can reach to a 97.63% accuracy.

physics.geo-ph↗

arXiv2019

Physics-informed semantic inpainting: Application to geostatistical modeling

A fundamental problem in geostatistical modeling is to infer the heterogeneous geological field based on limited measurements and some prior spatial statistics. Semantic inpainting, a technique for image processing using deep generative models, has been recently applied for this purpose, demonstrating its effectiveness in dealing with complex spatial patterns. However, the original semantic inpainting framework incorporates only information from direct measurements, while in geostatistics indirect measurements are often plentiful. To overcome this limitation, here we propose a physics-informed semantic inpainting framework, employing the Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) and jointly incorporating the direct and indirect measurements by exploiting the underlying physical laws. Our simulation results for a high-dimensional problem with 512 dimensions show that in the new method, the physical conservation laws are satisfied and contribute in enhancing the inpainting performance compared to using only the direct measurements.

eess.IV↗