Densely Connected Squeeze‐and‐Excitation Convolutional Encoder‐Decoder Networks for Identifying Preferential Channels in Highly Heterogeneous Porous Media. Issue 9 (15th September 2022)

• Record Type: Journal Article
• Authors: Zhou, Zhengkun; Shi, Liangsheng; Zha, Yuanyuan; Wang, Shuixian; Xu, Baokun; Tian, Lei; Zhang, Lanhui; Tian, Jie; Yang, Ruiting
• Is Part Of: Water resources research. Volume 58:Issue 9(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-09-15
• Subjects: deep learning -- connectivity -- hydraulic conductivity field -- heterogeneity -- solute transport; Hydrology -- Periodicals; 333.91
• DOI: 10.1029/2021WR031429
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: In highly heterogeneous formations, the preferential channels (PC) that usually exist can dramatically change the transport dynamics of a solute plume. Identifying PC is vital to delineate contaminant transport and conduct the environmental risk analysis. Conventional methods usually depend on the solution of the governing equations for flow and transport and struggle with high computational costs. In this paper, we transform PC identification into a particular semantic segmentation task for the first time and build a densely connected squeeze‐and‐excitation convolutional encoder‐decoder network. A training strategy combining segmentation loss and new regression loss (index distance loss) is used to improve the identification of PC locations. We evaluate the effectiveness and generalizability of the proposed deep learning model with conductivity fields with different variances. A new connectivity indicator (connectivity ratio) is proposed to describe the connectivity area of a field. The network trained by conductivity fields with larger variances (higher heterogeneity) produces better PC identification and has a stronger generalizability. A conductivity field with a large variance usually has small connected areas and induces tortuous and long PC. The small connected areas restrict the generation of multiple PCs. This makes the model learn discriminative features easily. And the long PC increase its length, which means an increasing of the number of positive pixels in training samples. Finally, we compare the model with the stream function and minimum resistance methods. The results show that the deep learning model is an efficient method and can provide more connectivity information than the other two methods. Plain Language Summary: Spatial distribution of hydraulic conductivity in porous media has a key role in controlling contaminant transport. As the heterogeneity of the spatial distribution increases, the size of low‐permeability areas increases and the hydraulic conductivity field tends to form high hydraulic conductivity channels, which is called preferential channels. These channels act like pipelines in which the contaminant plume moves quickly. They are a key factor governing the early time arrivals of contaminant plume and are correlated with the generation of long tails in contaminant plumes. Therefore, identifying the preferential channels from the hydraulic conductivity field is fundamental for risk assessment. In this work, we identify the preferential channels from the hydraulic conductivity field through a deep learning model. The identification of preferential channels is conducted as the segmentation task in computer vision. We build three data sets with different levels of heterogeneity in spatial distribution of hydraulic conductivity. We show that the model which is trained by data sets with higher heterogeneity tends to have better generalization performance. We compare the model with two other methods. We show that the deep learning model can provide more connectivity information than the other methods. A new connectivity indicator is proposed to characterize the connected areas in conductivity fields. Key Points: We transform preferential channel identification into a semantic segmentation task Preferential channels are identified only through the spatial distribution of hydraulic conductivity A new connectivity indicator is proposed to describe the connectivity of a conductivity field