Advancing spatiotemporal forecasts of CO2 plume migration using deep learning networks with transfer learning and interpretation analysis

Accurate and timely forecasts of CO2 plume distribution throughout the injection and post-injection phases are crucial for detecting plume migration, assessing leakage risks, and supporting operational decisions in geologic carbon storage (GCS). Current convolutional neural network-based approaches...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:International journal of greenhouse gas control Jg. 132; H. 1
Hauptverfasser: Fan, Ming, Wang, Hongsheng, Zhang, Jing, Hosseini, Seyyed A., Lu, Dan
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States Elsevier Ltd 01.02.2024
Elsevier
Schlagworte:
Auto-Encoder
Encoder-Decoder ConvLSTM networks
Explainable machine learning
Geologic carbon storage
GEOSCIENCES
MATHEMATICS AND COMPUTING
Transfer learning
Geologic carbon storage
Encoder-Decoder ConvLSTM networks
Transfer learning
Auto-Encoder
Explainable machine learning
ISSN:1750-5836
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Accurate and timely forecasts of CO2 plume distribution throughout the injection and post-injection phases are crucial for detecting plume migration, assessing leakage risks, and supporting operational decisions in geologic carbon storage (GCS). Current convolutional neural network-based approaches primarily focus on spatial information and overlook temporal dependencies in plume distributions, thus limiting their ability to capture dynamic movement effects and provide accurate predictions of plume migration. In this work, we propose two deep learning models, Auto-Encoder (AE)-LSTM and Encoder-Decoder (ED)-ConvLSTM, each uniquely designed to capture both spatial and temporal features. We apply the proposed methods to forecast the dynamic distribution of CO2 plumes based on 108 reservoir simulations over a 30-year injection and a 30-year post-injection period. The results indicate that the ED-ConvLSTM model outperforms the AE-LSTM model in accurately predicting the spatiotemporal dynamics of CO2 plume migration, achieving R2 values above 0.99. To provide a deeper understanding of these model predictions, we employ a gradient-based explanation method on the trained models. This approach provides insights into the influence of input variables on plume migration forecasts and uncovers the underlying prediction mechanisms of the proposed models. Furthermore, we introduce a transfer learning technique, enabling fast and accurate plume migration forecasting in the post-injection phase by leveraging the trained model during the injection phase. This reduces the necessity for extensive data collection or re-training. The methods proposed in our work enhances the performance and interpretability of CO2 plume migration forecasts, thereby facilitating informed decision-making throughout the entire lifecycle of GCS applications. •Develop deep-learning models to forecast plume migration in geological reservoirs.•The methods effectively capture spatiotemporal features, enabling accurate forecasts.•Propose a gradient-based explanation method to interpret the prediction mechanisms.•Apply transfer learning for fast and accurate plume migration forecasts.•Our methods aid informed decisions across the geological carbon storage lifecycle.
Bibliographie:USDOE Laboratory Directed Research and Development (LDRD) Program
USDOE Office of Fossil Energy and Carbon Management (FECM)
AC05-00OR22725
ISSN:1750-5836
DOI:10.1016/j.ijggc.2024.104061