CO2 geo-storage is a promising approach in reducing greenhouse gas emissions and controlling global temperature rise. Although numerous studies have reported that offshore saline aquifers have greater storage potential and safety, current suitability evaluation models for CO2 geo-storage primarily focus on onshore saline aquifers, and site-level evaluations for offshore CO2 geo-storage remain unreported. In this study, we propose a framework to evaluate the site-level offshore CO2 geo-storage suitability with a multi-tiered indicator system, which considers three types of factors: engineering geology, storage potential, and socio-economy. Compared to the onshore CO2 geo-storage suitability evaluation models, the proposed indicator system considers the unique conditions of offshore CO2 geo-storage, including water depth, offshore distance, and distance from drilling platforms. The Analytic Hierarchy Process (AHP) and Fuzzy Comprehensive Evaluation (FCE) methods were integrated and applied to the analysis of the Ying–Qiong Basin, South China Sea. The results indicated that the average suitability score in the Yinggehai Basin (0.762) was higher than that in the Qiongdongnan Basin (0.691). This difference was attributed to more extensive fault development in the Qiongdongnan Basin, suggesting that the Yinggehai Basin is more suitable for CO2 geo-storage. In addition, the DF-I reservoir in the Yinggehai Basin and the BD-A reservoir in the Qiongdongnan Basin were selected as the optimal CO2 geo-storage targets for the two sub-basins, with storage potentials of 1.09 × 108 t and 2.40 × 107 t, respectively. This study advances the methodology for assessing site-level potential of CO2 geo-storage in offshore saline aquifers and provides valuable insights for engineering applications and decision-making in future CO2 geo-storage projects in the Ying–Qiong Basin.
Data-driven deep learning models usually perform well in terms of improving computational efficiency for predicting heat transfer processes in heterogeneous riparian zones. However, traditional deep learning models often suffer from accuracy when data availability is limited. In this study, a novel deep transfer learning (DTL) approach is proposed to improve the accuracy of spatiotemporal temperature distribution predictions. The proposed DTL model integrates the physical mechanisms described by an analytical model into the standard Deep Neural Networks (DNN) model using a transfer learning technique. To test the robustness of the proposed DTL model, the influence of the number of observation points at different locations, streambed heterogeneity (𝜎²lnK =0, 0.2, 0.5, and 1.0), and observation noise levels (𝜎𝑁𝑜𝑖𝑠𝑒 =0.025, 0.05, 0.075) on the MSE values between the observed and predicted temperature fields. Results indicate that the DTL model significantly outperforms the DNN model in scenarios with scarce training data, and the mean MSE values decrease with increasing observation points for both DTL and DNN models. The mean MSE values for both the DTL and DNN models approach zero as the number of observation points increases to 200, indicating that both DTL and DNN models perform satisfactorily. Furthermore, increasing 𝜎²lnK and 𝜎𝑁𝑜𝑖𝑠𝑒 raises the mean MSE values of the DTL and DNN models, with the DTL model exhibiting greater robustness than the DNN model, highlighting its potential for practical applications in riparian zone management.
Data-driven deep learning models usually perform well in terms of improving computational efficiency for predicting heat transfer processes in heterogeneous riparian zones. However, traditional deep learning models often suffer from accuracy when data availability is limited. In this study, a novel physics-informed deep transfer learning (PDTL) approach is proposed to improve the accuracy of spatiotemporal temperature distribution predictions. The proposed PDTL model integrates the physical mechanisms described by an analytical model into the standard deep neural network (DNN) model using a transfer learning technique. To test the robustness of the proposed PDTL model, we analyze the influence of the number of observation points at different locations, streambed heterogeneity, and observation noise levels on the mean squared error MSE values between the observed and predicted temperature fields. Results indicate that the PDTL model significantly outperforms the DNN model in scenarios with scarce training data, and the MSE values decrease with increasing observation points for both PDTL and DNN models. Furthermore, increasing streambed heterogeneity and observation noise levels raises the MSE values of the PDTL and DNN models, with the PDTL model exhibiting greater robustness than the DNN model, highlighting its potential for practical applications in riparian zone management.