Objectives/Scope: The performance of ML-based rock properties prediction from seismic with limited and sparse well data is very often inadequate. To address this limitation, we propose a novel automatic well log regularization (ALR) method with specially designed feature augmentation strategy to improve the prediction accuracy. The effectiveness of ALR method is showcased on field data in Malay basin where we successfully predict elastic logs with 30% higher accuracy, while using only 28% less training dataset. Methods, Procedures, Process: The ALR workflow (Figure 1): (1) feature selection and augmentation; (2) training and prediction and (3) prediction optimizations. The workflow starts with predicting any logs type which are available at training but not in blind wells using standard ML workflow for all blind wells (Step 1-2). Then, these intermediately predicted logs at blind well were jointly used as input features together with seismic-derived attributes using a specially designed feature augmentation strategy (Step 3). Finally, Step 1and 2 are then repeated to predict the elastic logs using these augmented input features. Results, Observations, Conclusions: The ALR method was applied on an oil/gas field data in Malay basin to predict elastic logs (AI and SI) at five blind wells from seismic data only and compared to the standard ML workflow. Two wells were used as training (28% of all data). The prediction performance of standard ML workflow (Figure 2a) is poor and can only capture general mean values of the actual AI/SI logs. The results of ALR workflow (Figure 2b) shows 30% better prediction performance compared to the standard ML workflow. In general, the background and high-resolution trend are well captured, and the overall prediction performance is improved using the new proposed prediction method. There are conceivably two explanations for this result: a) the background (low frequency) trend of the well log is properly reconstructed in ALR using only using seismic data. This could mainly lie in the ability of augmented features in better learning the uncertain reflection-reception relationship between seismic data and elastic logs, as well as the spatial/time-varying property of seismic data; (b) The ability to learn meaningful nonlinear feature relationship between input (feature) and output (label) variables with little or no supervision seems to work properly using specially designed feature augmentation. Novel/Additive Information: The ALR method is an ML-based pseudo log generation from seismic data using specially designed feature augmentation strategy. The novel ALR implementation relaxes the requirement of having a massive amount of high-quality labeled data for training and can therefore be applied in areas with limited well data information. ALR method is proven to be highly accurate for direct elastic logs prediction and can potentially be extended to estimate petrophysical properties from seismic data.

Submitted Abstract The aim in characterizing reservoir on seismic is to be able to precisely assess rock and fluid properties from seismic data. Despite several approaches available, geoscientists still facing issue to accurately characterize the elastic & reservoir properties from seismic. This mainly because of the implementation of simplified linearized algorithms and assumptions which unable to fully address problem with non-linearity and non-uniqueness solution. Besides, the conventional approaches also typically take a very long turnaround time due to extensive computational works. Hence, implementation of deep learning-based seismic inversion with incorporation of a generalized rock physics knowledge for elastic properties prediction perhaps might be able to address the issues. The proposed approach consists of 4 main phases. The first phase is to develop rock physics understanding based on in-situ available reference well data. In the next phase, realistic synthetic dataset library at various pseudo well location is generated that mimic the actual subsurface reservoir condition and the corresponding seismic data from those pseudos well library is subsequently simulated. The third phase is to implement U-net deep learning network architecture with residual blocks of ResNet-18 for training, validation, and testing, utilizing synthetic data library generated in the previous phase. The fourth phase is to apply the pre-trained predictive network model on the actual seismic dataset as input for rock properties estimation. The robustness of the proposed approach is first evaluated based on synthetic dataset. Exceptionally good coefficient cross-correlation is obtained after evaluation on the predicted and true elastic properties; while results on the blind test dataset are slightly less compared to the training performance. This shows that the implemented model architecture with specially designed residual blocks can generalize quite well. The network model able to qualitatively distinguish the properties variation and capture the background trend of each elastic property. Besides, the proof-of-concept exercise also is successfully verified on a fluvial dominated field in Malay basin. Qualitatively, the predictive network model able to capture the background trend with some minor overfitting and identify the relative deviation of the elastic properties. The promising results are also comparable with results produced from commercial software, yet with faster turn-around time and more efficient workflow. In this work, a new approach of deep learning-based seismic inversion is introduced for elastic properties prediction which incorporates a generalized rock physics knowledge and a complex deep learning networks architecture of U-Net with residual block ResNet-18.