The F-Area Seepage Basins at the U.S. Department of Energy (DOE)’s Savannah River Site (SRS) have been a long-standing source of groundwater contamination due to the disposal of low-level radioactive waste during the Cold War. To mitigate further contamination, the basins were dewatered and sealed with low-permeability caps designed to prevent rainwater infiltration and minimize migration of contaminants into the groundwater. Monitoring the structural integrity of these caps is crucial to maintaining the effectiveness of remediation efforts. Electrical Resistivity Tomography (ERT) has been deployed to detect water seepage through the cap by measuring changes in electrical conductivity beneath the cap. However, although ERT data collection and processing can be automated, analysis of the resultant images requires human effort, which is time-consuming and resource-intensive over the timescales required for long-term monitoring. An AI-based automated system was developed to process and analyze ERT images in real time. This system identifies anomalies in conductivity values, locates potential seepage areas, and evaluates their severity. The results are automatically visualized in an interactive 3D model, enabling stakeholders to efficiently assess cap integrity and determine whether further investigation is warranted. By eliminating the need for daily manual data analysis, this approach significantly reduces monitoring costs and enhances early detection capabilities. The AI system is adaptable to any ERT-based monitoring application beyond the F-Area Seepage Basin caps, enabling the detection, localization, and assessment of anomalies across a wide range of domains, both within and beyond the DOE complex.

Dynamic topic modeling is useful at discovering the development and change in latent topics over time. However, present methodology relies on algorithms that separate document and word representations. This prevents the creation of a meaningful embedding space where changes in word usage and documents can be directly analyzed in a temporal context. This paper proposes an expansion of the compass-aligned temporal Word2Vec methodology into dynamic topic modeling. Such a method allows for the direct comparison of word and document embeddings across time in dynamic topics. This enables the creation of visualizations that incorporate temporal word embeddings within the context of documents into topic visualizations. In experiments against the current state-of-the-art, our proposed method demonstrates overall competitive performance in topic relevancy and diversity across temporal datasets of varying size. Simultaneously, it provides insightful visualizations focused on temporal word embeddings while maintaining the insights provided by global topic evolution, advancing our understanding of how topics evolve over time.