Coastal zones are where riverine, atmospheric, oceanic, and human systems converge to drive flooding, erosion, and water-quality change — and where practical problems require innovative solutions to: what is happening now, how will the ecosystem change in the near future, and what is the impact of different management or socioeconomic decisions? These what-now, what-next, and what-if questions are exactly what an Earth Science Digital Twin is meant to answer, by maintaining a dynamic, interactive representation of an ecosystem kept in step with a continuously updating record of observations, model inputs, and impact assessments.
The Coastal Zone Digital Twin (CZDT), funded by the NASA Earth Science Technology Office (ESTO), is an extensible digital-twin capability for coastal flooding, water quality impairment, and impacts. This presentation will describe the continuing enhancements to CZDT including the broad use of remote sensing inputs, machine learning-based predictive modeling, and the application over the Eastern U.S. A central emphasis of CZDT is enabling users to interact directly with models for what-if scenario impact assessments. In addition, we are developing a data-wrangling agent that speeds the onboarding of new data sources as Essential Variables (EVs): retrieving, ingesting, and normalizing them into the digital replica. For the front-end, we are developing an LLM-based agentic layer that lets users engage through natural language and composes the CZDT subsystems on demand by assembling those needed to answer a given question rather than running a fixed pipeline. For scientifically defensible output, the architecture emphasizes standards-based interfaces, provenance tracking, and communication of confidence/uncertainty, and is designed to support future federation with partner digital twins and models. Together, these advances are intended to improve the fidelity, timeliness, access, and breadth of coastal information products, enabling richer characterization of processes that govern coastal flooding and water quality, their impacts, and a wider set of decision-relevant scenarios.