Abstract
<jats:p>The purpose of this work is to develop a digital twin architecture of an aquatic ecosystem to preserve environmental sustainability, including monitoring, forecasting, and management of water resources in the context of anthropogenic impact and climate change. The paper proposes a modular digital twin architecture that includes a module for collecting and combining multimodal data, digital ecosystem models, an expert system, a regulatory framework module, a geographic information system, and a graphical data representation module. The architecture is based on the integration of data from sensors, satellites, weather stations, and remote sensing systems, using machine learning models and large language models to analyze regulatory documentation. The developed architecture of the digital twin of a water body allows monitoring ecosystem parameters in real time, assessing compliance with regulatory requirements, and generating reports and recommendations for organizations involved in natural resource management and governing bodies. The effectiveness of various models of legal text processing was tested, and optimal approaches to their semantic analysis were determined. The expediency of integrating the digital twin into the state environmental management system is substantiated. The digital twin architecture proposed by the authors is a comprehensive tool for sustainable water resources management that allows predicting the state of water bodies, timely identifying risks and developing scientifically sound management solutions, thereby reducing environmental threats and preserving water resources.</jats:p>