MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641093920 A) filed by Vardhaman College Of Engineering on August 03, 2026, for Ai-Driven Climate Risk Mapping Using Spatiotemporal Graph Neural Networks.
Inventors include Dr. Sreenivasulu Gogula; Ms. Swapna B; Dr. E Ravi Kumar; Mr. Kachi Anvesh; Mr. Kulkarni Vivek; and Ms. Yadla Sunanda.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: AI-Driven Climate Risk Mapping Using Spatiotemporal Graph Neural Networks is the proposed invention. The invention proposed discloses an AI-driven climate risk mapping system that uses a Geospatial Foundation Model (GeoFM) coupled with Physics-Informed Spatiotemporal Graph Neural Networks (PI-STGNNs) for accurate and interpretable climate risk assessment. The system collects multisource environmental data, such as satellite images, meteorological data, IoT sensor data, hydrological data, digital elevation data, land use data, and historical climate data. The Geospatial Foundation Model learns general spatial representations and then organises them into a dynamic climate knowledge graph that encodes the spatial-temporal interactions among environmental entities. Physics-Informed ST-GNN models capture the evolution of climate dynamics by enforcing physical constraints of atmospheric, hydrological and ecological processes for scientifically consistent predictions. An explainable reasoning engine computes climate hazard probability, vulnerability, exposure and composite risk indices with associated confidence levels. The system creates high-resolution climate risk maps, hotspot analysis, future climate scenarios and adaptive mitigation recommendations. The proposed invention improves prediction accuracy, interpretability, scalability, and decision support for climate resilience planning.
Disclaimer: Curated by HT Syndication.
