MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202611068326 A) filed by Manipal University Jaipur on June 01, 2026, for Zenithvapourcast - An Ai/Ml-Powered Web Platform For High-Resolution Tropospheric Precipitable Water Estimation Using Gnss Zenith-Wet Delay.
Inventors include Dr. Prashant Vats; Abhay Singh Bisht; Dharyansh Achlas; Kriti Khanijo; and Mooksh Jain.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: The present invention relates to a full-stack web system and machine learning pipeline that converts GNSS Zenith Total Delay (ZTD) observations into near-real-time tropospheric Precipitable Water Vapor (PWV) estimates. The web system comprises a data ingestion module to accept GNSS Zenith Total Delay (ZTD) observations from compressed RINEX files or direct coordinate-based API queries; a meteorological co-variate acquisition module to retrieve real-time temperature, pressure, and relative humidity for queried coordinates via an external weather API; a physics-informed feature engineering pipeline configured to encode nine input features including ZTD, meteorological variables, station elevation, and cyclic sin/cos representations of hour and month; a machine learning inference engine comprising a trained XGBoost regressor operating on RobustScaler-normalized features, with automatic fallback to Gradient Boosting Regressor; a joblib-serialized model artifact storing the trained model, fitted scaler, feature list, and validation metrics; and a RESTful API layer with JWT-based authentication exposing PWV prediction endpoints. The system is specifically designed for deployment over ISRO's CORS network of 200+ Indian GPS stations, transforming existing positioning infrastructure into real-time water vapor monitors for monsoon forecasting and flood early warning.
Disclaimer: Curated by HT Syndication.
