MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202621073886 A) filed by Symbiosis International Deemed University on June 12, 2026, for Hybrid Real-Time Rainfall Prediction System Integrating Neural Network And Numerical Weather Prediction Models.
Inventors include Dr. Princy Diwan; Akshan Motghare; and Siddhant Virmani.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT HYBRID REAL-TIME RAINFALL PREDICTION SYSTEM INTEGRATING NEURAL NETWORK AND NUMERICAL WEATHER PREDICTION MODELS A hybrid real-time rainfall prediction system (100) for generating locally calibrated daily precipitation forecasts is disclosed. The system (100) comprises a data acquisition module (120) that receives real-time meteorological data from an external numerical weather prediction API and historical training data, a feature engineering module (125) that derives 21 engineered features from raw meteorological variables, a two-stage prediction engine comprising a calibrated gradient boosting classifier (130) with Platt scaling (132) for rain occurrence probability estimation and seasonally decomposed gradient boosting regressors (140) for conditional precipitation quantity estimation, a hybrid blending module (150) that combines machine learning predictions with NWP forecast output using physics- informed decision rules and configurable blending weights, and a web-based presentation layer (160) providing a dashboard with seven-day rolling forecasts and comparative analytics. [
Disclaimer: Curated by HT Syndication.
