MUMBAI, India, Oct. 5 -- Intellectual Property India has published a patent application (202611101009 A) filed by Manipal University Jaipur on August 21, 2026, for Automated Urban Heat Island Intensity Prediction System Using Deep Learning.
Inventors include Siddharth Pal; and Kavita Jhajharia.
The application for the patent was published on October 02, 2026, under issue no. 40/2026.
Abstract: The present invention relates to an automated urban heat island intensity prediction system using deep learning. The system comprises a Data Ingestion and Pre- processing Module that collects NASA SEDAC LST data, handles missing values via alternative-period imputation, performs normalisation to remove geographic and climatic bias, and aligns seasonal hemispheric data; a Landscape Metrics Extraction Module that computes urban area size (SQKM_FINAL), population proxies (ES95POP), daytime and nighttime LST averages for urban and buffer zones, and geographic coordinates (LATITUDE, LONGITUDE); a Multi-Model Prediction Engine comprising RF, XGBoost, LightGBM, MLP, and LSTM with attention mechanism trained to predict D_T_DIFF (daytime UHI intensity) and N_T_DIFF (nighttime UHI intensity); an Evaluation and Comparison Module assessing RMSE, MSE, MAE, and R² across all models; and a Spatial-Temporal Analysis Module generating KDE plots and scatter visualisations to identify high-risk urban zones. The invention provides a way to accurately predict UHI at low cost and on a global scale without the need for extensive sensor infrastructure.
Disclaimer: Curated by HT Syndication.
