Research in AI & Modeling at STORM Lab focuses on developing data-driven and hybrid approaches to understand, retrieve, and predict atmospheric processes, with particular emphasis on precipitation systems and multi-sensor observations.

The lab has extensively used artificial neural networks (ANNs) for rainfall retrieval, classification, and enhancement of observational datasets. These include methodologies to estimate surface rainfall rates from combined radar–radiometer observations, upscale ground radar measurements to match satellite resolution, and retrieve three-dimensional rainfall structure from multi-sensor data .

A major contribution of the lab lies in integrating machine learning with physical and observational frameworks, enabling improved consistency between satellite and ground-based measurements. These approaches help address challenges such as resolution mismatch, data sparsity, and uncertainty in precipitation estimation.

The lab has also explored information-theoretic and data-driven techniques to quantify relationships between different sensing systems, enhancing cross-retrieval capabilities and improving the reliability of atmospheric datasets.

More recently, the research has expanded to include AI-based anomaly detection and early warning systems, such as machine learning frameworks for forest fire detection and environmental monitoring, demonstrating the broader applicability of data-driven methods in Earth system science .

Overall, the AI & Modeling research at STORM Lab emphasizes the integration of machine learning, physical insight, and observational data to develop robust, scalable, and interpretable models for atmospheric and environmental applications.