MoES Secretary Calls for AI-Physics Fusion to Predict India’s Next Climate Extremes

AI, he explained, provides the missing capability to interpret these fine-scale “time series” variations, enabling more precise forecasting where lives and infrastructure are most at risk.

In a strong call for a new era in meteorological science, Dr. M. Ravichandran, Secretary, Ministry of Earth Sciences (MoES), today urged the strategic integration of Artificial Intelligence (AI) with traditional physics-based weather models to address the rising unpredictability of climate extremes across India.

Speaking at the high-level panel discussion, “Harnessing AI to Manage Climate Extremes and Build Sustainable Systems,” held on the final day of the India AI Impact Summit 2026 at Bharat Mandapam, Dr. Ravichandran emphasized that climate change is fundamentally reshaping the way forecasting systems must operate.

“Traditional models help us see the big picture, but AI will help us understand the local rhythm of extreme events,” he said, underlining the urgent need for next-generation forecasting tools.

From Tracking the Elephant to Tracking the Ant

In one of the most striking moments of his address, Dr. Ravichandran offered an analogy that captured the evolving complexity of weather prediction in the climate era:

“Earlier, we only had to track the elephant—the large-scale weather systems. Today, because of climate change, we must also track the ant sitting on that elephant.”

The metaphor illustrated a major scientific challenge: while numerical weather prediction models excel at broad spatial forecasting, they struggle with hyper-local, time-sensitive extremes such as cloudbursts, flash floods, and sudden convective storms.

AI, he explained, provides the missing capability to interpret these fine-scale “time series” variations, enabling more precise forecasting where lives and infrastructure are most at risk.

Why Forecasting Must Evolve Now

India is witnessing a sharp rise in high-impact climate events—from Himalayan cloudbursts to urban flooding in megacities. These extremes demand systems that can deliver:

• Faster local warnings

• Kilometer-scale precision

• Reduced uncertainty in model outputs

• Better disaster preparedness

Dr. Ravichandran noted that the solution lies not in replacing physics-based models, but in fusing AI with them, creating a hybrid forecasting architecture.

Reducing Errors by Improving Initial Conditions

A major limitation of conventional numerical models is their dependence on multiple assumptions, which can lead to error accumulation over time.

Dr. Ravichandran highlighted that AI can help:

• Reduce model bias

• Correct errors early in simulations

• Improve initial atmospheric conditions

• Enhance prediction accuracy for extreme events

This hybrid approach could significantly strengthen India’s early warning systems.

Unlocking 150 Years of India’s Meteorological Legacy

India possesses one of the world’s richest climate archives through the India Meteorological Department (IMD), spanning nearly 150 years of observational weather data.

Dr. Ravichandran called for this vast repository to be opened up to:

• Young researchers

• Data scientists

• Multi-disciplinary experts

• Innovation-driven institutions

Such access, he noted, could accelerate breakthroughs in climate resilience and forecasting innovation.

AI Downscaling: Forecasting at 1-Kilometre Resolution

One of the most promising advances discussed was AI’s ability to “downscale” large-scale models into highly localized forecasts.

Dr. Ravichandran identified 1-km resolution forecasting as a game-changer for:

• District-level disaster response

• Agricultural advisories

• Urban flood risk prediction

• Localized heatwave alerts

This capability could redefine weather protection systems across India.

Building Trust Through Validation and Ethics

Addressing concerns around AI reliability, Dr. Ravichandran stressed that trust is the foundation of any public forecast system.

He advocated for rigorous:

• Validation

• Verification

• Ethical safeguards

• Transparent benchmarking

to ensure AI-generated insights are dependable for public safety and decision-making.

A Call to Break Scientific Silos

Dr. Ravichandran concluded with a strong appeal for cross-sector collaboration:

“We shouldn’t think in one way. We need biology experts, data scientists, and researchers from multiple disciplines to look at our data through different lenses.”

He emphasized that only a data-driven, multi-disciplinary approach can build a climate-resilient India capable of managing future extremes.

Distinguished Panel of National and Global Experts

The session brought together leading voices from governance, research, industry, and global academia, including:

• Er. Manish Bhardwaj, Secretary, NDMA

• Dr. Shivkumar Kalyanaraman, CEO, Anusandhan National Research Foundation

• Prof. Amit Sheth, Founding Director, IAIRO & Professor, University of South Carolina

• Dr. Praphul Chandra, Dean (Research), Atria University

• Dr. Karthik Kashinath, Distinguished Scientist & Engineer, NVIDIA (USA)

• Prof. Dev Niyogi, UNESCO Chair, University of Texas at Austin

• Sandeep Singhal, Senior Adviser, Avaana Capital

• Dr. Akshara Kaginalkar, Professor of Practice, Centre for Climate Change, Atria University

Organisers

The panel discussion was jointly organized by:

Indian AI Research Organisation (IAIRO), Atria University, C-DAC, IITM/MoES, and LokNeeti.

Towards a Climate-Resilient India

The discussion underscored a national imperative: combining India’s deep scientific expertise with AI-driven innovation to create forecasting systems that are faster, finer, and more reliable—capable of protecting communities in an era of intensifying climate extremes.