A Billion Butterflies: Insights from Dr Jagadish Shukla

Excerpts:

Q. This book beautifully details your achievements in weather and climate science. How did it come about?
A. This book owes its origin to three main points—one personal and two professional. That’s why the book is a combination of both.
Personal: The book is dedicated to my three granddaughters. I wanted them to know that their grandfather didn’t just talk about climate—he wanted to act. They are the ones who will face the most serious consequences of climate change , especially as we continue to ignore recommendations to reduce greenhouse gases.
Professional: First, I wanted to explain to a general audience that weather prediction, seasonal forecasting, and climate change are three distinct areas—each with different methods and mathematics. While weather prediction is familiar and climate change is long-term, my work focused on seasonal climate prediction. At the time, the dominant belief was that the butterfly effect made predictions beyond 10 days impossible. But having grown up in India, where I experienced monsoon droughts and floods lasting an entire season, I couldn’t accept that such patterns were purely random. I believed other factors must be involved.
When I went to MIT, I discovered that the very champion of the butterfly effect was a professor there. So challenging that paradigm was daunting. But my professors encouraged me to pursue the problem. That’s where the title ‘ A Billion Butterflies ’ comes in. The butterfly effect suggests small changes can lead to big outcomes—like a butterfly’s flutter affecting the weather. I asked: what if billions of butterflies fluttered? That’s equivalent to a major shift in initial conditions. I found that ocean temperature was far more influential than countless small perturbations.
This interest in seasonal prediction, especially for the monsoon, shaped my research. I showed that if we know ocean temperatures, we can predict monsoons.
Finally, I come from a poor village with no electricity, drinking water, or proper school—I studied under a banyan tree using kerosene lamps. I walked barefoot three miles to school. Despite that, I believe we all have the inherent ability to achieve something if we don’t give up. I wanted to share my journey so children, especially in underdeveloped regions, wouldn’t lose hope.

Q. You opened up an entire vista for the layperson through your generous sharing and collaboration. How did that approach shape your work?
A. Thank you. I went out of my way to ensure people don’t feel it’s all very complicated. It’s actually simple—we just need the right kind of education and explanation.
On the first page of my memoir, I give an example from my class at George Mason University . I asked my students why nights are colder than days. They all answered, “Because there’s no sun.” I told them that answer would get a B-minus. They were surprised. The real reason is that the Earth is constantly losing energy to space, 24 hours a day. At night, without sunlight to offset that loss, we feel the cooling. In the day, the sun compensates for that loss, so we don’t feel cold.
Understanding climate is about energy balance: the sun gives energy, and the Earth loses it. Over a year, they roughly balance out—that’s why the climate has been stable for thousands of years. But now, with increasing CO₂, that balance is shifting. CO₂ acts like a blanket, trapping energy and preventing it from escaping to space. That’s global warming . So the explanation for why nights are cold and why the climate is changing is deeply connected—and actually very simple.

Q. You talk in greater detail about global warming and how you were a skeptic before joining the IPCC , but later changed your views. Could you talk about that shift?
A. I wasn’t sceptical about the basic physics—that increasing CO₂ warms the planet. My scepticism was about whether we could already say it was happening, because by 1988, the observed warming wasn’t significant enough. My concern was that making bold claims too early could damage the credibility of science.
By 2007, nearly 20 years later, our IPCC panel received the Nobel Peace Prize. That was the first time scientists globally agreed that humans were indeed changing the climate. It took years of data, modelling, and debate to reach that level of confidence.
Some conservatives dismissed global warming as a political agenda, but as you might’ve seen from the book, we debated thoroughly before making any statement. In 2007, we finally had enough evidence to say: yes, human activity is affecting the planet’s climate—an enormous statement considering the scale of Earth’s systems.
The 2015 Paris Climate Conference was one of the best I’ve attended. I was part of the Indian team, and thanks to leaders like PM Modi, former US President Obama, and the Chinese presidency, it was a historic moment. For the first time, nearly 200 countries agreed that humans are driving climate change and committed to action.
Of course, not all countries are doing enough—especially the United States, which has been among the worst performers. And with recent political changes, we’re uncertain about the future. If action is not taken, the consequences could be severe. But right now, politics has taken over.

Q. You spent decades working on the Indian summer monsoon. What were the key innovations and discoveries you made? And are there still aspects you would love to explore further?
A. That’s a very good question—I could write another memoir just to answer it.
As far as weather forecasting is concerned, I didn’t introduce any technical innovation. But I did play a key role in pushing the Indian govt to acquire a supercomputer and advanced models. When former Prime Minister Rajiv Gandhi met President Reagan , and Reagan agreed to provide a supercomputer, I was asked to help set up a weather forecasting centre in New Delhi. I’m proud to say that this centre, which uses a global model, now produces 10-day forecasts that are comparable to many international agencies.
The real innovation I brought was recognising that butterflies—used metaphorically for small disturbances—can’t affect seasonal averages like the monsoon. They can influence weather over a few days, but not an entire season. The weather tomorrow depends entirely on the weather today, and since we never know the exact state of the atmosphere everywhere, we can’t predict the weather perfectly beyond a point.
But even if we can’t predict daily weather beyond 10 days, I argued that we could predict the seasonal average. That was the basis for our paper on Monsoon Predictability , which laid the groundwork for the science of seasonal forecasting. My focus began with the monsoon, but the idea extended far beyond that.
Climate change, of course, is a different challenge altogether. And here’s an important point: we have more confidence in predicting climate 100 years from now than weather 100 days from now. That’s because short-term weather is shaped by initial conditions and chaotic variations, while long-term climate depends mainly on external factors—especially how much carbon dioxide we release and how much deforestation takes place.
Over the last century, solar changes and volcanic eruptions have had only minimal impact on global temperature. The main drivers of recent climate change are human: deforestation and greenhouse gas emissions. Deforestation accounts for about 20 to 25 percent of the impact, while carbon dioxide and other greenhouse gases make up about 75 to 80 percent. That’s why climate change is such a complex problem—it’s driven by activities that are essential to modern life, like transport and heating, yet they also shape the future of the planet.

Q. You mention the relationship between war, weather, and technology, and also note that accurate models and predictions are a matter of national security. Could you elaborate on that connection?
A. The reason India wanted its own supercomputer and weather forecasting model—and why I supported it fully—was because it’s a matter of both national security and building local expertise. Some might argue, why not just rely on the European Centre for Medium-Range Weather Forecasts , which is among the best in the world? But that’s not enough.
National security becomes critical here. For example, during the Falklands War, when Britain attacked Argentina, the Europeans reportedly blocked Argentina’s access to weather forecasts. Some say this severely impacted Argentina’s response—one of their ships was hit by a British torpedo while high-ranking officers were on deck, unaware of the weather conditions.
There are many such examples that highlight why having your own weather prediction system is crucial—especially in wartime. It’s also essential for disaster management. If a cyclone is about to hit, you need precise, hour-by-hour forecasts to prepare and minimise damage. So, accurate and independent weather models are vital not just for science, but for national preparedness and security.

Q. Your book emphasises the importance of accurate data, and that forecasts are only reliable up to around 10 days. In that context, do you think weather apps are actually reliable?
A. You’re absolutely right to raise that question. All these weather apps ultimately depend on the output generated by supercomputers running billions of mathematical equations—something I emphasise throughout my book. These models solve equations every few minutes to produce accurate weather forecasts. Once that data is generated, television channels and apps take it and present it with visuals, animations, and stories—but the core forecast still comes from those scientific models.
Until about two years ago, those supercomputer-based models were the only source of detailed and reliable forecasts. This approach—refining mathematical models over decades—has been one of the great scientific achievements of the 20th century. We’ve steadily improved weather prediction, hour by hour and even minute by minute, thanks to this method.
But two years ago, things changed. Companies like Google and Microsoft introduced artificial intelligence–based methods. These AI systems don’t use physical models at all. Instead, they feed decades of forecast data—generated by traditional models—into AI systems and try to find patterns to make predictions. In a way, it’s like going back to what we were doing 30 or 40 years ago: relying purely on past data, without physical understanding.
That said, AI has produced surprisingly good results. The European Centre for Medium-Range Weather Forecasts—widely considered the best in the world—recently acknowledged that some AI forecasts are now as good as theirs. And AI developers are confident they can surpass traditional models.
This is a completely different kind of revolution. Unlike our approach, which is grounded in physical laws and deep understanding, AI focuses purely on prediction. And for many people—those selling umbrellas or deciding what to plant in a field—understanding doesn’t matter as much as having a usable forecast.
That’s why AI-based apps are exploding in popularity. I mentioned in my book that there are already around 10,000 weather-related apps, and that number is rapidly growing.