Rainfall in India rarely behaves the way the calendar promises. A monsoon can drench one district and bypass its neighbour entirely; a ‘rainy’ season can pass with barely a drop, only for an entire month’s worth to fall in an afternoon. Behind this unevenness lies a stubborn scientific gap — weather radar, for all its sophistication, still estimates rainfall using raindrop size distribution as an assumed input rather than a measured one. It was this quiet, largely invisible problem that set Prof Kirti Chandra Sahu of IIT Hyderabad’s Department of Chemical Engineering on a research path that would eventually earn him a place among the 21 recipients of the National Awards to Teachers 2026, conferred by the Ministry of Education for excellence in teaching, research, innovation and mentoring. In this conversation, he walks us through the making of that facility, the physics of a falling raindrop, and what it will take to give India rainfall models built for its own sky.
Excerpts
Congratulations on the National Award. What went through your mind when you learned you had been selected for this award?
I was obviously very happy because teaching students is our profession. Although we also do research, at an institute like IIT Hyderabad, classroom teaching, research, PhD guidance and student mentoring all play an important role in the selection process. I was very happy, but at the same time, I was thinking about what more I should do in teaching, research and mentoring.
The award recognises your contributions across teaching, research, innovation and mentoring. Which aspect do you feel most personally connected to, and why?
Even in teaching, I teach students from a research point of view because at IIT Hyderabad, we get the brightest among the brightest students. If you only teach what is there in books, they will not be excited. So even my classroom teaching involves a component of research. I would consider the research component including research within teaching — to be what I am most excited about.
Your research on the shape and size distribution of raindrops at different altitudes, which contributes to more accurate rainfall prediction, was one of the things that led to the National Award. Can you take us through the research?
India is a diverse country, particularly climatically. Hyderabad’s weather may be different from Warangal, Odisha or Assam. The geophysical and geological conditions are also different — some places have rivers, some have hills. Although we receive rainfall because of the monsoon, different parts of the country receive different amounts. Because of climate change and other factors, most parts of the country are now experiencing uneven rainfall, localised in space and time. A region may receive a large amount of rainfall in a short time, while sometimes there is no rainfall even during the rainy season.
This disturbed me around 2013, when I was starting my career. I had worked in fluid mechanics from the beginning, but this was not an area I had expertise in at that time. Still, I thought I could contribute. I read several hundred research papers and tried to identify the gaps — the questions that had not been answered.
I found that weather radar measures rainfall using a formula. It sends an electric signal, receives the reflectivity and derives rainfall from it, but uses drop size distribution as an input parameter. Raindrop size distribution is actually an unknown parameter; it is based on assumptions or statistical data rather than being scientifically derived. I thought that was the question that needed to be answered.
I started working on the problem and ran numerical simulations of droplets, but they were not answering my key questions. When a raindrop starts falling from a cloud, the conditions in the cloud are different — the temperature is below zero degrees Celsius and the humidity is supersaturated. At ground level, on a rainy day, the temperature could be 35 degrees Celsius and humidity as low as 10–20%.
Raindrops are deformable. They can coalesce, merge, fragment into different pieces, evaporate and condense. The dynamics are therefore quite complex as a droplet travels from cloud to ground through changing atmospheric conditions.
I wanted to study this, but cloud height is around 15–20 km. To study these microphysical processes directly, you would need a wind tunnel 15 km high, which is impossible. So I decided to take the reverse approach.
You might have seen people blow air through a whistle to keep a ping-pong ball suspended in the air. Rather than moving the drop from the cloud to the ground, I could make the drop stationary by blowing air upward — it is an inverse problem.
We condition the air so that the suspended droplet experiences the local conditions, as if it were actually falling from the cloud to the ground. Since we know the temperature and humidity gradients, we condition the inlet air accordingly. Over time, the suspended droplet experiences the atmospheric conditions it would encounter while falling.
That is the idea behind the Raindrop Research Facility at IIT Hyderabad. It is a unique facility, and no such facility is available anywhere, not just in our country but across the globe. We developed it from scratch — myself, my students and my collaborators — designing, fabricating and testing it. We are now using it to obtain rainfall rates for different parts of the country.
Can the research eventually help improve how India prepares for extreme rainfall, cloudbursts or flooding?
Eventually, it should. There are three ways of handling any such phenomenon. First, you need a fundamental understanding of it. Second, you should be able to predict what is going to happen after, say, 15 or 20 days. Third is control.
Right now, we are developing the fundamental understanding and obtaining empirical formulas for weather radar. It should soon be able to contribute to weather prediction — whether an extreme event is coming or not.
Unfortunately, these phenomena occur very quickly. A cloud can form in about 20 minutes, so it is a very complex process. But even getting close to prediction would eventually help with control. We are not yet at the stage of controlling or fully predicting these events, but I think this study will definitely help with weather forecasting and climate modelling. Once these studies are coupled with climate modelling, it should become possible to predict the weather and, eventually, control it.
Can the data generated through the facility eventually feed into computational models or AI-based prediction?
Absolutely. This should be done not only through AI, but by combining it with satellite data and direct, in-situ measurements.
The Indian government undertook initiatives in 2011 and 2015 to predict rainfall in different parts of the country. We published an article on this in 2025. We were able to predict rainfall for Bhubaneswar, Vijayawada and Hyderabad and compared it with the government’s Cypex project. We showed that data produced in our experiments could validate the in-situ measurements conducted by the Indian government.
The data generated in our facility for different parts of the country should be coupled with atmospheric or climate science modelling. It is a global phenomenon that affects things locally. There is no real partition in the atmosphere. Local phenomena contribute to the global atmosphere, and the global atmosphere contributes to local phenomena. It is a two-way process.
This local, instantaneous data should give us better rainfall prediction in specific regions and also help with global climate modelling, particularly in the Indian context.
What’s the roadmap ahead?
So far, we have only validated the weather data for Bhubaneswar, so we have confidence that the data produced in our facility is correct, or good enough for better rainfall estimation.
The next step is to conduct this under different conditions. We do not specify this by state, but by ground-level temperature and humidity, to develop region-specific rainfall models.
A lot of work remains because the country is so diverse. It is not just 10 relationships that we need to establish, but potentially more than a thousand. We therefore have to run many experiments to develop region-specific rainfall models for different parts of the country.
On the personal front, we are also working on something else. At airports and defence-related sites, a device called a disdrometer is used to measure rainfall rate. It is very costly — one crore rupees or more, sometimes over ₹70 lakh — and has traditionally been imported from Germany or the US.
We wanted to develop an indigenous version. Its working principle is different, but it should provide better, more detailed data at low cost and be made indigenously. We currently have a patent on it and want to establish it as well.
So, essentially, our future goal is to establish this rainfall research in a much more rigorous way.
