Subscribe
In a major scientific breakthrough with direct relevance to climate resilience, the Cochin University of Science and Technology (CUSAT) has secured a national patent for a new machine learning-based technology that can accurately extract rainfall information from atmospheric radar signals. The innovation, developed at the university’s Advanced Centre for Atmospheric Radar Research (ACARR), is expected to transform how weather is observed, analysed and predicted in Kerala — a state repeatedly tested by floods and extreme monsoon events.
The new system merges advanced signal processing with artificial intelligence to solve one of the oldest challenges in radar meteorology — separating echoes from falling raindrops and those generated by atmospheric turbulence. In tropical regions like Kerala, radar signals are often contaminated by this overlap, leading to inaccurate rainfall and wind measurements.
The research team at CUSAT devised what they describe as a Hybrid Adaptive Bi-Gaussian Fitting Algorithm in combination with a Gaussian Mixture Model (GMM). Together, they automatically identify and separate rain and clear-air echoes in radar signals within the lower troposphere, up to about five kilometres. The method works even under rapidly changing weather conditions, when rainfall intensity fluctuates sharply.
“This innovation allows the radar to clean its own data in real time,” said S Abhilash, director of ACARR and one of the inventors. “During a heavy downpour, the radar receives mixed echoes from air motion and raindrops. Our algorithm uses machine learning to recognise those patterns and separate them instantly. The result is a clearer, more accurate picture of rainfall and atmospheric dynamics. It is a small but vital step in improving flood forecasting and early warning systems in Kerala.”
The patented technique, officially titled “System and Method for Distinguishing Atmospheric Echoes within a Radar Spectrum”, builds on more than a decade of radar-based atmospheric research at CUSAT. The university operates India’s first 205 MHz Stratosphere–Troposphere (ST) Radar, a national facility dedicated to studying the tropical atmosphere and monsoon dynamics.
This powerful radar, located on the CUSAT campus at Kalamassery, can profile winds, turbulence and echoes from a few hundred metres above the ground to nearly 20 kilometres into the sky. Unlike conventional weather radars that lose accuracy during heavy rain, the 205 MHz system maintains signal integrity and provides continuous data even in intense monsoon conditions.
Over the years, the radar has generated one of the richest datasets on tropical atmospheric processes in India. It captures everything from low-level monsoon jets and convective bursts to upper-air wind shears and moisture transport. Such information is essential for understanding extreme rainfall, cyclogenesis, and the onset and withdrawal phases of the monsoon.
Abhilash said the patented system will further strengthen this radar’s capabilities and make it more relevant to Kerala’s specific climate challenges. “Kerala’s weather is highly variable and often violent. Within a few hours, we can move from light showers to cloudbursts. Conventional algorithms developed for temperate climates do not work well here. Our method is customised for the tropical atmosphere and trained on real data from our radar. It can distinguish rainfall echoes even in mixed or transitional conditions, giving us much cleaner signals to work with,” he explained.
The development comes at a time when the state is looking to improve its weather intelligence and early-warning systems after a series of devastating floods. The catastrophic floods of 2018 and 2019 exposed the limitations of existing rainfall monitoring networks. In several instances, inaccurate rainfall inputs contributed to delayed flood alerts. The CUSAT innovation could fill that gap by providing real-time, high-resolution rainfall and wind data that can be integrated directly into forecasting models.
“This is not just about academic curiosity,” said Abhilash. “We want this technology to reach the operational level. Once integrated with the Kerala State Disaster Management Authority (KSDMA) and the India Meteorological Department (IMD), it can help provide more accurate warnings for floods, landslides and urban waterlogging.”
CUSAT’s achievement also reflects the growing strength of indigenous science in Kerala. The ACARR facility is one of the few centres in the country where engineers, atmospheric scientists and data analysts work under one roof. Apart from the 205 MHz radar, it houses a suite of instruments including micro-rain radars, microwave radiometers, disdrometers, ceilometers, and automatic weather stations. The centre has built collaborations with the Indian Institute of Tropical Meteorology (IITM), ISRO, and other research bodies.
In recent years, CUSAT’s radar observations have contributed to improvements in monsoon prediction models and better understanding of tropical convection. Its scientists have published extensively in national and international journals, exploring everything from the behaviour of boundary-layer winds to the vertical coupling of the troposphere and stratosphere.
Beyond scientific advancement, the new patent points toward practical solutions for climate adaptation. More accurate rainfall information can support not only disaster management but also agriculture, water resource planning, and fisheries. In Kerala, where coastal and inland livelihoods are deeply tied to the rhythms of the monsoon, timely rainfall data can make a crucial difference.
The technology could also find use across India’s coastal and tropical zones, from Odisha to Tamil Nadu and the Andaman Islands, where similar radar challenges exist. Researchers believe the patented approach can be adapted to other radar frequencies and eventually become part of a national radar-based rainfall monitoring network.
“India is entering a phase where local innovations must shape our response to climate risks,” said Abhilash. “We can no longer rely entirely on imported models or technologies built for different weather systems. The monsoon is unique and understanding it demands our own tools and perspectives. This patent is part of that effort.”
CUSAT officials said the next phase will involve pilot projects with state and national agencies to operationalise the algorithm in real-time forecasting systems. If implemented successfully, Kerala could soon become the first state in India to use machine learning-enabled radar data for flood and rainfall alerts.
“The atmosphere above Kerala is our laboratory,” said Dr Abhilash. “Every monsoon teaches us something new. With this technology, we are getting better at listening to those lessons.”
The patented research emerged from a collaborative effort involving Dhanya R, Assistant Professor at ToCH Institute of Science and Technology and research scholar at the Department of Electronics, CUSAT; Anju Pradeep, Professor at the School of Engineering, CUSAT; Abhilash S, Director of the Radar Centre, CUSAT; and Abhiram Nirmal from the Radar Centre team.