Earthquake Early Warning in Kathmandu

In 2015, the Gorkha Earthquake devastated Nepal, killing nearly 9,000 people, injuring more than 22,000 and displacing 3.5 million. The $10 billion in economic losses equaled roughly half of the country’s annual GDP. Studies suggest that another rupture along the same fault line could exceed the scale of the 2015 disaster. Yet Nepal still does not have a comprehensive earthquake early warning system.

Through a Bass Connections project, Duke University and Tribhuvan University’s Institute of Engineering (IOE) in Kathmandu are working to build one.

The 2025–2026 team is led by Henri Gavin, professor and the W.H. Gardner Jr. Chair of Civil and Environmental Engineering at Duke; Prashamsa Koirala, graduating senior in electrical and computer engineering; Amna Rauf, master’s student in global health; and Kjitij C. Shrestha, associate professor of civil engineering at Tribhuvan University IOE. Graduate team member Trailokaya Bajgain, a second-year PhD student in civil and environmental engineering, is also helping lead the project.

Bajgain, who studies air pollution, first joined the initiative as a Duke Kunshan University student. A Nepali citizen who moved to the United States in 2021, he said the work felt personal from the start. “It was something close to my heart.” The project is built around what he calls “student-to-student co-development.” Each year, a cohort of students at Duke and a parallel cohort in Kathmandu are selected to work on related research questions. They meet regularly over Zoom and present their findings each summer at a research symposium in Nepal.

“The partnership we have built is very valuable,” Bajgain said. “We are learning a lot from the Nepalese studentsand the Nepalese students are learning a lot from Duke students.” For many Nepali participants, he added, “it’s one of the few projects where they get international exposure.” The collaboration between Duke and IOE has now continued for more than six years.

The work spans engineering, social science, data science, and public policy. Some students study how political trust shapes disaster preparedness. Others design earthquake- resistant furniture or develop mobile messaging systems for early warnings. On the engineering side, students are building a smart seismic sensing network using custom-printed circuit boards and motion sensors.

Earthquake early warning depends on detecting the first seconds of ground motion. When an earthquake begins, it produces a fast- moving primary wave, or P-wave, followed by a slower but more destructive secondary wave, or S-wave. Bajgain explained that researchers analyze the initial P-wave signal to estimate the potential intensity of the S-wave. That analysis can provide a window of a few seconds to up to a minute, allowing emergency protocols to begin before the most destructive waves arrive.

The technology relies on an array of motion sensors, similar in principle to a smartphone’s gyroscope, distributed across the region. If one sensor records movement, it could be an isolated disturbance. But if the network detects consistent signals across multiple sensors, the system can identify an earthquake in progress and estimate its likely impact.

Although early warning systems are well established in places like Japan and California, Nepal presents distinct challenges. Its position along the Himalayan tectonic belt affects how seismic waves travel. “You can’t copy and paste the system that works in California to Nepal,” Bajgain explained. So far, the team has trained its models using data from Japan, which has one of the world’s most extensive seismic monitoring networks. “We aim to have our first set of working sensors by this summer,” Bajgain said. Then the team will start collecting data specific to Nepal’s geography.

Siri Gullapalli, an electrical and computer engineering and computer science major, is part of the team developing the sensing device. “I really liked the fact that the product that we were making was being tested in the environment we were making it for,” she said.

The device collects analog motion signals and converts them into digital data that feed into a machine learning model trained to distinguish between P-waves and S-waves. “Timing is super crucial when working with early earthquake warning,” Gullapalli said. Even small processing delays can reduce accuracy and shorten the response window. To minimize latency, the team is redesigning parts of the circuitry to process signals more efficiently.

Gullapalli said working across interdisciplinary and intercultural subteams has been one of the most rewarding aspects of the project. “I’ve learned a lot about PCB design from the team in Nepal,” she said. She also collaborates with students in the social sciences who study how to communicate warnings effectively so people will trust and act on them, and with structural engineering teams focused on designing earthquake-resistant structures. “Being able to interface with these teams is something I’ve really enjoyed doing.”

This year, the team expanded its focus to air pollution, another major hazard in Nepal. “Air pollution is now the number one health risk factor for Nepal, and the average life expectancy in Nepal is reduced by 3.3 years as a result,” Bajgain said. According to the Air Quality Life Index, Nepal ranks as the third-most polluted country in the world, yet its monitoring infrastructure remains limited.

Drawing on data from a previous Duke project that deployed 70 air quality sensors in Nepal, the team is conducting a feasibility study for a new monitoring network. The goal is to produce a white paper outlining how and where additional sensors should be deployed, while building local capacity to maintain the system long term.

For Bajgain, a central question guides the work: “Once the project is over, how do you sustain these systems?” he asked. “One of the things we believe in is local capacity building and local experts.” Beyond developing early earthquake warning technologies and air quality monitoring systems, the project aims to ensure Nepal has the expertise and infrastructure to continue the work long after the research ends.

Claire Andreason is a graduating senior double majoring in electrical and computer engineering and physics.