Noemi Vergopolan: satellite data and machine learning for predicting droughts

Written by
Sharon Adarlo
March 1, 2021

Noemi Vergopolan, 29, doctoral student

Studies:

Noemi Vergopolan is a doctoral student in the civil and environmental engineering department and expects to receive her PhD this year. In addition, she is completing the Graduate Certificate Program in Statistics and Machine Learning at the Center for Statistics and Machine Learning (CSML) and another certificate in computational science and engineering from the Princeton Institute for Computational Science and Engineering. Before coming to Princeton, Vergopolan earned her bachelor's degree in environmental engineering from Universidade Federal do Paraná in Brazil.

During her undergraduate studies, she was an exchange student at North Carolina State University in Raleigh. She developed her undergraduate senior thesis at NASA's Jet Propulsion Lab, where she used satellite data to study the impact of deforestation on the hydrological cycle in the Amazon rainforest.

 

Research:

Vergopolan's research aims to understand the impact of extreme hydroclimate, such as droughts and floods, on human activities. One focus of her research is predicting soil moisture, droughts, and their impact on agricultural production, such as corn yields.

"I use machine learning with satellite data and hydrological modeling to predict terrestrial water distribution. My research goal is to obtain accurate and detailed soil moisture estimates, a key variable for many water resource applications, including agriculture. From there, I use machine learning to predict maize production," Vergopolan said.

In specific terms, she takes high-resolution remote sensing observations from satellites, which are helpful because they provide global and frequent monitoring. She combines it with HydroBlocks, a numerical cluster-based model that simulates the land surface's hydrological processes. With this framework, she can obtain soil moisture content across the United States at very high-resolution. Further, Vergopolan combines soil moisture simulations and machine learning to improve the prediction of extreme events, such as floods and droughts, water scarcity, irrigation water demands, and crop yields at high spatial resolution.

Vergopolan said this kind of work is important because improving the accuracy and predictability of extreme hydroclimate and agriculture yields can help government and farmers make better decisions, especially if a drought is coming.

Her research focuses not only on the continental United States but also on Africa. Many developing countries do not have as much existing data for farmers and governments to draw from for decision-making. As such, Vergopolan's framework provides a solution to bridge this data gap by leveraging satellite-data and machine learning.

Vergopolan has secured a Postdoctoral Fellowship from the NOAA Geophysical Fluid Dynamics Laboratory. She will take up this position after obtaining her doctoral degree. "I am going to implement a part of my research into one of GFDL forecast models, and I am very excited about it," she said.


Extracurricular Activities:

Vergopolan was treasurer of the Princeton Latino Graduate Student Association, a Resident Graduate Student at Mathey College, and editor for Princeton's online magazine Highwire Earth, an interdisciplinary publication on sustainable development. She also hosts Energy Table, a Princeton event that holds discussions on energy and the environment.

 

For Fun:

Vergopolan likes to dance, hike, and work out.