Manzhu Yu joined the Department of Geography in summer 2019 as an assistant professor of geography. Her research focuses on spatiotemporal theories and applications, atmospheric modeling, environmental analytics, and big data and cloud computing to solve pressing issues in natural hazards and sustainability.
Yu’s interest in natural hazards came from growing up in Northeast China.
“We experienced dust storms that turned the sky orange and dropped visibility drastically,” Yu said. “I’ve always been curious about why these dust storms happen, how they transport in the atmosphere, and where they deposit to the Earth surface.”
Yu said her curiosity about dust storms later expanded into a broader interest in all kinds of extreme weather events and natural hazards, and how to use big data to understand them.
“The emergence of big data started to influence my generation of scholars around 2010,” she said. “Big data technologies pose great opportunities to enhance the analysis and prediction of natural hazards, especially in the efficiency and timeliness aspects.”
Yu started to integrate big data technologies into her research to address computational and data handling challenges during natural disaster management.
“For the computational challenge, I used high-performance computing to conduct large-scale simulations of environmental hazards for improved computational efficiency,” Yu said. “For the data handling challenge, I adopted a deep learning model that classified different Twitter topics that emerged during hurricanes Sandy, Harvey, and Irma.”
By analyzing these topics, Yu is developing a model to identify especially vulnerable populations requiring relief.
The core methodology of Yu’s research is spatiotemporal data mining, modeling, analysis, and visualization.
“Extreme weather events are intrinsically spatiotemporal and highly dynamic, so to better understand the complex patterns of these events, spatiotemporal methodologies are essential,” Yu said.
Yu said she is most interested in working with data sources like in situ observations, meteorological observations, numerical simulations, volunteered geographic information, social media data and especially the Internet of Things (IoT).
“IoT sensors can monitor environmental status, such as temperature, humidity, and wind speed, to predict the occurrence of a disaster and the potential impacted areas,” she said. “However, the IoT data is usually incomplete and unstable, so one of my research activities in the near future will be addressing those issues to leverage the capabilities of IoT data to enhance disaster relief efforts.”
Other challenges Yu sees are intelligent processing and knowledge integration of massive datasets acquired from greater observation capabilities. She asks, “How can one extract the most valuable and related information, i.e. intelligent processing. And how can one integrate the extracted information and translate to knowledge from multiple sources covering different spatial and temporal durations, i.e. knowledge integration?”
Yu is seeking graduate students who share her curiosity about extreme weather events, their spatiotemporal patterns and contributing physical and social factors; and their associated mitigation, response, and early warning activities.
“Due to climate change, we are experiencing more frequent natural hazards with higher intensity,” Yu said. “Patterns demonstrate that the characteristics of natural hazards are changing. For example, hurricanes are moving more slowly after landfall, dumping more rain on the affected areas and creating more severe floods. Dust events are transporting in abnormal pathways, affecting areas that have never experienced dust storms. Extreme weather events are breaking records at a faster pace, making it harder to use historical statistics to explain and predict future events. Therefore, mitigating the negative impacts of these natural hazards is becoming more challenging.”