There is an increase in the frequency and intensity of natural disasters. Hurricanes, wildfires, floods, winter storms, and tornados have devastated many communities in the United States and across the globe. The Billion Dollar Disaster website of NOAA shows the tens of thousands of deaths and human injuries and billions of dollars in damages and costs. Improving disaster resilience and enhancing disaster management are critical for mitigating these losses due to a disaster.
Geospatial data capture human-environment interactions and their changes under a disaster context. Datasets, such as remote sensing images, weather and climate data, environmental data, and points of interest data, provide information about our natural and built environment in which a disaster is happening. Meanwhile, datasets, such as Census data, anonymized mobile phone location data, social media data, and traffic data, capture the characteristics of the affected communities and human behavior changes in respond to a disaster. By linking these geospatial datasets, we can better understand the complex human-environment-disaster system.
We integrate GeoAI methods and geospatial data to improve disaster resilience. We employ a wide range of machine learning and deep learning methods, from linear regression and random forest to deep neural networks and large foundation models, and integrate them with geographic knowledge and spatial principles. We ask research questions on various aspects of a disaster, from assessing disaster damages and predicting future affected areas to examining impact disparities and understanding related community factors. Together, we hope our research can inform disaster management and improve disaster resilience.
Join our lab
Interested in joining our lab? Please check our available positions on the Opportunities page.