Coupled Statistics-Physics Guided Learning to Harness Heterogeneous Earth Data at Large Scales

This project is funded by NASA's Advanced Information Systems and Technology (AIST) program. It aims to advance machine learning for Earth Science problems. Specifically, we will develop new technologies that address two major challenges facing machine learning for broad Earth Science applications—spatial heterogeneity, where satellite observations and their relationships to the prediction targets vary over space, and the limited and highly localized nature of ground-truth data that are needed to train the algorithms.

Zhao, Yingrui

Yingrui
Zhao
Yingrui Zhao

Yingrui Zhao is a Ph.D. candidate in Geography at the University of Maryland, College Park, specializing in Spatiotemporal Data Modeling, Travel Behavior Modeling, and GeoAI. Her research focuses on analyzing human mobility patterns by integrating big mobility data, machine learning, and GIScience methods to model travel behavior and analyze the spatial dynamics of transportation systems. She applied explainable machine learning and LLMs to analyze large-scale trip and social media datasets, offering new perspectives on how mobility interacts with demographics and built-environment. Her work has been published in leading journals such as the Journal of Transport Geography.

Before joining UMD, she earned her Master’s degree in Community and Regional Planning from the University of Texas at Austin, where she developed the Node-Place-People (NPP) model for evaluating transit-oriented development and conducted research on affordable housing supply through land use optimization for corridor TOD.

At UMD’s Master of Science in GIS Program, she serves as a teaching assistant and lab instructor, supporting a range of graduate courses including GEOG 656: Advanced Programming for GIS, GEOG 651: Spatial Statistics, GEOG 653: Spatial Analysis, GEOG 687: Applied GEOINT – Regional GeoStrategic Issues, GEOG 683: Hazards and Emergency Management, and GEOG 661: Fundamentals of Geospatial Intelligence.

Area of Interest
Spatiotemporal Data Modeling
Travel Behavior and Urban Form
Big Mobile Device Data
Geographical Artificial Intelligence (GeoAI)
Application of Large Language Models (LLMs) in Geoscience
Email
yzhao120@umd.edu
Room and Building
4600 River Road, Suite 334 (send mail to 2181 LeFrak)
Degrees Held

The University of Texas at Austin - MS

China Agriculture University - BE

Student Status
PhD Advanced to Candidacy
Research
Zhao, Yingrui, and Kathleen Stewart. 2025. “Analyzing Travel Behavior Differences across Population Groups: An Explainable Machine Learning Approach with Big Mobility Data.” Journal of Transport Geography 128 (October): 104368. https://doi.org/10.1016/j.j
Faculty Advisors

Rakowski, Judith

Judith
Rakowski
Judith J. Rakowski

Geographer with a focus on conservation, human-environmental system dynamics, protected areas management, restoration, ecosystem services, wildlife crime prevention.

BSc & MSc at Humboldt-University Berlin/Germany.

Working experience:
- WWF Azerbaijan & Armenia (2017)
- Helmholtz Zentrum for Environmental Research (UFZ) (2020)
- German Centre for Integrative Biodiversity Research (iDiv) (2021-2022)

- RA at Humboldt University at the lab of Urban Landscape Ecology (2018-2019)
- RA at Humboldt University at the lab of Conservation Biogeography (2019-2020)
- TA at Humboldt University at the lab of Applied Geoinformatics (2016-2017) 

 

Obtained grants:
- female career fund (Geography Department HU) 
- DAAD (German Academic Exchange Service) 
- ERASMUS+ grant
- PROMOS 
- GEOG Student Fellowship for International Collaborations
 

Human Dimensions of Global Change - Coupled Human and Natural Systems
Area of Interest
conservation, human-environmental system dynamics, protected areas management, restoration, ecosystem services, wildlife crime prevention, participatory methods, environmental justice
Position Title
PhD candidate, TA
Email
jrakowsk@umd.edu
Phone
2409 279072
Degrees Held

Geography and Agricultural Science - BSc

Global Change Geography - MSc

Student Status
PhD
Faculty Advisors

Irekponor, Victor

Victor
Irekponor
vireks

Victor is a Ph.D. candidate in the Center for Geographical Information Science, University of Maryland, College Park. He is broadly interested at the cutting-edge of research and development in urban informatics, smart cities, and geospatial data science; specifically, the science that investigates the ‘where’ and ‘why’ of various human and natural phenomena in cities.

Before coming to Maryland, he studied at the University of Lagos, Nigeria, and graduated with a first-class bachelor’s degree in urban planning. Beyond academics, Victor has over three years of software engineering experience as well as experience building and deploying machine learning models. His major tools include Python, Scikit-learn, Keras, Tensorflow, and React.js. He hopes to deploy his skills in software engineering, data science, and urban planning to conceptualize and contribute significantly to smart cities research.

Geospatial Information Science and Remote Sensing
Area of Interest
Spatial Data Science
Urban Informatics
GeoAI
GIS Programming (Python, PyTorch, TensorFlow)
Web GIS Mapping
Smart Cities
Multiscale local modeling
Position Title
Doctoral Candidate
Email
vireks@terpmail.umd.edu
Room and Building
4600, River road
Personal Website
Degrees Held

Urban Planning; major in urban informatics and computational urban planning. - BSc.

Student Status
PhD Advanced to Candidacy
Research
Handling Uncertainty in Multiscale Local Models
Faculty Advisors

Basak, Paromita

Paromita
Basak
Paromita Basak
Carbon, Vegetation Dynamics and Landscape-Scale Processes
Geospatial Information Science and Remote Sensing
Land Cover - Land Use Change
Area of Interest
Remote Sensing
Forest Biomass/Carbon Monitoring
Geographic Information System
Climate Change
Earth Dynamics
Environmental Conservation
Position Title
Doctoral Candidate, Graduate Research Assistant
Email
pbasak@umd.edu
Room and Building
4600 River Road
Degrees Held

Central European University (CEU) - M.Sc.

Asian University for Women (AUW) - B.Sc.

Student Status
PhD
Research
Basak, Paromita. 2020. Using Open Geospatial Data to Analyze Climate Change Impact on National Food Security Factors of Central Asia (Kyrgyzstan). Master of Science Thesis, Central European University, Budapest.
Basak, Paromita; Mohee, Fai; Gormaly, Marina Freire. 2020. Case study of Arsenic Contamination in Sitakunda District, Chittagong, Bangladesh. CSCE Annual Conference 'Tradition and the Future – La Tradition et L’Avenir '. Saskatoon, Saskatchewan, Canada.
Faculty Advisors

Li, Haijun

Haijun
Li
Haijun Li
Land Cover - Land Use Change
Remote Sensing
Area of Interest
Large-scale high-resolution crop type mapping with machine learning
Satellite monitoring on agriculture sustainability
High-performance computing for Analysis Ready Data (ARD) from satellite imagery
Position Title
Graduate Research Assistant
Email
haijunli@umd.edu
Room and Building
4600 River Road, Suite 358
Degrees Held

Cartography and Geography Information System, Wuhan University, 2017 - MS

Geographic Information System, Wuhan University, 2014 - BS

Student Status
PhD
Faculty Advisors