Inference and Uncertainty Quantification for High Dimensional Systems in Remote Sensing

NSF DMS-2053668 (2021-2025)

Overview

This project develops statistical methods and computational tools for inverse problems in high-dimensional systems, with applications to remote sensing and carbon monitoring. The research focuses on building scalable statistical emulators using joint dimension reduction for input and output spaces, enabling efficient uncertainty quantification for remote sensing data products.

Grant Information:

  • Funding Agency: National Science Foundation
  • Program: Division of Mathematical Sciences
  • Award Number: DMS-2053668
  • Period: 2021 - 2025
  • PI: Emily L. Kang, University of Cincinnati
  • Type: Collaborative Research

Publications

2025

  1. Enhancing Gaussian Processes for Surrogate Modeling: A Review of Dimension Reduction Techniques for Input Variables
    Eric Herrison Gyamfi, Bledar A. Konomi, Guang Lin, and Emily L. Kang
    In Handbook of Statistical Methods for Computer Models: Uncertainty Quantification, 2025
    In press

2025

  1. A multivariate spatial statistical model for statistical downscaling of sea surface temperature in the Great Barrier Reef region
    Ayesha Ekanayaka, Emily L. Kang, Amy Braverman, and Peter Kalmus
    Journal of the Royal Statistical Society Series C: Applied Statistics, Nov 2025
    HTML DOI

2025

  1. A Practical Tool for Visualizing and Measuring Model Selection Uncertainty
    Sheng Ren, Peng Wang, and Emily L. Kang
    Stat, Mar 2025
    HTML DOI

2024

  1. Recursive nearest neighbor co-kriging models for big multi-fidelity spatial data sets
    Si Cheng, Bledar A. Konomi, Georgios Karagiannis, and Emily L. Kang
    Environmetrics, Feb 2024
    HTML DOI

2023

  1. Bayesian Latent Variable Co-kriging Model in Remote Sensing for Quality Flagged Observations
    Bledar A. Konomi, Emily L. Kang, Ayat Almomani, and Jonathan Hobbs
    Journal of Agricultural, Biological and Environmental Statistics, 2023
    HTML DOI

2022

  1. Statistical Downscaling of Sea Surface Temperature Projections with a Multivariate Gaussian Process Model
    Ayesha Ekanayaka, Peter Kalmus, Emily L. Kang, and Amy Braverman
    In Workshop on Gaussian Processes, Spatiotemporal Modeling, and Decision-Making Systems (GPSMDMS) at the International Conference on Neural Information Processing Systems (NeurIPS), 2022
    PDF HTML

2021

  1. Modeling Large Multivariate Spatial Data with a Multivariate Fused Gaussian Process
    Emily L. Kang, Miaoqi Li, Kerry Cawse-Nicholson, and Amy Braverman
    Journal of the Indian Statistical Association, 2021
    PDF

Software

MFGP

Multivariate Fused Gaussian Process model for modeling large multivariate spatial data.

Downscaling-demo

Statistical downscaling methods for climate and environmental data.

AutoBasisFDA

Automatic basis selection for functional data analysis.

Input_Dimension_reduction

Dimension reduction methods for high-dimensional input spaces.

Course Materials

STAT 8025 Spatial Statistics (Spring 2023)

This graduate course on spatial statistics was supported by this grant. Course materials cover statistical methods for analyzing spatial data, including kriging, co-kriging, and Gaussian processes.

Course Materials:

Lecture Notes:

STAT 7020 Surrogates: Gaussian Process Modeling, Design and Optimization (Fall 2025)

A new graduate course developed by Dr. Bledar A. Konomi (Co-PI on this project). This course focuses on Surrogate Models, which are fundamental to Uncertainty Quantification (UQ) in natural sciences, biological sciences, and engineering. Topics include Response Surface methods, Space Filling Design, Gaussian Processes, Model-based Design, Bayesian Optimization, Multifidelity and Calibration Models, and Sensitivity Analysis.

Course Materials:

Presentations

Selected Conference and Seminar Presentations:

Team

Principal Investigators

Co-Principal Investigator

  • Bledar A. Konomi, Associate Professor, Division of Statistics and Data Science, University of Cincinnati

Collaborators

  • Amy Braverman, Principal Statistician, Jet Propulsion Laboratory, California Institute of Technology
  • Jonathan Hobbs, Data Scientist, Jet Propulsion Laboratory, California Institute of Technology
  • Peter Kalmus, Data Scientist, Jet Propulsion Laboratory, NASA
  • Georgios Karagiannis, Associate Professor, Department of Mathematical Sciences, Durham University, UK
  • Kerry Cawse-Nicholson, Scientist, Jet Propulsion Laboratory, NASA

Students Supported

Graduate Students (Ph.D.)

Graduated:

  • Ayesha Kumari Ekanayaka Katugoda Gedara, Ph.D. 2024, University of Cincinnati (Currently Postdoctoral Fellow, UNC Chapel Hill)
  • Tzu-Chun Wu, Ph.D. 2022, University of Cincinnati (First position: Data Scientist, UC College of Medicine)
  • Jieyan Zhang, Ph.D. 2022, University of Cincinnati (Currently at BASF)
  • Gang Yang, Ph.D. 2022, University of Cincinnati (Currently at Bristol Myers Squibb)

Current:

  • Rick Lucas, Ph.D. student (2021-present), University of Cincinnati
  • Eric Herrison Gyamfi, Ph.D. student (2022-present), University of Cincinnati
  • Hancheng Li, Ph.D. student (2022-present, Joint with B. A. Konomi), University of Cincinnati
  • Ying Zhang, Ph.D. student (2024-present), University of Cincinnati
  • Lloyd Goldstein, Ph.D. student (2024-present), University of Cincinnati

Undergraduate Students

  • Linh Tran (Fall 2025), University of Cincinnati