Statistical Emulation with Dimension Reduction for Complex Forward Models in Remote Sensing 

Zoom

Abstract: The retrieval algorithms in remote sensing generally involve complex physical forward models that are nonlinear and computationally expensive to evaluate. Statistical emulation provides an alternative with cheap computation and can be used to quantify uncertainty, calibrate model parameters and improve computational efficiency of the retrieval algorithms. Motivated by this, we introduce a framework of building statistical emulators by combining dimension reduction of input and output spaces and Gaussian process modeling. The functional principal component analysis (FPCA) via a conditional expectation method is chosen to reduce the dimension of the output space of the forward model. In addition, instead of making restrictive assumptions regarding the dependence structure of the high-dimensional input space, we apply the gradient-based kernel dimension reduction method to reduce the dimension of input space when the gradients of the complex forward model are unavailable or computationally prohibitive. A Gaussian process with feasible computation is then constructed at the low-dimensional input and output spaces. Theoretical properties of the resulting statistical emulator are explored, and the performance of our approach is illustrated with numerical studies using data from NASA’s Orbiting Carbon Observatory-2 (OCO2) data. 

Bio:  Emily Kang is Associate Professor of Statistics in the Division of Statistics and Data Sciences and Department of Mathematical Sciences at University of Cincinnati. She received a PhD in Statistics from The Ohio State University. Before joining UC, she was a postdoctoral fellow at the Statistical and Applied Mathematical Sciences Institute. Her research has focused on statistical methodologies and computational statistics for large data with complex dependence structures, in particular spatial, spatio-temporal and network data; she has also been working on statistical emulation for computer experiments, uncertainty quantification, Bayesian hierarchical modeling with applications in environmental, climate, biological sciences and engineering.