TIES Webinar Series on Data Science for Environmental Sciences (DSES)! Next Week!
11 views
Skip to first unread message
yuzhou Chen
Apr 13, 2022, 3:52:54 PM4/13/22
to Climate Informatics News
Dear List,
The International Environmetrics Society (TIES) has launched a new TIES Webinar Series on Data Science for Environmental Sciences (DSES).
Our second webinar will be on April 22, at 11:00 am Central Standard Time (UTC-6) (attached flyer).
You can virtually access the webinar and register via our website: www.environmetrics.xyz
Speaker: Frederik Simons, Department of Geosciences, Princeton University
Title: Beyond Roughness: Efficient Parameter Estimation of Sampled Random Fields in Geology and Geophysics
Abstract: Describing and classifying the statistical structure of topography and bathymetry is of much interest across the geophysical sciences. Oceanographers are interested in the roughness of seafloor bathymetry as a parameter that can be linked to internal-wave generation and mixing of ocean currents. Tectonicists are searching for ways to link the shape and fracturing of the ocean floor to build detailed models of the evolution of the ocean basins in a plate-tectonic context. Geomorphologists are building time-dependent models of the surface that benefit from sparsely parameterized representations whose evolution can be described by differential equations. Geophysicists seek access to parameterized forms for the spectral shape of topographic or bathymetric loading at various (sub)surface interfaces in order to use the joint structure of topography and gravity for inversions for the effective elastic thickness of the lithosphere. Planetary scientists are in need of robust terrain-classification models to help unravel the cratering history and tectonic evolution of planetary surfaces, for the selection of suitable landing sites, and for purposes as mundane as the prediction of wear and tear on rover wheels. Finally, statisticians, mathematicians and computer scientists are interested in the analysis of texture for purposes of out-of-sample prediction, extension, and in-painting for application in fields as diverse as computer graphics and medical imaging.
A unified geostatistical framework for the description, characterization and study of surfaces of these various kinds and for such a multitude of applications is via the Matérn process, a theoretically well justified and mathematically attractive parameterized form for the spectral-domain covariance of Gaussian processes, both in isotropic form and considering various geometrical kinds anisotropy. We provide a computationally and statistically efficient method for estimating the parameters of a stochastic covariance model observed on a regular spatial grid in any number of dimensions. Our proposed method, which we call the Debiased Spatial Whittle likelihood, makes important corrections to the well-known Whittle likelihood to account for large sources of bias caused by boundary effects and aliasing. We generalise the approach to flexibly allow for significant volumes of missing data including those with lower-dimensional substructure, and for irregular sampling boundaries. We build a theoretical framework under relatively weak assumptions which ensures consistency and asymptotic normality in numerous practical settings including missing data and non-Gaussian processes. We also extend our consistency results to multivariate processes. We provide detailed implementation guidelines which maintains the computational scalability of Fourier and Whittle-based methods for large data sets. We validate our procedure over a range of simulated and real world settings, and compare with state-of-the-art alternatives, demonstrating the enduring practical appeal of Fourier-based methods, provided they are corrected by the procedures developed in our work.
Hope to see you all there!
Yuzhou Chen,
On behalf of the TIES Seminar Series' organizing committee
Yuzhou Chen
Princeton University (Postdoctoral Research Associate)
Lawrence Berkeley National Laboratory (NSF Research Fellow)
Our second webinar will be on April 22, at 11:00 am Central Standard Time (UTC-6) (attached flyer).
You can virtually access the webinar and register via our website: www.environmetrics.xyz
Speaker: Frederik Simons, Department of Geosciences, Princeton University
Title: Beyond Roughness: Efficient Parameter Estimation of Sampled Random Fields in Geology and Geophysics
Abstract: Describing and classifying the statistical structure of topography and bathymetry is of much interest across the geophysical sciences. Oceanographers are interested in the roughness of seafloor bathymetry as a parameter that can be linked to internal-wave generation and mixing of ocean currents. Tectonicists are searching for ways to link the shape and fracturing of the ocean floor to build detailed models of the evolution of the ocean basins in a plate-tectonic context. Geomorphologists are building time-dependent models of the surface that benefit from sparsely parameterized representations whose evolution can be described by differential equations. Geophysicists seek access to parameterized forms for the spectral shape of topographic or bathymetric loading at various (sub)surface interfaces in order to use the joint structure of topography and gravity for inversions for the effective elastic thickness of the lithosphere. Planetary scientists are in need of robust terrain-classification models to help unravel the cratering history and tectonic evolution of planetary surfaces, for the selection of suitable landing sites, and for purposes as mundane as the prediction of wear and tear on rover wheels. Finally, statisticians, mathematicians and computer scientists are interested in the analysis of texture for purposes of out-of-sample prediction, extension, and in-painting for application in fields as diverse as computer graphics and medical imaging.
A unified geostatistical framework for the description, characterization and study of surfaces of these various kinds and for such a multitude of applications is via the Matérn process, a theoretically well justified and mathematically attractive parameterized form for the spectral-domain covariance of Gaussian processes, both in isotropic form and considering various geometrical kinds anisotropy. We provide a computationally and statistically efficient method for estimating the parameters of a stochastic covariance model observed on a regular spatial grid in any number of dimensions. Our proposed method, which we call the Debiased Spatial Whittle likelihood, makes important corrections to the well-known Whittle likelihood to account for large sources of bias caused by boundary effects and aliasing. We generalise the approach to flexibly allow for significant volumes of missing data including those with lower-dimensional substructure, and for irregular sampling boundaries. We build a theoretical framework under relatively weak assumptions which ensures consistency and asymptotic normality in numerous practical settings including missing data and non-Gaussian processes. We also extend our consistency results to multivariate processes. We provide detailed implementation guidelines which maintains the computational scalability of Fourier and Whittle-based methods for large data sets. We validate our procedure over a range of simulated and real world settings, and compare with state-of-the-art alternatives, demonstrating the enduring practical appeal of Fourier-based methods, provided they are corrected by the procedures developed in our work.
Hope to see you all there!
Yuzhou Chen,
On behalf of the TIES Seminar Series' organizing committee
Yuzhou Chen
Princeton University (Postdoctoral Research Associate)
Lawrence Berkeley National Laboratory (NSF Research Fellow)
Reply all
Reply to author
Forward
0 new messages