An official website of the United States government
Summary For multivariate spatial Gaussian process models, customary specifications of cross-covariance functions do not exploit relational inter-variable graphs to ensure process-level conditional independence between the variables. This is undesirable, especially in highly multivariate settings, where popular cross-covariance functions, such as multivariate Matérn functions, suffer from a curse of dimensionality as the numbers of parameters and floating-point operations scale up in quadratic and cubic order, respectively, with the number of variables. We propose a class of multivariate graphical Gaussian processes using a general construction called stitching that crafts cross-covariance functions from graphs and ensures process-level conditional independence between variables. For the Matérn family of functions, stitching yields a multivariate Gaussian process whose univariate components are Matérn Gaussian processes, and which conforms to process-level conditional independence as specified by the graphical model. For highly multivariate settings and decomposable graphical models, stitching offers massive computational gains and parameter dimension reduction. We demonstrate the utility of the graphical Matérn Gaussian process to jointly model highly multivariate spatial data using simulation examples and an application to air-pollution modelling.
more »
« less
Vecchia Approximations and Optimization for Multivariate Matérn Models
We describe our implementation of the multivariate Matérn model for multivariate spatial datasets, using Vecchia’s approximation and a Fisher scoring optimization algorithm. We consider various pararameterizations for the multivariate Matérn that have been proposed in the literature for ensuring model validity, as well as an unconstrained model. A strength of our study is that the code is tested on many real-world multivariate spatial datasets. We use it to study the effect of ordering and conditioning in Vecchia’s approximation and the restrictions imposed by the various parameterizations. We also consider a model in which co-located nuggets are correlated across components and find that forcing this cross-component nugget correlation to be zero can have a serious impact on the other model parameters, so we suggest allowing cross-component correlation in co-located nugget terms.
more »
« less
- Award ID(s):
- 1953088
- PAR ID:
- 10384563
- Date Published:
- Journal Name:
- Journal of Data Science
- ISSN:
- 1680-743X
- Page Range / eLocation ID:
- 475 to 492
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract. Localization is widely used in data assimilation schemes to mitigate the impact of sampling errors on ensemble-derived background error covariance matrices. Strongly coupled data assimilation allows observations in one component of a coupled model to directly impact another component through the inclusion of cross-domain terms in the background error covariance matrix.When different components have disparate dominant spatial scales, localization between model domains must properly account for the multiple length scales at play. In this work, we develop two new multivariate localization functions, one of which is a multivariate extension of the fifth-order piecewise rational Gaspari–Cohn localization function; the within-component localization functions are standard Gaspari–Cohn with different localization radii, while the cross-localization function is newly constructed. The functions produce positive semidefinite localization matrices which are suitable for use in both Kalman filters and variational data assimilation schemes. We compare the performance of our two new multivariate localization functions to two other multivariate localization functions and to the univariate and weakly coupled analogs of all four functions in a simple experiment with the bivariate Lorenz 96 system. In our experiments, the multivariate Gaspari–Cohn function leads to better performance than any of the other multivariate localization functions.more » « less
-
Hinrichs, Ute; Perin, Charles (Ed.)As scientific data continues to grow in size, complexity, and density, the representation scope of three-dimensional spaces, data sampling methods, and transfer functions have improved in parallel, allowing visualization practitioners to produce richer multidimensional encodings. Glyphs, in particular, have become an essential encoding tool due to their versatile applications in co-located multivariate volumetric datasets. While prior work has been conducted investigating the perceptual attributes of computationally-generated three-dimensional glyph-forms for scientific visualization, their affective and expressive qualities have yet to be examined. Further, our prior work has demonstrated the benefits of artist hand-created glyph forms in contrast to commonly-used synthetic forms in increasing visual diversity, discrimination, and expressive association in complex environmental datasets. In order to begin to address this gap, we establish preliminary groundwork for an affective design space for hand-created glyph forms, produce a novel set of glyph forms based on this design space, describe a non-verbal method for discovering affective classifications of glyph-forms adopted from current affect theory, and report the results of two studies that explore how these three-dimensional forms produce consistent affective responses across assorted study cohorts.more » « less
-
null (Ed.)Abstract We conduct a study of the aliased spectral densities of Matérn covariance functions on a regular grid of points, providing clarity on the properties of a popular approximation based on stochastic partial differential equations. While others have shown that it can approximate the covariance function well, we find that it assigns too much power at high frequencies and does not provide increasingly accurate approximations to the inverse as the grid spacing goes to zero, except in the one-dimensional exponential covariance case.more » « less
-
Arecaceae (palms) are an important resource for indigenous communities as well as fauna populations across Amazonia. Understanding the spatial patterns and the environmental factors that determine the habitats of palms is of considerable interest to rainforest ecologists. Here, we utilize remotely sensed imagery in conjunction with topography and soil attribute data and employ a generalized cluster identification algorithm, Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN), to study the underlying patterns of palms in two areas of Guyana, South America. The results of the HDBSCAN assessment were cross-validated with several point pattern analysis methods commonly used by ecologists (the quadrat test for complete spatial randomness, Morista Index, Ripley’s L-function, and the pair correlation function). A spatial logistic regression model was generated to understand the multivariate environmental influences driving the placement of cluster and outlier palms. Our results showed that palms are strongly clustered in the areas of interest and that the HDBSCAN’s clustering output correlates well with traditional analytical methods. The environmental factors influencing palm clusters or outliers, as determined by logistic regression, exhibit qualitative similarities to those identified in conventional ground-based palm surveys. These findings are promising for prospective research aiming to integrate remote flora identification techniques with traditional data collection studies.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript
- Journal Article:
- https://doi.org/10.6339/22-JDS1074
