Online Program Home
My Program

Abstract Details

Activity Number: 531 - SPEED: Statistical Computing: Methods, Implementation, and Application, Part 2
Type: Contributed
Date/Time: Wednesday, July 31, 2019 : 11:35 AM to 12:20 PM
Sponsor: Section on Statistical Computing
Abstract #307944
Title: Rapid Numerical Approximation of Spatial Covariance Functions Over Irregular Data Regions
Author(s): Peter Simonson* and Doug Nychka and Soutir Bandyopadhyay
Companies: Colorado School of Mines and Colorado School of Mines and Colorado School of Mines
Keywords: spatial statistics; Fourier transforms; integral approximation; computational efficiency

In geospatial data analysis, observations can often be represented as integrals of a target spatial field over irregular regions. In order to compute standard statistical models for this kind of indirect data it is necessary to find integrals over these irregular data regions with respect to spatial covariance functions. The irregularity of the spatial regions and the nature of the integrands can make analytic evaluation impractical. Naïve, Riemann-sum based numerical approximations of these integrals are often computationally expensive, when executed at the resolutions required for high quality approximation. We have implemented algorithms for numerically approximating these integrals using discrete Fourier transforms that provide accuracy comparable to Riemann approaches with computational costs that are dramatically lower.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program