IMAGe Brown Bag- Global Spatial Statistics

05/27/2016 - 12:00pm to 1:00pm

Jaehong Jeong
King Abdullah University of Science and Technology (KAUST)

Global-scale geophysical, environmental, and climate science data sets require statistical models that explain the curvature of their spatial domain. Over the last few decades, statisticians have developed covariance models to capture their spatial and/or temporal behavior. Mathematical limitations have prevented the use of the geodesic distance, the most natural metric for measuring distance on the surface of a sphere, and instead some previous approaches have applied the Euclidean or chordal distance to approximate the covariance. However, because these approximations may result in physically unrealistic distortions on the sphere, covariance functions directly defined on the sphere using the geodesic distance are needed. We discuss the issues that arise when dealing with global-scale spherical data sets and review current geostatistical approaches. We illustrate the use of isotropic and nonstationary covariance models through deformations and geographical indicators for global surface temperature data. To assess the suitability of each method, we compare their log likelihood values and prediction score. This is a joint work with Mikyoung Jun and Marc G. Genton.

Friday, May 27, 2016
12:00-1:00 pm
Mesa Lab, Damon Room
(Bring your lunch)