Multi-resolution spatial methods for large data sets.

Doug Nychka (National Center for Atmospheric Research)

Thursday 8th May, 2014 15:00-16:00 Maths 203


Spatial data is ubiquitous and a basic problem is to reconstruct
surfaces from irregular observations or measurements and to quantify
the uncertainty in the estimates. Standard statistical methods break
when applied to large data sets and so alternative approaches are
needed that balance changes to the statistical models for increases in
computational efficiency. A useful method expands the field in a set
of compact basis functions and places a Gaussian Markov random field
latent model on the basis coefficients. The impact is that
evaluating the model likelihood and computing spatial predictions is
feasible even for tens of thousands of spatial observations on a
single computational core (e.g. a laptop). Moreover, by varying the
support of the basis functions and the correlations among basis
coefficients it is possible to entertain multi-resolution and
nonstationary spatial models that exploit the rich structure often
found in large data sets.

Add to your calendar

Download event information as iCalendar file (only this event)