Dr. Doug Nychka (NCAR)

Date: 

Tuesday, February 4, 2014, 3:00pm to 4:00pm

Location: 

HUCE Seminar Room

ClimaTea is held every Tuesday from 3-4pm in the HUCE Seminar room, which is on the 3rd floor of the Geo Museum in the Harvard University Center for the Environment.

Title: Statistical Models for Spatial Struct in Climate Simulations and Data

Background reading: Nychka, et al (NCAR Technical Note, 2013), 'A multi-resolution Gaussian process model for the analysis of large spatial data sets'

Abstract:
The interest in the regional effects of climate change has motivated the analysis of large spatial and space-time data that are the result of numerical models. Typically the model output involves grids of several thousand points and standard methods of spatial statistics break when applied to these large data sets.  This talk will present a flexible spatial model based on fixed rank Kriging that can handle a large number of spatial locations and also include non stationary spatial dependence. This is feasible using compactly supported basis functions and spatial dependence based on Markov random fields. Using this method we estimate the change in the seasonal cycle of temperature over the US from climate simulations from the North American Regional Climate Change and Assessment Program (NARCCAP). Part of this analysis is to account for topography and other covariates and to determine the effect of specific pairings of global and regional models on the results.

 

Refreshments will be provided during the presentation.