Bivariate spatial analysis of temperature and precipitation from general circulation models and observation proxies

R. Philbin, M. Jun

Research output: Contribution to journalArticlepeer-review


This study validates the near-surface temperature and precipitation output from decadal runs of eight atmospheric ocean general circulation models (AOGCMs) against observational proxy data from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis temperatures and Global Precipitation Climatology Project (GPCP) precipitation data. We model the joint distribution of these two fields with a parsimonious bivariate Matérn spatial covariance model, accounting for the two fields' spatial cross-correlation as well as their own smoothnesses. We fit output from each AOGCM (30-year seasonal averages from 1981 to 2010) to a statistical model on each of 21 land regions. Both variance and smoothness values agree for both fields over all latitude bands except southern mid-latitudes. Our results imply that temperature fields have smaller smoothness coefficients than precipitation fields, while both have decreasing smoothness coefficients with increasing latitude. Models predict fields with smaller smoothness coefficients than observational proxy data for the tropics. The estimated spatial cross-correlations of these two fields, however, are quite different for most GCMs in mid-latitudes. Model correlation estimates agree well with those for observational proxy data for Australia, at high northern latitudes across North America, Europe and Asia, as well as across the Sahara, India, and Southeast Asia, but elsewhere, little consistent agreement exists.
Original languageEnglish (US)
Pages (from-to)29-44
Number of pages16
JournalAdvances in Statistical Climatology Meteorology and Oceanography
Issue number1
StatePublished - May 22 2015
Externally publishedYes


Dive into the research topics of 'Bivariate spatial analysis of temperature and precipitation from general circulation models and observation proxies'. Together they form a unique fingerprint.

Cite this