IMAGe Seminar - Statistical Methods for Quantifying the Uncertainties in Climate Data Reconstruction and Climate Signal Detection

10/12/2016 - 1:30pm to 2:30pm
ML Damon Room

Sam Shen, San Diego State University

This talk will describe some modern statistical methods to optimally interpolate historical spatial climate data, to estimate errors in the assessment of past climate changes, and to quantify uncertainties in the climate change detection. Principal component analysis’ U eigenvectors, called empirical orthogonal functions in climate science, are extensively used to account for spatial inhomogeneity. Two products will be presented. One is the error estimates of the global averages annual mean of surface air temperature since 1861 for the reports of Intergovernmental Panel on Climate Change (IPCC). Another is the reconstructed global monthly precipitation dataset since January 1900 with 2.5 deg latitude-longitude resolution. Some fundamental issues on statistical and mathematical theories for the climate model uncertainties will be discussed, including non-stationarity, nonlinearity, and interactions of multiple spatiotemporal scales.

Wednesday, October 12, 2016
1:30 - 2:30 pm
Mesa Lab, Damon Room