A0587
Title: Mapping and ranking Australian precipitation system by data fusion and multivariate empirical dynamic quantiles
Authors: Guoqi Qian - The University of Melbourne (Australia) [presenting]
Abstract: Precipitation data from Australia are collected from both ground-based rain gauge networks and spaceborne satellites. This increases the volume, coverage, and resolution of the available data but introduces extra complications and heterogeneity to them. Two techniques are developed for large-scale data fusion and ranking that have been applied to the Australian precipitation system to deliver a high-quality fused dataset with location rankings for the following statistical analysis in the pipeline. The developed data fusion technique consists of multivariate linear regression and spatial autoregression and kriging. The developed data ranking technique involves ranking a large collection of multivariate time series by the multivariate empirical dynamic quantiles (MEDQ) method.
