Answers are grounded in this paper and cite their sources.
A new multilevel, hierarchical Bayesian model is developed to simultaneously identify clusters of stations with similar temporal patterns, the trend associated with each cluster and for the individual stations within each cluster. An application to the annual maximum daily precipitation in the United States for a common 70 year (1941–2010) period across the HADEX2 sites is presented. The model identifies statistically homogeneous regions, spatially clustering the data into groups according to the intensity and the trend. Partial pooling of model parameters for each group is considered. Spatially consistent trends are detected in the Midwest and Northeast U.S., at the cluster and at the station level. The new approach can dramatically improve trend identification for precipitation extremes; e.g., all 14 stations in the Midwest report a significant trend as opposed to only 4 stations based on single site analysis. The method is generally applicable for improving trend identification over a heterogeneous region.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2015) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: