Time-varying quantiles for climate variability assessment
Bruno Sanso, University of California Santa Cruz
We develop a flexible dynamic quantile linear model (exDQLM) whichenables versatile, structured, and informative estimation of theevolution of a given quantile of an environmental variable ofinterest. In addition, our approach considers a transfer functionextension to our exDQLM as a method to quantify nonlinearrelationships between a specific quantile of a climatologicalresponse and a lagged input. We apply our method to the estimationof a high quantile over time of the integrated water vapor transport(IVT) magnitude. IVT is the primary component of many detectionschemes for atmospheric rivers. Our transfer function exDQLM is usedto to describe both the immediate and lagged effects of ENSO on theestimation of the .85 quantiles of IVT of the California coastduring the last four decades. The results show a significant butcomplex association with past values of ELI, a recently developedoceanic index that provides a summary of the sea surface temperaturefor the equatorial Pacific alternative to ENSO.