Zonal Wind Indices to Reconstruct CONUS Winter Precipitation

Abstract

Seasonal precipitation forecasts over the contiguous United States (CONUS) during the 2015-2016 El Niño exhibited significant bias over many regions, especially in the western United States where seasonal information is particularly valuable for reservoir operation. Diagnosing the origin of this bias requires understanding the empirical signal from tropical heating to midlatitude precipitation. In this paper, we find that atmospheric zonal wind indices computed over the region typically associated with the winter jet stream provide a skillful, spatially distributed, linear prediction of precipitation over CONUS, over all winters (January-March; JFM). Furthermore, we show that more (less) central (eastern) Pacific Ocean heating may have contributed to the unexpected 2016 JFM CONUS precipitation and that this was likely predictable based on antecedent (December) sea surface temperatures. The zonal wind indices act as intermediate variables in a causal chain, and our analyses provide support for the potential for empirical prediction and also a diagnostic for physics-based models to help improve forecasts.

A Subjective Bayesian Framework for Synthesizing Deep Uncertainties in Climate Risk Management

“…Using the lag 1 and 2 partial autocorrelations, sample maximum, sample minimum, sample median, and Mann-Kendall test value as Bayesian test statistics, we find that draws from the posterior predictive distribution match the observed test statistics credibly (Figure S9 in Supporting Information S1) although panels (a and b) suggest the possibility of temporal structure not captured by our stationary IID model. Future efforts could represent this structure by conditioning the parameters of the distribution on relevant climate indices (as in Farnham et al, 2018Farnham et al, , 2017Ossandón et al, 2021;Wong, 2018).…”

A subjective Bayesian framework for synthesizing deep uncertainties in climate risk management

“…Diagnostics suggest (though cannot guarantee) convergence (see table S1). We evaluate the model's fit using posterior predictive checks (see Wong, 2018;Farnham et al, 2018Farnham et al, , 2017Ossandón et al, 2021).…”

Machine Learning of Key Variables Impacting Extreme Precipitation in Various Regions of the Contiguous United States

“…For extreme precipitation in the continental United States (CONUS), numerous studies have examined the characteristics of extreme precipitation events and linked extreme precipitation to large‐scale variables, for example, El Niño–Southern Oscillation (ENSO), North Atlantic Oscillation, Pacific Decadal Oscillation(L. Agel et al., 2018; Barlow et al., 2019; Farnham et al., 2017; Huang et al., 2018). The Madden–Julian oscillation (MJO) may impact rainfall probabilities at sub‐seasonal and seasonal scales (Becker et al., 2009).…”