Week 3\u20134 predictability over the United States assessed from two operational ensemble prediction systems

Lei Wang,Andrew W Robertson

doi:10.1007/s00382-018-4484-9

Abstract

The subseasonal predictability of surface temperature and precipitation is examined using two global ensemble prediction systems (ECMWF VarEPS and NCEP CFSv2), with an emphasis on the week 3–4 lead (i.e. 15–28 days ahead) fortnight-average anomaly correlation skill over the United States, in each calendar season. Although the ECMWF system exhibits slightly higher skill for both temperature and precipitation in general, these two systems show similar geographical variations in the week 3–4 skill in all seasons and encouraging skill in certain regions. The regions of skill are then interpreted in terms of large-scale teleconnection patterns. Over the southwest US in summer, the North American monsoon system leads to higher skill in precipitation and surface temperature, while high skill over northern California in spring is found to be associated with the seasonal variability of the Arctic Oscillation (AO). During winter, in particular, week 3–4 predictability is found to be higher during extreme phases of the El Niño–Southern Oscillation, Pacific-North American (PNA)/Tropical-Northern Hemisphere mode, and AO/North Atlantic Oscillation (NAO). Both forecast systems are found to predict these teleconnection indices quite skillfully, with the anomaly correlation of the wintertime NAO and PNA exceeding 0.5 for both models. In both models, the subseasonal contribution to the PNA skill is found to be larger than for the NAO, where the seasonal component is large.

Highlights

The prospect of substantial subseasonal to seasonal (S2S) climate predictability (2 weeks to a season ahead) has drawn increasing attention in recent years (Brunet et al 2010), especially for the week 3–4 forecast range (i.e. 15–28 days ahead) where many operational weather forecast centers have started to issue experimental forecasts (Vitart et al 2017)
T2m forecast skill suggests the influence of North American Monsoon on temperatures and shows a local maximum in southern Arizona and New Mexico in summer (Fig. 2e)
Composites of monthly-mean week 3–4 correlation of anomalous patterns (CORP) from the semi-weekly reforecasts for the negative (Nino3.4 index < − 1), neutral, and positive (> 1) phase of El Niño–Southern Oscillation (ENSO) are plotted in Fig. 10 for each variable

Summary

Introduction

The prospect of substantial subseasonal to seasonal (S2S) climate predictability (2 weeks to a season ahead) has drawn increasing attention in recent years (Brunet et al 2010), especially for the week 3–4 forecast range (i.e. 15–28 days ahead) where many operational weather forecast centers have started to issue experimental forecasts (Vitart et al 2017). Great advances have been achieved in the past few decades in medium-range (i.e., 4–10 days) weather forecasts (Bauer et al 2015) and seasonal (3–6 months) climate predictions (Kirtman et al 2014), whereas the subseasonal forecast

Present Address

Data and methods

Temporal correlation

Sources of week 3–4 predictability

Pattern correlation