Extended wind stress analyses for ENSO

Abstract

Surface wind stresses are fundamental to understanding El Niño, yet most observational stress products are too short to permit multidecadal ENSO studies. Two exceptions are the Florida State University subjective analysis (FSU1) and the NCEP-NCAR reanalysis (NCEP1), which are widely used in climate research. Here, the focus is on the aspects of the stress most relevant to ENSO-namely, the climatological background, anomaly spectrum, response to SST changes, subannual "noise" forcing, and seasonal phase locking-and how these differ between FSU1 and NCEP1 over the tropical Pacific for 1961-99. The NCEP1 stress climatology is distinguished from FSU1 by weaker equatorial easterlies, stronger off-equatorial cyclonic curl, stronger southerlies along the Peruvian coast, and weaker convergence zones with weaker seasonality. Compared to FSU1, the NCEP1 zonal stress anomalies (τ1x) are weaker, less noisy, and show less persistent westerly peaks during El Niño events. NCEP1 also shows a more stationary spectrum that more closely resembles that of equatorial east Pacific SST anomalies. After the 1970s, the equatorial trade winds and stress variability shift east and strengthen in FSU1, while the opposite occurs in NCEP1. Both products show increased mean convergence in the equatorial far west Pacific in recent decades, with weaker mean easterlies near the date line, an increased stress response to SST anomalies, and stronger interannual and subannual τ1x in the central equatorial Pacific (Niño-4; 5°N-5°S, 160°E-150°W). The variance of Niño-4 τ1x is highly seasonal in both datasets, with an interannual peak in October-November and a subannual peak in November-February; yet apart from interannual Niño-4 τ1x after 1980, stress anomalies are not well correlated between the products. Newer and more reliable stress estimates generally fall between NCEP1 and FSU1, with most closer to FSU1. © 2004 American Meteorological Society.