Five different analyses of 1982-83 monthly average surface wind stress fields have been used to force an ocean general circulation model of the tropical Pacific, in a series of El Niño hindcast experiments, like the one reported by Philander and Seigel. Although there were prominent common departures from climatology in the surface wind stress field during 1982-83 according to each wind analysis, there are also very substantial differences between analyses. This study was done to investigate the sensitivity of such hindcasts to our uncertain knowledge of the surface wind stress field. We concentrate here on the behavior along the Pacific ship-of-opportunity tracks. According to the ship-of-opportunity XBT data, the ocean underwent major changes during this period. The vertical temperature gradients and mixed layer temperatures, as well as the depth of the thermocline, underwent substantial changes. There were also major changes in the geostrophic flow of the major current systems, as revealed by upper ocean dynamic height differences. Comparing the hindcasts with observations, we find that the gross large-scale changes of the ENSO event - surface warming in the second half of 1982, continued warmth into 1983 and cooling in mid-1983, together with major thermocline depth changes - are found in each hindcast. However, major quantitative differences exist between each hindcast and the observations in at least some region for some time and some variable. Within the waveguide, dynamic height changes generally are hindcast with quantitative skill using each wind stress field and the best hindcasts differ from the observations by only a few dyn-cm more than the estimated uncertainty in the observations. Such hindcast skill is unlikely to be fortuitous: evidently the major elements of the waveguide variability are forced by the 1982-83 surface wind stress field rather than evolving out of some aspect of the state of the ocean during late 1981. Sea surface temperature changes are generally hindcast with qualitative skill, but rms erros of 2-3°C are frequent. Subsurface temperature variability skill varies with hindcast, location and depth; skill is greatest in the thermocline. Outside the waveguide, hindcast skill tends to be reduced, and varies greatly with location and hindcast. Quantitative hindcast skill is found near 10°S and 10°N in some hindcasts in the WP, and near 10°S in most hindcasts in the CP, but there is never quantative skill in the NECC region. The most striking inconsistency found involves the behavior of the NMC hindcast in the region of the North Equatorial Counter Current. Wind stress curl-forced Ekman pumping appears to be a significant factor in the variations in the more successful hindcasts. In almost every comparison, the range of hindcast results brackets the observations, suggesting that the model physics is plausible. Overall, the special research effort wind fields produced better dynamic height results than did the operational wind product fields, but the operational fields produced generally better waveguide SST results. Improved knowledge of the surface wind stress field (and its curl) is a minimum requirement if we are to assess more critically model performance, and to identify needed model improvements.
Download the paper as it appeared in JPO (pdf, 1.8 Mb)
Dr. William S. Kessler
NOAA / PMEL / OCRD
7600 Sand Point Way NE
Seattle WA 98115 USA
|See also:||Kessler home page Kessler publications PMEL home page|