Sensitivity of wind-driven tropical Pacific Ocean simulations on seasonal and interannual time scales

The purposes of this study are (1) to characterize differences in the time/space structure of various multiyear surface wind products for the tropical Pacific; and (2) to quantify the impact these differences may have on our ability to model oceanic wind-forced variability on seasonal and interannual time scales. Three coincident wind field analyses are used, viz. the Florida State University (FSU) subjective analysis, the University of Hawaii (SAWIN) subjective analysis and the Fleet Numerical Oceanography Center (FNOC) operational analysis. The five years chosen for study, 1979-1983, encompass three years of a fairly regular seasonal cycle leading up to the 1982-1983 El Nino. A linear multi-vertical model model is forced with these analyses; model dynamic height and sea level are then compared with observations based on expendable bathythermograph and island tide gauge data. The mean seasonal cycle prior to El Nino (1979-1981) is considered first, which then serves as a self-consistent basis for analyzing the interannual variability, particularly the significant anomalies about the mean in 1982-1983. The impact of discrepancies in the forcing functions is discussed relative to the dominant seasonal and interannual scales of variability for the wind-driven oceanic response. The analyses of the wind products and model solutions indicate the need for special attention to the wind stress curl fields when evaluating wind products for use in tropical oceanographic applications. On seasonal time scales, critical differences in the wind stress products, of order 0.2-0.4 dyn cm(-2), are in wind regimes of surface convergence and significant gradients such as the ITCZ and SPCZ. These uncertainties in the wind fields, or more appropriately the wind stress curl distributions, are manifested in model sea level solutions as 6-12 cm discrepancies near the NECC Trough and east of New Guinea. On average, the seasonal amplitude of the wind-forced sea level response in any one simulation is of the same order as the differences between any two sea level simulations. The interannual variability is dominated by the anomalies associated with the 1982-1983 El Nino. Root mean square differences between the product versus product wind stress anomalies range from 0.1-0.2 dyn cm(-2) along the major ship tracks and up to 0.5 dyn cm(-2) away from the shipping lanes. The basin-wide average of the rms differences, approximately 0.25 dyn cm(-2), is of similar magnitude to the average wind stress anomaly. Discrepancies in the interannual variability of the wind products lead to large-scale differences in the model sea level anomalies of up to 9-21 cm. The anomalous year to year variability of model sea level is of similar order in each of the simulations. Results for the FSU and SAWIN forced simulations are generally in better agreement with the observations than the FNOC simulation, especially during the 1982-1983 El Nino.