Winds are used in this study to estimate Ekman pumping and to force a Rossby wave model (Section 4.2.1). Since the winds in this region are known to have small spatial and temporal scales (Chelton et al., 2000a), especially in the case of the mountain-gap wind jets that are found to have great importance, and because these calculations require spatial derivatives of the winds, excellent sampling is a necessity. This points to the utility of satellite scatterometer winds that can resolve the wind jets better than any in situ or present-generation reanalysis wind product (Schlax et al., 2001). The Quikscat satellite carrying a Seawinds scatterometer (referred to here as Quikscat) was launched in June 1999 and its operational products became available in late July 1999. The satellite circles the earth in a 4-day repeat, sun-synchronous orbit with a spacing of 6.3° at the equator, sampling a 1600-km wide swath centered on its ground track. The first three years of Quikscat wind stresses were obtained on a 1° × 1° grid from the web site of Center for Ocean-Atmosphere Prediction Studies (http://www.coaps.fsu.edu/scatterometry/) at Florida State University (FSU), and used to represent the mean winds.
The European Research Satellite (ERS) was launched in July 1991, and its nearly identical follow-on mission was launched in April 1995 and remains in operation (but a gyroscope failed in January 2001, degrading the data). Data used here include the period August 1991 through December 2000. ERS data are processed onto a 1° × 1° monthly grid by the Centre ERS d'Archivage et de Traitement (CERSAT), at IFREMER, Plouzan'e (France), and are available at http://www.ifremer.fr/cersat. The ERS sampling is not as good as Quikscat, with a 500-km wide swath and 35-day repeat cycle. However, the nearly 10 year record is more suitable for estimating the average annual cycle, and the ERS winds are used here to force a Rossby wave model (Section 4.2.1). Extensive comparison has been made between the two scatterometer products, and also with in situ wind products (the widely used Florida State University (FSU) product, which spans 40 years (Stricherz et al., 1992)). All these winds show quite a consistent picture in the eastern tropical Pacific; the differences are essentially that Quikscat gives greater spatial detail than ERS, which in turn gives greater detail than FSU. Given this, we have chosen to use Quikscat to represent the mean winds, since the spatial detail is valuable to show the effects of the Central American wind jets, but to use ERS for the average annual cycle used to force the Rossby wave model since its longer record suggests a better resolution of the time variability.
Return to previous section or go to next section