Noninertial flow in NSCAT observations of Tehuantepec winds

NASA scatterometer (NSCAT) observations provide high temporal and spatial resolution wind fields, which are used to examine gap flow through the Chivela Pass and the influence of Hurricane Marco. For approximately 1 week in November 1996, Hurricane Marco caused the flow through the gap to change from its usual flow pattern: the winds turned to the left rather than the right. Previous studies of this gap flow have used monthly averages of sparse ship observations, European Centre for Medium-Range Weather Forecasts fields, mesoscale models, or proxies such as cloud motion or sea surface temperatures (SSTs). NSCAT provided unprecedented accuracy (rms differences less than 1.5 m s(-1) and 15 degrees) and resolution (daily and 1 degrees x 1 degrees in this study). The observations show that the gap flow often turns to the right (dominated by the Coriolis force); however, unusual events can cause highly noninertial flow for several days. The SSTs respond slowly to changes in the wind pattern, and they are a poor proxy for shortterm wind patterns. The NSCAT winds, gridded with a new methodology presented herein, have the spatial and temporal resolution required to show the evolution of the gap outflow in the presence of a hurricane. The wind fields are used to generate parcel pseudotrajectories, which show that Marco caused Tehuantepec winds to turn to the left and pass through the Gulf of Papagayo into the Caribbean Sea. Gridding techniques-are required to fill gaps in fields of NSCAT observations. NSCAT coverage of the lice free) global oceans is approximately 77% in 1 day and 90% in 2 days. The new gridding technique temporally averages the winds in a manner that leaves very little evidence of the pattern of satellite tracks in the wind fields and little evidence of wind field curl and divergence. Daily fields of 1 degrees X 1 degrees resolution are generated. The temporal sampling characteristics of the wind fields are shown to be nonhomogeneous, with the distribution of characteristic sampling times peaking at 1 day and usually within the range of 0.75 to 1.75 days.