Abstract
This study examines ERS-1 3-day repeat orbit scatterometer wind data from January through March 1992. The study region encompasses the North Pacific from 30° to 50°N and 160°E to 130°W longitude. The data are separated by orbit trajectory and binned to 26 km. These data are examined by direct comparative analysis to surface European Centre for Medium-Range Weather Forecasts (ECMWF) model analyses on daily, monthly, and 3-month timescales. The scatterometer wind fields compare favorably, but distinct, nonisolated differences exist. These differences, exhibited in the scatterometer winds, include slightly stronger wind speeds, more distinct curvature, and detail on structures smaller than the ECMWF resolution. Systematic relative northward displacements of cyclonic centers, 1°–3° in latitude, in ECMWF surface winds are also indicated. The scatterometer wind retrieval algorithm (CMODFD/NSCAT MLE) demonstrates some difficulty in selecting the true wind vector. Problems are generally identifiable by inspection. Complex empirical orthogonal function (EOF) analysis on the ascending and descending scatterometer wind fields reveal frequency and amplitude information about the sampled variance. The first four EOFs, for which the results suggest physically motivated phenomena, account for 50%–60% of the total variance sampled in the data. The EOF results partition the sampled variance in the ascending and descending data and suggest the more significant EOFs depict spatiotemporal “bands” of 18–21, 8–10, and 6–8 days, reflecting the planetary wave cycle. large-scale general circulation systems, and smaller-scale storm structures, respectively. The partitioning of the variance demonstrates only limited filtering capability in identifying erroneous ERS-1 wind vectors.
