Regional accuracy of QuikSCAT gridded winds

The quality of gridded 00 UTC and 12 UTC QuikSCAT wind speed fields provided by the Florida State University (FSU) and NASA Jet Propulsion Laboratory (JPL) are analysed over the Bay of Bengal during May–August 2001. Additionally, an examination of these fields is performed over the Gulf of Mexico for the May–August period from 2001 to 2003. Both 00 UTC and 12 UTC time almost coincide with QuikSCAT sampling times (twice a day) and correspond to either early morning or late evening local time over these regions. The primary restriction for generating accurate maps with a temporal resolution of 12 hours and less is a lack of adequate sampling. Due to non-uniform spatialtemporal sampling of the scatterometer, interpolation procedures cannot provide proper estimates in data gaps over the regions not covered by a swath. Wind speed autocorrelation coefficients for gridded datasets have been compared with that of original QuikSCAT data and buoy winds. It is shown that the spatial and temporal interpolation used to obtain these datasets results in smoothing of the QuikSCAT wind speeds, reducing their original variance. This smoothing is amplified where substantial diurnal wind variability occurs. A comparison with buoy data shows that FSU and JPL gridded fields are unable to reproduce correctly observed low correlations in wind speed between morning and evening breeze flows and diurnal variability of winds, leading to a reduced difference between 00 UTC and 12 UTC values in comparison with buoys and swath QuikSCAT data. Rather, the FSU and JPL maps describe daily mean fields. Another consequence of the spatial-temporal interpolation is that the winds are distorted at a frequency matching the dominant sampling interval (3–4 days) of QuikSCAT measurements over the Bay of Bengal.