460 ASSESSMENT OF OCEAN SURFACE WIND VECTOR CLIMATE DATA RECORDS FROM CROSS-CALIBRATED SATELLITE DATA SOURCES

Tuesday, 8 January 2013
Exhibit Hall 3 (Austin Convention Center)
David F. Moroni, JPL, Pasadena, CA; and M. A. Bourassa and P. J. Hughes

Three data sets containing an assimilated analysis of satellite derived ocean surface wind vector retrievals are evaluated to assess uncertainty and suitability as a Climate Data Record (CDR). Two of the data sets utilize an NWP-derived background analysis wind field for gap-filling at 6-hourly intervals; the two datasets assimilate identical satellite data, but differ in how the data is assimilated as well as which background wind field was chosen. The third data set is pre-averaged at monthly intervals and approximates a merged satellite-only data set with no background filling. The satellite data sources for each data set have been thoroughly cross-calibrated and quality controlled to help provide for stability in the time series and reduction of sensor-to-sensor bias. Derivative fields, such as the curl, are used as a proxy for the uncertainty for each data set. Results show consistently the uncertainty as a function of the curl is not consistent with time. In particular, two distinct points in the time series, both representing minima and maxima in the number of observations respectively, show uniquely different surface features and spatial resolutions. Dynamical features that resemble transient storms, eddies, and currents are clearly visible during the year representing the maxima, but remain unrecognizable or nonexistent induring the minima. The pre-averaged monthly satellite-only data set depicts substantial noisiness and aliasing in the curl, likely due to the interpolation error induced by the NWP-assigned surface wind directions during months where scatterometer observations are minimal or unvavailable; even with the presence of two scatterometer sources providing a nominal global coverage of four daily observations, there is substantial residual aliasing. These results suggest that quality-controlled data assimilation at fine temporal spatial scales can be used to remove most if not all aliasing artifacts that are due to inherent limitations such as the availabity of satellite input data and the corresponding spatial resolutions of the input data. Such improvements can and have already been utilized to generate a high quality CDR for ocean surface wind vectors over the globe.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner