Wednesday, 25 January 2017: 2:15 PM
605 (Washington State Convention Center )
There are limited ocean surface vector wind observations in the Tropical Atlantic. These are mainly the sixteen PIRATA buoys along with some
observations from volunteer ship observations, which are largely constrained to a few narrow shipping tracks that themselves have seasonal
variability. Because satellite scatterometers provide very high quality ocean vector winds, we seek a method by which a useful climate data
record can be constructed from them without resorting to the use of "background" fields from numerical analyses. This can then be used in data
assimilation and as an independent metric for validating model diagnostics. It requires consideration of the satellite sampling pattern on the resulting averaged data. Even though approximately 95% of the ocean surface is sampled daily by QuikSCAT, the Equatorial region is less well-sampled than regions nearer to the poles. Satellite ocean vector wind data are provided by polar-orbiting scatterometers and along inclined swaths of wind vector cells with nominal spacings of 12.5 or 25 km. For the case of the NASA QuikSCAT scatterometer, the swath width is ~1800 km and each orbit is about 100 minutes long. In those 100 minutes, at the equator the earth will have rotated approximately 2800 km. This means that there are approximately 1000 km wide gaps between successive ascending overpasses. We construct the gridded data as a weighted average of those points that fall within a pair of time and space windows surrounding the time and location of interest. Input data that are closer in time and space are weighted higher than less recent or more distant data. We choose the time and space windows by matching estimates of the sub-grid wind variability measured spatially and temporally using collocated swath and ocean surface PIRATA buoy data adjusted to neutral stratification and 10 m anemometer height. For example, we find that the sub-grid wind variability for a time window of 72 hours matches that of a spatial window of approximately 1000 km. The resulting gridded wind fields are evaluated against ICOADS and ECMWF surface wind fields using a variety of methods including triple collocation analysis.
observations from volunteer ship observations, which are largely constrained to a few narrow shipping tracks that themselves have seasonal
variability. Because satellite scatterometers provide very high quality ocean vector winds, we seek a method by which a useful climate data
record can be constructed from them without resorting to the use of "background" fields from numerical analyses. This can then be used in data
assimilation and as an independent metric for validating model diagnostics. It requires consideration of the satellite sampling pattern on the resulting averaged data. Even though approximately 95% of the ocean surface is sampled daily by QuikSCAT, the Equatorial region is less well-sampled than regions nearer to the poles. Satellite ocean vector wind data are provided by polar-orbiting scatterometers and along inclined swaths of wind vector cells with nominal spacings of 12.5 or 25 km. For the case of the NASA QuikSCAT scatterometer, the swath width is ~1800 km and each orbit is about 100 minutes long. In those 100 minutes, at the equator the earth will have rotated approximately 2800 km. This means that there are approximately 1000 km wide gaps between successive ascending overpasses. We construct the gridded data as a weighted average of those points that fall within a pair of time and space windows surrounding the time and location of interest. Input data that are closer in time and space are weighted higher than less recent or more distant data. We choose the time and space windows by matching estimates of the sub-grid wind variability measured spatially and temporally using collocated swath and ocean surface PIRATA buoy data adjusted to neutral stratification and 10 m anemometer height. For example, we find that the sub-grid wind variability for a time window of 72 hours matches that of a spatial window of approximately 1000 km. The resulting gridded wind fields are evaluated against ICOADS and ECMWF surface wind fields using a variety of methods including triple collocation analysis.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner