July 28, 2011

The MetOp-A ASCAT Level 2 Ocean Surface Wind Vectors Optimized for Coastal Ocean dataset provides enhanced coastal coverage of ocean surface wind vectors over the global ice-free oceans at a sampling resolution of 12.5 km. The Ocean and Sea Ice Scatterometer Application Facility (OSI SAF) provides this experimental dataset in near-real-time (NRT) to supplement the existing Level 2 NRT ASCAT ocean wind vector datasets.

Figure Caption: "This image compares the wind vectors from the OSI SAF Operational 12.5 km ASCAT data set (left panel) with wind vectors from the OSI SAF Coastal 12.5 km ASCAT data set (right panel) in the western Gulf of Mexico region on the morning (ascending node) of 6 September 2010. Tropical Storm Hermine can be seen just west of the Yucatan Peninsula."
This image compares the wind vectors from the OSI SAF Operational 12.5 km ASCAT data set (left panel) with wind vectors from the OSI SAF Coastal 12.5 km ASCAT data set (right panel) in the western Gulf of Mexico region on the morning (ascending node) of 6 September 2010. Tropical Storm Hermine can be seen just west of the Yucatan Peninsula.

The enhanced coastal coverage is made possible using a spatial box filtering technique (rather than the traditional Hamming filter) which is used to generate a spatial average of the full resolution (Level 1) backscatter (Sigma-0) measurements within a 15 km radius of the wind vector cell center; prior to averaging, all land/ice contaminated Sigma-0 measurements are discarded. This unique sampling approach provides wind retrievals as close as approximately 15 km from the coast, which is in contrast to the static 35 (70) km land mask in the standard (i.e., operational) 12.5 (25) km dataset. The European Organization for the Exploitation of Meteorological Satellites (EUMETSAT) OSI SAF production and quality monitoring information is available at the Royal Netherlands Meteorological Institute (KNMI) through www.knmi.nl/scatterometer/.

Cal/Val Summary (Verhoef and Stoffelen, 2011): The data characteristics of the coastal dataset in the open ocean (beyond 50 km from coastline) have been shown to be very similar to the operational 12.5 km dataset. Direct comparisons with the operational 12.5 km dataset show there is no wind speed bias and very low standard deviations in the vector components (0.39 m/s for U; 0.53 m/s for V). The coastal dataset, however, flags less points in the QC step than the operational dataset does, probably due to its reduced spatial smearing. Comparisons to in situ buoys (winds adjusted to 10 m and equivalent neutral) reveal a slight negative speed bias with the Coastal dataset in the coastal areas (-0.23 m/s) which is a bit less than the negative speed bias observed in open ocean (-0.29 m/s); the latter is very similar to the open-ocean negative speed bias observed with the operational 12.5 km dataset (-0.28 m/s). Standard deviation of the difference with respect to buoys is notably better in open-ocean regions than coastal regions for the U (i.e., Zonal) vector component (1.57 m/s for coastal; 1.48 m/s for open-ocean), where the difference in the V component is statistically identical for both open-ocean (1.61 m/s) and coastal buoys (1.60 m/s). This result was anticipated as coastal winds are known to be more variable than open-ocean winds and thus buoys do not represent the 30-km winds as well here, which leads to the slightly larger differences. The speed standard deviation of difference with buoys is very similar for collocated coastal and operational 12.5 km dataset in open ocean regions (resp. 1.44 and 1.43 m/s for U; 1.54 and 1.56 m/s for V). Spectral analysis shows very similar and consistent spectral slopes for the 12.5 km operational and coastal datasets with no significant flattening at the higher spatial frequencies, the latter of which indicates no significant white noise in both of the wind datasets.
Data: This experimental dataset is provided in the same netCDF, CF-compliant format as the operational ASCAT wind datasets. Due to the experimental nature of this dataset, its usage and any scientific results derived from its usage should be treated with caution and discretion. Unlike the operational wind datasets, the historical data record of this dataset is considerably recent, beginning on 18 August 2010. The spatial distribution of wind vector cells is identical to the operational 12.5 km dataset, with the exception of an additional number of valid wind vector retrievals in coastal regions, which comes as a bi-product of the spatial box filtering technique. All quality flags, metadata, and file naming conventions are consistent with the operational datasets. This data may be accessed through the PO.DAAC web portal. Please direct all inquiries to the PO.DAAC Help Desk: podaac@podaac.jpl.nasa.gov

References:

  • Verhoef, A. and A. Stoffelen (2011), Validation of ASCAT coastal winds, Tech. Report, version 1.3, SAF/OSI/CDOP/KNMI/TEC/RP/176, Sponsored by EUMETSAT.
  • Verhoef, A., M. Portabella and A. Stoffelen (2011), High-resolution ASCAT scatterometer winds near the coast, submitted, IEEE Trans. Geosci. Remote Sens.