Scatterometers and other satellite winds

by Mary Forsythe — last modified Jun 11, 2014 04:48 PM

Satellite vector winds from scatterometers are available in swaths of Wind Vector Cells of size 10-25 km. Their resolution is typically 25-50 km and no evidence of error correlation beyond these scales exist. The estimated accuracy of these scatterometer vector winds is typically 1 m/s VRMS. Ocean surface wind component measurements are further available from Synthetic Aperture Radar (single-look backscatter and Doppler) and from microwave radiometers.

Moreover, ESA will launch a Doppler Wind Lidar, called Aeolus, in 2015, which will provide profiles of (mainly) the zonal wind component in clear air and at cloud tops. After aggregation over 80 km, a component accuracy of typically 2-3 m/s is expected. Stereoscopic cloud top winds are obtained from multi-look imagers such as MISR at, e.g., 17 km sampling.

Wind products

• Spatial representation of scatterometer winds is investigated by spectral and spatial analyses; triple collocation analyses provide calibration and error budgets of scatterometer, NWP and in situ (reference) winds (e.g., Vogelzang et al., 2011);
• Recent work shows that ASCAT scatterometer winds provide good winds and probably wind variability near convection;
• Quality control is challenged by spatial representation errors in dynamical areas (with rain);
• Ambiguity removal by 2DVAR at full scatterometer resolution works well;
• Extreme surface winds from scatterometers are calibrated against the NOAA hurricane hunter dropsondes and Stepped-Frequency Microwave Radiometer (SFMR);

Data assimilation

• Assimilated winds need to be consistent with the NWP model climatology in both calibration and deterministic spatial characteristics; such consistency generally reduces o-b departures;
• A 3-hour data assimilation window, implying 1.5 hour time differences, is insufficient to capture fast-moving mesoscale wind features (3Dvar);
• Insufficient temporal sampling of scatterometer winds is available to allow a deterministic initialisation and evolution of NWP winds on the (scatterometer) mesoscale; however, e.g., in coastal areas, deterministic forcing by orography or land may show real NWP wind features in the mesoscale;
• Ambiguity removal after thinning in a 3D of 4D Var DAS approach may be more challenging and coordinated tests will be useful, e.g., assimilating 2DVAR products;
• Good quality, but unrepresentative scatterometer winds may exist, e.g.,  near convection, where a more strict QC could be beneficial in NWP (TBC);

References and links

• EUMETSAT OSI SAF, produces NRT scatterometer winds; EU Copernicus MyOcean provides L3 and L4 gridded wind products;
• EUMETSAT NWP SAF, develops scatterometer wind processors, monitoring and data assimilation guidance;
• IOVWST for in-depth ocean vector wind studies;

Speed calibration, spatial representativeness studies, QC, ambiguity removal?