Abstract/Summary

CloudSat observations are used in combination with collocated European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis to simulate spaceborne W-band Doppler observations from slant-looking radars. The simulator also includes cross-polarization effects which are relevant if the Doppler velocities are derived from polarization diversity pulse pair correlation. A specific conically scanning radar configuration (WIVERN), recently proposed to the ESA-Earth Explorer 10 call that aims to provide global in-cloud winds for data assimilation, is analysed in detail in this study. One hundred granules of CloudSat data are exploited to investigate the impact on Doppler velocity estimates from three specific effects: (1) non-uniform beam filling, (2) wind shear and (3) crosstalk between orthogonal polarization channels induced by hydrometeors and surface targets. Errors associated with non-uniform beam filling constitute the most important source of error and can account for almost 1 m s−1 standard deviation, but this can be reduced effectively to less than 0.5 m s−1 by adopting corrections based on estimates of vertical reflectivity gradients. Wind-shear-induced errors are generally much smaller (∼ 0.2 m s−1 ). A methodology for correcting these errors has been developed based on estimates of the vertical wind shear and the reflectivity gradient. Low signal-to-noise ratios lead to higher random errors (especially in winds) and therefore the correction (particularly the one related to the wind-shear-induced error) is less effective at low signal-to-noise ratio. Both errors can be underestimated in our model because the CloudSat data do not fully sample the spatial variability of the reflectivity fields, whereas the ECMWF reanalysis may have smoother velocity fields than in reality (e.g. they underestimate vertical wind shear). The simulator allows for quantification of the average number of accurate measurements that could be gathered by the Doppler radar for each polar orbit, which is strongly impacted by the selection of the polarization diversity H − V pulse separation, Thv. For WIVERN a selection close to 20 µs (with a corresponding folding velocity equal to 40 m s−1 ) seems to achieve the right balance between maximizing the number of accurate wind measurements (exceeding 10 % of the time at any particular level in the mid-troposphere) and minimizing aliasing effects in the presence of high winds. The study lays the foundation for future studies towards a thorough assessment of the performance of polar orbiting wide-swath W-band Doppler radars on a global scale. The next generation of scanning cloud radar systems and reanalyses with improved resolution will enable a full capture of the spatial variability of the cloud reflectivity and the in-cloud wind fields, thus refining the results of this study.