EnKF assimilation of simulated spaceborne Doppler observations of vertical velocity: impact on the simulation of a supercell thunderstorm and implications for model-based retrievals
Résumé
It is shown that the RMS errors of the analyzed fields produced by assimilation of vertical velocity alone are in general better than those obtained in previous studies: in most cases assimilation of vertical velocity alone leads to analyses with small errors (e.g. <1 ms-1 for velocity components) after only 3 or 4 assimilation cycles. The microphysical fields are notable exceptions, exhibiting lower errors when observations of reflectivity are assimilated together with observations of vertical velocity, likely a result of the closer relationship between reflectivity and the microphysical fields themselves. It is also shown that the spatial distribution of the error estimates improves (i.e. approaches the true errors) as more assimilation cycles are carried out, which could be a significant advantage of EnKF model-based retrievals.