Simulation of W-band radar reflectivity for model validation and data assimilation
Abstract
This paper describes a reflectivity forward operator developed for the validation and assimilation of W-band radar data into regional Arome (Applications de la Recherche à l'Opérationnel à Méso-Echelle) class Numerical Weather Prediction models. The forward operator is consistent with the Arome ICE3 one-moment microphysical scheme and is devised for vertically pointing radars. A new neighbourhood validation method, called the Most Resembling Column (MRC) method, is designed to disentangle spatial location model errors from errors in the forward operator. This novel method is used to validate the forward operator using data collected in diverse conditions by the airborne cloud radar RASTA (Radar Airborne System Tool for Atmosphere) during a two-month period over a region of the Mediterranean. The MRC method is then applied to retrieve the optimal effective shapes (i.e. the mean axis ratios) of the predicted graupel, snow and pristine ice, by minimising the standard deviation between observations and simulations. The optimal mean axis ratio is approximately 0.7 for snow and 0.8 for graupel. It is shown that treating snow and graupel particles as oblate spheroids with axis ratios close to their optimal values leads to good agreement between the observations and simulations of the ice levels. Conversely, there is a large bias if snow and graupel particles are considered to be either spherical or overly flattened. The results also indicate that pristine ice can be approximated by a sphere, but this conclusion should be taken cautiously since the amount of pristine ice particles is probably overestimated in the ICE3 microphysical scheme.