Using Model Cloud Information to Reassign Low-Level Atmospheric Motion Vectors in the ECMWF Assimilation System

Abstract

Abstract This paper investigates the use of model cloud information in the assimilation of low-level atmospheric motion vectors (AMVs) in the ECMWF global data assimilation system, with the aim to characterize and address issues encountered in the assimilation of these observations. An analysis of background departure statistics (comparison of observations with the model background) shows that AMVs placed above the model cloud show larger deviations from the model background relative to those placed unrealistically close to the surface. Reassigning the pressure of AMVs diagnosed above the model cloud layer to either the model cloud top, cloud base, or average pressure leads to improvements in root-mean-square vector difference (RMSVD) and speed bias against the background wind fields. In assimilation experiments, reassigning AMVs placed above the model cloud to the model cloud top, cloud base, or average pressure results overall in a positive impact on subsequent forecasts. The reassignment to an average model cloud pressure performs best in this respect, and this approach has been implemented in the operational ECMWF system in October 2021.