Vertical Mode Initialization in a Limited Area Data Assimilation Experiment

Abstract

A previously reported 9-day limited area data assimilation experiment has been reported, incorporating a recently developed nonlinear vertical mode initialization scheme. It is shown that the initialization scheme significantly reduces surfaces pressure oscillations during the model integration, producing more accurate guess fields for each analysis. It is demonstrated, by means of objective verification statistics and by means of a case study, that these more accurate guess fields result in improved analyses and a small increase in skill of 24 h prognoses based on these analyses.