We examine the linearity of the Ensemble of Data Assimilations (EDA) technique with
respect to the amplitude of the applied observation perturbations. We provide explicit
examples to assess the linear relationship between such modifications of the observing
system and the resulting changes in the EDA ensemble spread. The results demonstrate
that, for a state-of-the-art numerical weather prediction (NWP) system, such linearity
between the applied observation perturbations and the EDA ensemble spread holds well
for temporal and spatial regimes relevant to global medium-range weather prediction:
specifically, for forecast lead-times of up to approximately 5 days, in the vertical throughout
the troposphere up to the lower and middle stratosphere and for broad horizontal scales.
L Megner, DGH Tan, H Kornich, A Horanyi, A Stoffelen, GJ Marseille. Linearity aspects of the ensemble of data assimilations technique
Status: published, Journal: Quart. J. Royal Meteor. Soc., Year: 2014, doi: DOI:10.1002/qj.2362