Ensemble-Based Estimation of Observation Impact-Simplified AGCM Perfect Experiments

Experiments of assessing observation impact with a simplified parameterization AGCM model are carried out to examine how the ensemble-based estimation method proposed by Liu and Kalnay (2008) (LK08) can be applied to the state-of-art models. The results show that LK08 can successfully estimate each observation's impact at once and then the impact values can be simply summed and grouped according to various types of observations or different areas. Although the summed observation impacts for the two hemispheres are similar, the individual observation impact in the Southern Hemisphere is much larger than that in the Northern Hemisphere due to the sparse observations there. The impacts for different observational types are also different. The estimated total observation impact accounts for 70%－80% of the actual impact and captures the variations of actual impact very well.