Observing-System Research and Ensemble Data Assimilation at JAMSTEC - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 508

Fig. 21.3 The analysis
ensemble spread of sea-level
pressure in the Arctic ( a )on
10 January 2006 in ALERA
and ( b ) on 10 January 2008 in
ALERA2
to suppress precipitation. Recent convective parameterization schemes are designed
to suppress such precipitation by including sensitivity to environmental humidity
( Emori et al. 2001 ; Betchold et al. 2008 ; Chikira and Sugiyama 2010 ). We also mod-
ified the Emanuel scheme this way, but it has not yet been adopted in ALEDAS2.
There is another reason for the excessive weak precipitation: precipitation is a
highly non-Gaussian process. Under a conditionally unstable environment, a small
perturbation can ignite convective parameterization. If precipitation is generated
with only a few members with modest intensity, the ensemble mean precipitation
will be weak rather than non-existent. With a convectively unstable profile, almost
all members yield precipitation, and the distribution will be closer to Gaussian.
Consequently, the ensemble mean agrees well with observations in the regions with