An idealized dry primitive equation model on the f-plane is used to study u
pstream (and downstream) baroclinic wave development. The simulations are i
nitiated with localized finite amplitude and vertically evanescent perturba
tions, specified either as upper-level potential vorticity or surface poten
tial temperature anomalies. The nonlinear evolution of these nonmodal pertu
rbations leads to the generation of large-scale upper-level induced primary
and downstream surface cyclones, and of distinctively smaller, shallow and
more slowly intensifying upstream systems. It is shown that in particular
the genesis and evolution of upstream cyclones is highly sensitive to the s
cale of the initial perturbation. Narrow upper-level troughs (or zonally co
nfined surface temperature anomalies) are favorable for upstream developmen
t, whereas no or only weak upstream activity occurs with broad planetary-sc
ale troughs (or zonally extended surface temperature anomalies) as initial
perturbations. It is proposed that this sensitivity property of upstream de
velopment is qualitatively related to the dispersion characteristics of sur
face edge waves.
The shortcomings of the present approach are discussed, and some considerat
ion is given to the occurrence of upstream cyclogenesis in the real atmosph
ere, to the relationship with earlier concepts of secondary cyclogenesis, a
nd to possible implications for the issue of predictability of extratropica
l weather systems.