I. Szunyogh et al., A COMPARISON OF LYAPUNOV AND OPTIMAL VECTORS IN A LOW-RESOLUTION GCM, Tellus. Series A, Dynamic meteorology and oceanography, 49(2), 1997, pp. 200-227
We compare the first local Lyapunov vector (LLV) and the leading optim
al vectors in a T10/18 level truncated version of the National Centers
for Environmental Prediction global spectral model. The leading LLV i
s a vector toward which all other perturbations turn and hence it is c
haracterized by the fastest possible growth over infinitely long time
periods, while the optimal vectors are perturbations that maximize gro
wth for a finite time period, with respect to a chosen norm. Linear ta
ngent model breeding experiments without convection at T10 resolution
show that arbitrary random perturbations converge within a transition
period of 3 to 4 days to a single LLV. We computed optimal vectors wit
h the Lanczos algorithm, using the total energy norm. For optimization
periods shorter than the transition period (about 3 days), the horizo
ntal structure of the leading initial optimal vectors differs substant
ially From that of the leading LLV, which provides maximum sustainable
growth, There are also profound differences between the two types of
vectors in their vertical structure. While the 24-hour optimal vectors
rapidly become similar to the LLV in their vertical structure, change
s in their horizontal structure are: very slow. As a consequence, thei
r amplification factor drops and stays well below that of the LLV for
an extended period after the optimization period ends, This may have a
n adverse effect when optimal vectors with short optimization periods
are used as initial perturbations for medium-range ensemble forecasts.
The optimal vectors computed for 3 days or longer are different. In t
hese vectors, the fastest growing initial perturbation has a horizonta
l structure similar to that of the leading LLV, and its major differen
ce from the LLV, in the vertical structure, tends to disappear by the
end of the optimization period. Initially, the optimal vectors are hig
hly unbalanced and the rapid changes in their vertical structure are a
ssociated with geostrophic adjustment. The kinetic energy of the initi
al optimal vectors peaks in the lower troposphere, whereas in the LLV
the maximum is around the jet level. During the integration the phase
of the streamfunction held of the optimal vectors, with respect to the
ir corresponding temperature field, is rapidly shifted 180 degrees. An
d, due to drastic changes that also rake place in the vertical tempera
ture distribution, the maximum baroclinic shear shifts from the lower
troposphere to just below the jet level. Just after initial rime, when
the geostrophic adjustment dominates, the leading optimal vectors exh
ibit a growth rate significantly higher than that of the LLV. By the e
nd of the period of optimization, however, the growth rate associated
with the leading optimal vectors drops to or below the level of the Ly
apunov exponent. The transient super-lyapunov growth associated with t
he leading optimal vectors is due to a one-time-only rapid rotation of
the optimal vectors toward the leading LLVs. The nature of this rapid
rotation depends on the length of the optimization period and the nor
m chosen, We speculate that the initial optimal vectors computed with
commonly used norms may not be realizable perturbations in a dynamical
system (that is not forced by random noise) since the leading optimal
vectors are directions in the phase space from which perturbations ar
e ''repelled''. By contrast, the leading LLV is the vector toward whic
h all random perturbations, including the optimal vectors, are attract
ed, which gives the LLV a unique role in linear perturbation developme
nt.