Bonnie R. Brown and Gregory J. Hakim
Department of Atmospheric Sciences, University of Washington,Seattle, WA
Journal of the Atmospheric Sciences, 69, submitted.
The internal variability and predictability of three-dimensional
hurricanes is investigated using 100-day-long, statistically steady
simulations in a compressible, non-hydrostatic, cloud-resolving model. The equilibrium solution is free of the
confounding effects of initial conditions and environmental
variability in order to isolate the ``intrinsic'' characteristics of the
hurricane.
Results show that the variance of the axisymmetric azimuthal velocity
is greatest inside the radius of maximum wind and is dominated by two
patterns: one characterize by a radial shift of the maximum wind, and
the other by intensity modulation at the radius of maximum wind. These
patterns are associated with bands of anomalous wind speed that
propagate inward from large radii with a period of roughly 5 days. The
largest increases in axisymmetric tangential wind associated with the
shifting pattern are opposed, on average, by the asymmetric component
of the wind. An analysis of the asymmetric azimuthal velocity
component shows that it is generally strongest inward of the
axisymmetric radius of maximum wind. The time-mean asymmetric contribution to the tendency of the
azimuthal-mean tangential wind is diffusive, with a negative tendency near the
radius of maximum wind.
Predictability of axisymmetric storm structure is measured through the
autocorrelation $e$-folding time and linear inverse modeling. Results
from both methods reveal an intrinsic predictable timescale of about
two days. The predictability as well as the variability of the
axisymmetric component are consistent with recently obtained results
from idealized two-dimensional hurricane modeling.