Data to accompany the article "The Predictability of Midlatitude Cyclones: Don't Sweat the Small Stuff"

Abstract

We use convection-permitting idealized simulations of moist midlatitude cyclones to compare the growth of synoptic-scale initial perturbations derived from an adjoint model with the growth of monochromatic-wavelength perturbations at the smallest resolved scale. When the initial-error magnitudes are comparable to those from present-day data assimilation systems, the adjoint-derived perturbations significantly change the intensity and location of the cyclone by projecting onto the synoptic-scale baroclinically unstable growth. In contrast, the wave-like perturbations grow with the moist convection, but the errors remain on the mesoscale and localized to the precipitating regions through lead times of up to 4~days, causing only negligible changes to the midlatitude-cyclone forecast. The adjoint-derived perturbations produce the most significant changes despite the fact that the adjoint model is run at a much coarser resolution to satisfy the tangent-linear approximation, and as such has the potential to miss nonlinear contributions to error growth from moist convection. When the initial-error magnitudes are reduced by factors of 10 and 100, the adjoint-derived and wave-like perturbations both grow rapidly with the moist convection, but this growth does not significantly alter the cyclone's development. Using quasigeostrophic potential vorticity inversion, we investigate the balanced response produced by the perturbations after 4~days of growth. Only the full-magnitude adjoint-derived perturbations lead to balanced growth with the synoptic-scale baroclinic instability; the balanced error signatures in the other experiments remain on the mesoscale and localized to the convection, in contrast to a widely cited conceptual model for error growth. These results suggest that synoptic-scale initial errors that project strongly onto regions of high adjoint sensitivity are more important for 2--4-day practical midlatitude-cyclone forecasts than arbitrarily small-scale initial errors that grow with the moist convection.