hide
Free keywords:
-
Abstract:
How well can we predict the weather? Questions about the limits of predictability pose an intriguing and societally highly relevant problem. The limit of predictability is directly related to the growth of forecast errors in scale and amplitude. Following seminal work by Ed Lorenz in the late 1960’s, many studies have taken a broad-brush view on the dynamics that underly the growth of forecast errors, namely that scale interactions occur as if the atmosphere behaved like homogenous and isotropic turbulence. Applying a quantitative potential-vorticity framework, new insight into the growth of forecast errors in the midlatitudes is gained. In a conceptual, multi-stage error-growth model, upper-tropospheric divergent outflow is identified as the most efficient process that projects uncertainties in the representation of moist processes onto the tropopause dynamics. Nonlinearities in the Rossby wave dynamics then dominates further growth, while the direct role of baroclinic instability is small, on average, in contrast to widely held previous ideas. Our improved mechanistic understanding of error growth finally suggests an avenue to identify forecasts situations in that the limits of predictability are already reached, i.e., in that no further improvements to state-of-the-art forecast systems would yield a measurable increase in forecast skill. Informing model evaluation about the closeness of forecasts to the intrinsic limit may be helpful in future strategies to examine model error.