Multiple-scale error growth in a convection-resolving model

Abstract

Abstract. The properties of the multiple-scale instabilities present in a non-hydrostatic forecast model are investigated. The model simulates intense convection episodes occurring in northern Italy. A breeding technique is used to construct ensembles of perturbations of the model trajectories aimed at representing the instabilities that are responsible for error growth on various timescales and space scales. By means of perfect model twin experiments it is found that, for initial errors of the order of present-day analysis error, a nonnegligible fraction of the forecast error can be explained by a bred vector ensemble of reasonable size representing the growth of errors on intermediate scales. In contrast, when the initial error is much smaller, the spectrum of bred vectors representing the fast convective-scale instabilities becomes flat, and the number of ensemble members needed to explain even a small fraction of the forecast error becomes extremely large. The conclusion is that as the analysis error is decreased, it becomes more and more computationally demanding to construct an ensemble that can describe the highdimensional subspace of convective instabilities and that can thus be potentially useful for controlling the error growth.

Comparison of initial perturbation methods for ensemble prediction at convective scale

“…This behaviour is in agreement with findings from (e.g.) Uboldi and Trevisan (2015), Harnisch and Keil (2015), Wang et al (2014), and results from an early saturation of ensemble perturbations at smaller scales. At longer ranges, namely from 12 to 21/24 h, the ensemble spread increases similarly whatever the ICs.…”

“…Experiments with a twelve‐member EDA at 2.5 km provide the same conclusions (not shown). It is worth mentioning that this result is in agreement with the study of Uboldi and Trevisan (), who find that, in a convection‐resolving non‐hydrostatic model, an initial error comparable to current analysis error can be well described by an ensemble of reasonable size.…”

“…This behaviour is in agreement with findings from (e.g.) Uboldi and Trevisan (), Harnisch and Keil (), Wang et al . (), and results from an early saturation of ensemble perturbations at smaller scales.…”

Scientific Challenges of Convective-Scale Numerical Weather Prediction

“…Generating such a homogeneous sample with a reasonable, finite ensemble size is not a simple matter, and it becomes more difficult for a system with an increasing number of unstable modes (cf. Uboldi and Trevisan 2015).…”

Evaluating a stochastic parametrization for a fast–slow system using the Wasserstein distance

“…In particular, it is crucial to derive efficient and accurate ways to surrogate the effect of dynamical processes occurring on the small spatial and temporal scales that are not explicitly resolved (e.g., because of excessive computational or storage costs) by the model. The operation of constructing so-called parametrizations is key to the development of geophysical fluid dynamical models and stimulates the investigation of the fundamental laws defining the multiscale properties of the coupled atmosphere-ocean dynamics (Uboldi and Trevisan, 2015;Vannitsem and Lucarini, 2016). Traditionally, the development of parametrizations boiled down to deriving deterministic empirical laws able to describe the effect of the small-scale dynamical processes.…”