Abstract. A modification of the widely used SPPT (Stochastically
Perturbed Parametrisation Tendencies) scheme is proposed and tested in a
Convection-permitting – Limited Area Ensemble Forecasting system (C-LAEF) developed at
ZAMG (Zentralanstalt für Meteorologie und Geodynamik). The tendencies from four
physical parametrization schemes are perturbed: radiation, shallow convection,
turbulence, and microphysics. Whereas in SPPT the total model tendencies are perturbed,
in the present approach (pSPPT hereinafter) the partial tendencies of the physics
parametrization schemes are sequentially perturbed. Thus, in pSPPT an interaction between
the uncertainties of the different physics parametrization schemes is sustained and a
more physically consistent relationship between the processes is kept. Two configurations
of pSPPT are evaluated over two separate months (one in summer and another in winter).
Both schemes increase the stability of the model and lead to statistically significant
improvements in the probabilistic performance compared to a reference run without
stochastic physics. An evaluation of selected test cases shows that the positive effect
of stochastic physics is much more pronounced on days with high convective activity.
Small discrepancies in the humidity analysis can be dedicated to the use of a very simple
supersaturation adjustment. This and other adjustments are discussed to provide some
suggestions for future investigations.