Examination of a Stochastic and Deterministic Convection Parameterization in the COSMO Model

Abstract

Stochastic parameterizations allow the representation of the small-scale variability of parameterized physical processes. This study investigates whether additional variability introduced by a stochastic convection parameterization leads to improvements in the precipitation forecasts. Forecasts are calculated with two different ensembles: one considering large-scale and convective variability with the stochastic Plant-Craig convection parameterization and one considering only large-scale variability with the standard Tiedtke convection parameterization. The forecast quality of both ensembles is evaluated in comparison with radar observations for two case studies with weak and strong synoptic forcing of convection and measured with neighborhood and probabilistic verification methods. The skill of the ensemble based on the Plant-Craig convection parameterization relative to the ensemble with the Tiedtke parameterization strongly depends on the synoptic situation in which convection occurs. In the weak forcing case, where the convective precipitation is highly intermittent, the ensemble based on the stochastic parameterization is superior, but the scheme produces too much small-scale variability in the strong forcing case. In the future, the degree of stochastic variability could be tuned, and these results show that parameters should be chosen in a regimedependent manner.

Evaluation of the Plant–Craig stochastic convection scheme (v2.0) in the ensemble forecasting system MOGREPS-R (24 km) based on the Unified Model (v7.3)

“…This score (denoted FSS) was developed by Roberts and Lean (2008), and was used by Kober et al (2015) to assess the quality of deterministic forecasts produced using the Plant-Craig scheme for two case studies. Note that we use the term "deterministic", in this manuscript, to refer to forecasts providing a single quantity (for example, a singlemember forecast, or the ensemble mean), and "probabilistic" to refer to forecasts providing a probabilistic distribution (or, at the very least, a deterministic forecast, with, in addition, an assessment of its uncertainty).…”

“…In particular, the stronger the largescale forcing, the lower the proportion of the variability that comes from the stochastic scheme. Kober et al (2015) investigated two of the case studies further, by verifying forecasts using the Plant-Craig scheme and using a non-stochastic convection scheme. They found that the improvement in forecast quality from using the PlantCraig scheme was significantly higher for the more weakly forced of the two cases, since the additional grid-scale variability introduced by the stochastic scheme is more important.…”

“…As part of the present study, we extend the work of Kober et al (2015) by separating our validation period into dates for which the synoptic forcing is relatively weak or strong. We then compare any improvement in the forecasts using the Plant-Craig scheme, over those using the GregoryRowntree scheme, for the two sets of forecasts, to assess over an extended period whether the benefit of using a stochastic scheme is indeed greater when the synoptic forcing is weaker.…”

“…As well as this, the present study aims to reveal more about the behaviour of the scheme itself, building on work referenced above, as well as on recent work by Kober et al (2015), which focused on individual case studies.…”

“…Such "stochastic" convection parameterization schemes have been developed over the last 10 years and are just beginning to be implemented and verified in operational forecasting set-ups, with some promise for the improvement of probabilistic ensemble forecasts (e.g Teixeira and Reynolds, 2008;Bengtsson et al, 2013;Kober et al, 2015). The purpose of the present study is to continue this pioneering work of verifying probabilistic forecasts using stochastic convection parameterizations, by investigating the performance of the Plant and Craig (2008) (PC) scheme in the Met Office Global and Regional Ensemble Prediction System (MO-GREPS) (Bowler et al, 2008).…”