Nat. Hazards Earth Syst. Sci., 12, 2671–2688, 2012 www.nat-hazards-earth-syst-sci.net/12/2671/2012/ Natural Hazards doi:10.5194/nhess-12-2671-2012 and Earth © Author(s) 2012. CC Attribution 3.0 License. System Sciences Heavy precipitation events in the Mediterranean: sensitivity to cloud physics parameterisation uncertainties S. Fresnay1, A. Hally1, C. Garnaud1,*, E. Richard1, and D. Lambert1 1Laboratoire d’Aerologie,´ UMR5560, CNRS and Universite´ de Toulouse, Toulouse, France *currently at: ESCER Center, Universite´ du Quebec´ a` Montreal,´ Montreal,´ Canada Correspondence to: S. Fresnay ([email protected]) Received: 28 January 2012 – Revised: 18 June 2012 – Accepted: 29 June 2012 – Published: 24 August 2012 Abstract. In autumn, southeastern France is often affected the microphysical perturbations. These results clearly show by heavy precipitation events which may result in damag- that the usefulness of an ensemble prediction system based ing flash-floods. The 20 October and 1 November 2008 are upon microphysical perturbations is case dependent. Addi- two archetypes of the meteorological situations under which tional experiments indicate a greater potential for the multi- these events occur: an upper-level trough directing a warm process ensemble when the model resolution is increased to and moist flow from the Mediterranean towards the Cevennes´ 500 m. ridge or a quasi stationary meso-scale convective complex developing over the Rhone valley. These two types of events exhibit a contrasting level of predictability; the former being usually better forecast than the latter. Control experiments 1 Introduction performed with the Meso-NH model run with a 2.5 km res- olution confirm these predictability issues. The determinis- The Mediterranean basin is a region prone to high rain- tic forecast of the November case (Cevennes´ ridge) is found fall and stormy events causing severe damage and human to be much more skilful than the one for the October case losses almost every year. These high-impact meteorological (Rhone valley). These two contrasting situations are used events need to be accurately forecasted, justifying impor- to investigate the sensitivity of the model for cloud physics tant cooperative research efforts, such as the WMO World parameterisation uncertainties. Three 9-member ensembles Weather Research Programme MEDiterranean EXperiment are constructed. In the first one, the rain distribution inter- (MEDEX, http://medex.aemet.uib.es/) or the forth-coming cept parameter is varied within its range of allowed val- HYdrological cycle in Mediterranean EXperiment (HyMeX, ues. In the second one, random perturbations are applied to http://www.hymex.org/). The former is dedicated to the study the rain evaporation rate, whereas in the third one, random of high-impact cyclones over the Mediterranean, whereas the perturbations are simultaneously applied to the cloud auto- latter aims at a better understanding and forecasting of the conversion, rain accretion, and rain evaporation rates. Re- whole water cycle in the Mediterranean. In particular, the sults are assessed by comparing the time and space distri- first HyMeX Special Observing Period (SOP), scheduled for bution of the observed and forecasted precipitation. For the autumn 2012, will specifically focus on heavy precipitation Rhone valley case, it is shown that not one of the ensem- events in the northwestern Mediterranean and will provide a bles is able to drastically improve the skill of the forecast. unique opportunity to implement and assess new forecasting Taylor diagrams indicate that the microphysical perturba- methodologies aiming to improve the predictability of these tions are more efficient in modulating the rainfall intensi- events. ties than in altering their localization. Among the three en- In southeastern France, climatology exhibits a peak of sembles, the multi-process perturbation ensemble is found to high precipitation events in October with more than 90 high yield the largest spread for most parameters. In contrast, the rainfall days – daily rainfalls greater than 150 mm – reg- results of the Cevennes´ case exhibit almost no sensitivity to istered for the 1967–2006 period. These events mainly oc- cur over the Cevennes´ range (see Fig. 1 for geographical Published by Copernicus Publications on behalf of the European Geosciences Union. 12 S. Fresnay, A.Hally, E. Richard, C. Garnaud and D. Lambert: Heavy precipitation events in the Mediterranean Ensemble name Initial conditions Horizontal resolution Type of perturbation E1 20-10-08 00 UTC 2.5 km Rain intercept parameter E2 20-10-08 00 UTC 2.5 km Rain evaporation E3 20-10-08 00 UTC 2.5 km Rain evaporation, autoconversion, accretion E4 01-11-08 00 UTC 2.5 km Rain evaporation, autoconversion, accretion E5 20-10-08 00 UTC 2.5 km Initial conditions E6 20-10-08 06 UTC 2.5 km Rain evaporation, autoconversion, accretion E7 20-10-08 06 UTC 0.5 km Rain evaporation, autoconversion, accretion Table 1. Characteristics of the different ensembles 2672 S. Fresnay et al.: Heavy precipitation events in the Mediterranean To overcome this current limitation, ensemble prediction is a promising strategy. From a set of perturbed scenarios, which model the uncertainties of the initial atmospheric state, ensemble prediction gives a probabilistic picture of the likeli- hood of the phenomena. The main difficulty associated with this strategy is to design appropriate and realistic perturba- tions to get enough dispersion of the scenarios and signifi- cant statistical scores. At the global and synoptic scales, en- semble prediction systems are based on well-proved method- ologies but much less knowledge exists for the convective scale. Hohenegger and Schar¨ (2007) investigated how dif- ferent mesoscale errors propagate through a cloud-resolving model. It appears that within a dozen hours, various per- turbations, even randomly distributed, lead to a maximum spread over regions where moist convection takes place. This is an indication that precise meso-scale perturbations could only be valuable for a very short forecasting range, whereas random perturbations might induce similar spread for more extended forecasts. Furthermore, from these results, which clearly underline the role of the convective processes, it could Fig. 1. Map of the domain of Southern France used for the sim- be hypothesised that the microphysical cloud scheme uncer- Fig. 1. Map of the domain of Southern Franceulations, used with for a the 2.5-km simulations, resolution. with Solid a line 2.5- indicateskm resolution. box where Solid linetainties indicates also play box wherea major statistics role in the have error propagation and statistics have been performed, that is the Languedoc and Southern been performed, that is the Languedoc and Southern Massif Central area. Shading represents altitudes (intherefore meters) should over 250alsom. be taken Geographica into accountl in a convective- Massif Central area. Shading represents altitudes (in meters) over names and French administrative regions250 are m. recalled, Geographical in particular names and 4 departements French administrativeof the regions Languedoc are areascale with ensemble their standard prediction number. system. recalled, in particular 4 departements of the Languedoc area with Past studies have addressed the issue of physical param- their standard number. eterisation uncertainties. They are cast into two categories involving either the use of different physical parameterisa- tion schemes (e.g. Houtekamer et al., 1996; Stensrud et al., locations) but also hit the highly-populated plains close by, 2000) or stochastic physical parameterisations (e.g. Buizza posing a serious threat to the population and to economic ac- et al., 1999b; Palmer, 2001; Grell and Dev´ enyi´ , 2002). In the tivities. Catastrophic floods, such as the 1999 Aude case or first category, model uncertainties are accounted for by using the 2002 Gard case, deserved several studies (e.g. Bechtold various combinations of different boundary layers, micro- et al., 2001, Nuissier et al., 2008, Ducrocq et al., 2008) help- physics, and land surface schemes. To be valuable, such an ing to identify the main factors responsible for the severity of ensemble requires the availability of a wide range of parame- these events. In particular, the stationarity and location of the terisations within the model. In the second category, stochas- mesoscale convective system (MCS) for the Gard case were tic physics aim at representing model uncertainty that results shown to be very sensitive to evaporation processes and re- from interactions with unresolved scales. Therefore, this ap- sulting cold pool developments. The presence of a cold pool proach was mainly used at large scales to compensate for was also suspected for an unusually high precipitation event deficiencies in turbulence and deep convection parameterisa- on the eastern shore of Corsica (Lambert and Argence, 2008). tions. Despite all the progress in numerical weather predic- Our goal in this study is to explore the importance of tion, predictability of these events, which involve intricate model uncertainties associated with the cloud representa- multi-scale processes, remains fairly limited. The theoretical tion at the convective scale. Two cases of high precipitation framework of predictability was pioneered by Lorenz (1963), over southern France, exhibiting very different levels of pre- who studied the propagation of small initial errors through dictability, are considered. The sensitivity