Vous l’avez sans doute déjà repéré : sur la plateforme OpenEdition Books, une nouvelle interface vient d’être mise en ligne.
En cas d’anomalies au cours de votre navigation, vous pouvez nous les signaler par mail à l’adresse feedback[at]openedition[point]org.

Précédent Suivant

Sub-chapter 3.4.2. Forecast of heavy precipitation events

p. 577-585

Texte intégral


1Forecasting and managing severe natural hazards are key topical issues, especially when considering the scale of damage, their social and economic consequences and their after-effects. Legal liabilities can only be met by a policy of management of natural hazards and selecting the appropriate means for the appropriate goals.

2In the Mediterranean region, intense rainfall events constitute a real danger to the population. Heavy rainfall, sometimes associated with strong winds, can cause floods, particularly during the autumn season, with dire consequences for people and the environment. Although the course of events cannot be changed (from a meteorological point of view), it is now possible to better understand these phenomena and their dynamics and thus to warn the populations of their occurrence from a few hours to several days in advance.

3These heavy precipitation events are favoured by the particular local geography of the Mediterranean basin, a nearly enclosed sea, surrounded by mountainous regions. All countries bordering the Mediterranean basin are subject to these severe events, for instance, North Africa with the flash-flood episode, which struck Morocco in December 2006, Algiers in November 2001, frequent episodes along the Spanish eastern coast, in the South of France and along the Italian coast. These natural disasters have been responsible for many casualties and huge economic losses.

4Although much progress has been made in recent years in understanding the driving mechanisms behind these systems and in improving their forecast, the precise forecasting of the location of such episodes remains a major challenge. The final quality of the weather forecast depends mainly on three factors. Accurate knowledge of the initial state of the atmosphere is provided by an optimal analysis of recent observations. The second factor consists of expertise able to provide an estimate of the future atmospheric state, which generally comes from a combination of numerical weather prediction models describing the evolution of atmospheric parameters (pressure, wind, temperature and humidity) and of forecasters who interpret model results and translate them into weather-sensitive elements not directly provided by modelling, such as visibility. The third factor represents the meteorological information focus on the end-users’ requirements (in terms of communication and information availability).

Numerical prediction of heavy precipitation systems

5Atmospheric dynamics and thermodynamics are relatively complex. Only the use of modelling provides an efficient means to gather knowledge and optimize forecasting and monitoring processes. In many countries, limited area models with a finer spatial resolution than global large-scale models are run daily. One such model is the 10 km resolution ALBACHIR model which simulates the atmospheric circulation over Morocco and is run in real time by the Moroccan weather service, DMN (fig. 1a). A kilometric horizontal resolution, together with detailed initial conditions, has been shown to significantly improve the convection simulation (Ducrocq et al. 2002) and has motivated the design of kilometric-resolution operational models such as the AROME model in Météo-France. AROME is currently run in many countries such as Morocco at a 2.5 km resolution (fig 1b). Since April 2015, the resolution of the French AROME version is 1.3 km and 90 levels (fig 1c, Brousseau et al. 2016), which helps the model to describe more realistic convective cells in terms of size, number, intensity and lifetime.

6An ensemble prediction system (EPS) based on AROME (Raynaud and Bouttier, 2016), complementing the deterministic system with a quantification of forecast uncertainties in terms of localisation and intensity, will become operational by the end of 2016 at Météo-France.


Figure 1
Domains of the operational regional numerical weather prediction models of the Moroccan Meteorological Service ALBACHIR a) AROME-MAROC b) and AROME domain at Météo-France c).

Observations and their use in forecast models

7Accurate initial conditions, especially in terms of wind and humidity description at low levels and in the mid-troposphere, are necessary for a realistic simulation of key aspects of precipitating systems (intensity and localisation, Ducrocq et al. 2000 and 2002). High-resolution models represent convective cells with significant small-scale memory: specific features from dissipation convection (such as gust fronts or cold pools) may influence the development of new convective systems. In this context, the AROME model may benefit from high-frequency observations, through frequent updates of the initial conditions. These updates are provided using data assimilation which combines information from the model (namely, a short term forecast) and observations. Data assimilation aims to readjust the trajectory of the numerical mode at regular intervals, so that it is the closest to the actual state of the atmosphere at the beginning of the forecast. Algorithms such as variational assimilation are used in many countries and are particularly efficient for handling large numbers of different types of observations. These algorithms provide a corrected initial condition, optimally combining information from a forecast and observations, over the whole model domain. In addition, they allow for the assimilation of many data which are indirectly linked to the model quantities, such as remote sensed data (like cloudy satellite observations or precipitating radar data).

8At mesoscale, as it is necessary to have information at a high spatial and temporal resolution, meteorological radars appear to be particularly relevant for characterizing rainy and precipitating events. They can provide wind measurements and reflectivities, delivering information on precipitating hydrometeors at a kilometric resolution every 15 minutes. The cycled data assimilation of joint Doppler winds and reflectivity significantly improved probabilistic forecast rainfall scores as well as the localization of convective cells (Wattrelot et al. 2014). Radar networks have been deployed in many countries. With the acquisition of weather radar (7 Doppler radars), a project to integrate radar data is being initiated for AROME-Morocco (fig. 2).


Figure 2
Moroccan radar network.

9At European level, the EUMETNET/OPERA programme aims to promote radar data exchange in Europe. The latest phase of this programme (OPERA4) focused on the preparation and the provision of radar data for numerical weather prediction models. In the context of an experimental campaign, a preliminary study at Météo-France showed the potential benefit in using both Spanish and French radar data together in the initial state of the AROME forecast (fig. 3).

Vigilance (weather alert) maps and real-time monitoring of heavy precipitation

10In many countries, vigilance maps are produced twice a day to warn populations of impending meteorological hazards. They publicize the likelihood of a severe event threatening one or more provinces within the next 24 hours. They are now available in many countries. Two examples are given for France and Morocco respectively (fig. 4). A four-color weather alert code is used to evaluate the risk level. For a very highly probable severe weather hazard, i. e. an orange or red alert, monitoring bulletins are issued as often as necessary. These describe the current phenomenon in progress, its evolution in the coming hours, likely consequences and potential damage and the behaviors to adopt.

11For example, the Moroccan weather service, fully aware of its national and international responsibilities, launched the Vigiobs project with a view to improving early warning systems for natural disasters. In the context of this project, significant work has been done on the reliability of the observation network at stake on the quality of the measurement, the concentration of data and supervision of the network. Work tools forecasting services have gradually adapted to incorporate new features implemented in the framework of Vigiobs.

12The financial effort made by DMN and the priority given to the Vigiobs network deployment enabled the forecasting services to have a network of ground observations that best meet their needs. The frequency of messages sent by the stations varies between one minute and 24 hours, which can be configured remotely from the central hub, depending on the weather conditions. As a rule, messages are sent every hour. In addition to the Vigiobs surface network, DMN operates a network of seven Doppler radars, three second-generation satellite reception stations and a lightning detection network. These observation systems have enabled better weather condition monitoring in real time and have improved the accuracy and timeliness of warnings.


Figure 3
For 28 September 2012 at 18 UTC: a comparison between the observed radar composite in dBz (top) with the location of radar (black triangles) and simulated reflectivities in dBz (at 700 hPa altitude) by the Mediterranean version of AROME, from a 18 hour forecast which assimilates radial winds and reflectivities from Spanish radars (bottom). The legend scale is the same for both images and indicated in equivalent rain rate in mm/h.


Figure 4
Example of vigilance maps with a four colour scale produced by Moroccan Weather Service (top) and by Météo-France (bottom).

Research to better understand and forecast heavy precipitation events

13The Hydrological cycle in the Mediterranean Experiment (HyMeX) (Drobinski et al. 2014) is a 10-year international programme. This project aims for a better understanding and quantification of the hydrological cycle and related processes in the Mediterranean Sea. It focuses on high-impact weather events, inter-annual to decennial variability of the Mediterranean coupled system, and associated trends in the context of global climate change. The first HyMeX special observation period (SOP1) took place from 05 September to 06 November 2012 over the north-western Mediterranean Sea including the French, Italian and Spanish coastal regions. SOP1 was dedicated to heavy precipitation and flash flooding (Ducrocq et al. 2014).

14This experiment also provided the opportunity to assimilate new observations or improve the assimilation of already existing ones. Studies on the radar parameters distinguishing the dominant hydrometeors, or indirectly linked with low-level humidity, have been conducted and experimental data such as lidar data have been assimilated. Furthermore, this experimental campaign represented a test-bed for future kilometre-scale ensemble prediction systems.


15Progress in modelling of heavy precipitation, flash-floods and vulnerabilities now enables us to consider designing seamless modelling approaches in order to meet societal expectations. Progress is also expected in the coupling of atmospheric, oceanic and hydrological models which will improve heavy precipitation and flash-flood forecasts. For example, the AROME EPS will drive a full hydrological ensemble prediction system for probabilistic flash-flood forecasting and flash-flood warning. In the longer term, improvements of exposure and vulnerability modelling capabilities will enable the coupling of these models with heavy precipitation and flash-flood hazard models towards the development of impact-based prediction models. Such models should help ensure effective responses from decision makers in disaster management.


Des DOI sont automatiquement ajoutés aux références bibliographiques par Bilbo, l’outil d’annotation bibliographique d’OpenEdition. Ces références bibliographiques peuvent être téléchargées dans les formats APA, Chicago et MLA.


10.1002/qj.2822 :

Brousseau P., Seity Y., Ricard D., Léger J., 2016
Improvement of the forecast of convective activity from the AROME-France system. Q.J.R. Meteorol. Soc.. doi: 10.1002/qj.2822

10.1175/BAMS-DD-1212-00242.1 :

Drobinski P, Ducrocq V, Alpert P, et al. 2014
HyMeX: A 10-year multidisciplinary program on the Mediterranean water cycle. Bull. Am. Meteorol. Soc. 95: 1063 – 1082, doi: 10.1175/BAMS-D-12-00242.1.

Ducrocq V., Lafore J.-P., Redelsperger J.-P., Orain F. 2000
Initialization of a fine-scale model for convective-system prediction: A case study. Q. J. R. Meteorol. Soc., 126: 3041 – 3065

Ducrocq V, Ricard D, Lafore JP, Orain F. 2002
Storm-scale numerical rainfall prediction for five precipitating events over France: On the importance of the initial humidity field. Weather and Forecasting, 17: 1236 – 1256.

10.1175/BAMS-DD-1212-00244.1 :

Ducrocq V, Braud I, Davolio S, et al. 2014
HyMeX-SOP1: The field campaign dedicated to heavy precipitation and flash flooding in the Northwestern Mediterranean. Bull. Am. Meteorol. Soc. 95: 1083 – 1100, doi: 10.1175/BAMS-D-12-00244.1.

10.1002/qj.2686 :

Raynaud L., Bouttier F., 2016
Comparison of initial perturbation methods for ensemble prediction at convective scale. Q.J.R. Meteorol. Soc., 142: 854 – 866. doi: 10.1002/qj.2686

10.1175/MWR-D-1313-00230.1 :

Wattrelot E., Caumont O., Mahfouf J. F., 2014
Operational implementation of the 1D+ 3D-Var assimilation method of radar reflectivity data in the AROME model. Monthly Weather Review, 142(5):1852-1873. DOI: http ;//dx.doi.org/10.1175/MWR-D-13-00230.1

Précédent Suivant

Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.