Weather forecasting nowadays often requires some estimation of uncertainties associated with the output of meteorological models, in order to better inform decision making, especially in the context of intense weather events.Ensemble prediction systems provide such information through sets of possible scenarios which are designed to represent various uncertainty sources, including model uncertainties. A wide variety of methods have been proposed to estimate model uncertainties, among which perturbation methods targeting uncertain processes are a promising research field. In this study, we focus on the representation of small-scale variability by process-oriented perturbation schemes applied to two key physical processes, namely turbulence and shallow convection. The perturbations are applied to a single-column version of the convection-permitting AROME model, in three idealized boundary-layer cases.Large-eddy simulations (LESs) of the same cases serve as a reference for the subgrid variability that has to be represented, and the results are also compared to those given by the Stochastically Perturbed Parametrization Tendencies (SPPT) method, which is a method commonly used by weather forecast centres to represent model uncertainty. The spread produced by our process-oriented perturbations of turbulence and shallow convection does not represent all the small-scale variability implied by the LESs for temperature and humidity. However, it is of a similar order of magnitude for the wind, thanks to perturbations generated by the stochastic turbulence scheme. The dispersion is structurally different from what is obtained with SPPT. It is non-negligible in the lower levels, where SPPT perturbations are usually suppressed because of numerical instabilities, indicating a possible complementarity between the schemes.

Various perturbation schemes have been proposed for representing model error in convectionpermitting ensemble prediction. Their evaluation usually relies on time-averaged ensemble prediction statistics and on complex case studies. In this work, their detailed physical behaviour is studied in order to understand their differences, and to help their optimization. A process-level intercomparison framework is used to investigate the widely used SPPT (stochastic perturbations of physics tendencies), independent SPPT (iSPPT), and random parameters (RP) model perturbation schemes. Ensemble predictions with the single-column version of the Arome numerical weather prediction model are evaluated on three different boundary-layer regimes: cumulus convection, stratocumulustopped boundary layer, and radiation fog. The iSPPT scheme is found to produce more dispersion than the SPPT scheme, particularly when several physics parametrizations are in near equilibrium. It also appears to be more numerically stable near the surface. The RP scheme perturbations are structurally very different from the other schemes, particularly regarding cloud structure. The iSPPT and RP ensembles have very different sensitivities to the atmospheric conditions, which suggests that intercomparisons of ensemble model error schemes should carefully account for situation dependency. Substantial forecast biases are produced by the RP scheme with respect to the unperturbed model. These results suggest that the iSPPT scheme can bring major improvements over the SPPT approach with minimal effort, that there is some complementarity between the iSPPT and RP approaches, but that implementing RP-type schemes in operational applications may require some careful tuning to avoid creating forecast biases.

Abstract. This paper presents a methodological framework designed for the event-based evaluation of short-range hydrometeorological ensemble forecasts, in the specific context of an intense flash-flood event characterized by high spatiotemporal variability. The proposed evaluation adopts the point of view of end users in charge of the organization of evacuations and rescue operations at a regional scale. Therefore, the local exceedance of discharge thresholds should be anticipated in time and accurately localized. A step-by-step approach is proposed, including first an evaluation of the rainfall forecasts. This first step helps us to define appropriate spatial and temporal scales for the evaluation of flood forecasts. The anticipation of the flood rising limb (discharge thresholds) is then analyzed at a large number of ungauged sub-catchments using simulated flows and zero-future rainfall forecasts as references. Based on this second step, several gauged sub-catchments are selected, at which a detailed evaluation of the forecast hydrographs is finally achieved. This methodology is tested and illustrated for the October 2018 flash flood which affected part of the Aude River basin (southeastern France). Three ensemble rainfall nowcasting research products recently proposed by Météo-France are evaluated and compared. The results show that, provided that the larger ensemble percentiles are considered (75th percentile for instance), these products correctly retrieve the area where the larger rainfall accumulations were observed but have a tendency to overestimate its spatial extent. The hydrological evaluation indicates that the discharge threshold exceedances are better localized and anticipated if compared to a naive zero-future rainfall scenario but at the price of a significant increase in false alarms. Some differences in the performances between the three ensemble rainfall forecast products are also identified. Finally, even if the evaluation of ensemble hydrometeorological forecasts based on a low number of documented flood events remains challenging due to the limited statistical representation of the available data, the evaluation framework proposed herein should contribute to draw first conclusions about the usefulness of newly developed rainfall forecast ensembles for flash-flood forecasting purpose and about their limits and possible improvements.

<p>A new methodological framework for the event-based evaluation of high resolution short-range hydro-meteorological ensemble forecasts is presented. It is specifically designed to address the questions of spatial and temporal scales at which ensemble forecasts should be evaluated, according to the characteristics of a flash-flood event. We adopt the point of view of end-users in charge of organizing evacuation and rescue operations. For this purpose, the potentially local exceedance of discharge thresholds need to be anticipated in time and accurately localized in space. A step-by-step approach is proposed, involving, first, an evaluation of the rainfall forecasts to define the spatial and temporal scales for the event-based evaluation. Second, a spatial analysis of the anticipation lead-times of hydrological responses is performed, focusing on the flood rising limbs, with the evaluation carried out against a reference forecast based on simulated flows. Based on this second step, several gauged sub-catchments are selected, at which a detailed evaluation of the hydrological forecasts is finally conducted.</p><p>This methodology has been tested and illustrated on the October 2018 flash-flood which affected part of the Aude River basin (south-eastern France). Three ensemble rainfall nowcasting research products recently developed by the French meteorological service (Météo-France) coupled with rainfall-runoff models (GRSDi and CINECAR) are evaluated and compared. The originality of the method is in the evaluation of the whole hydro-meteorological forecast chain by defining criteria corresponding to the users. The evaluation may seem limited (a single event and a limited number of outlets) but flash floods justify the implementation of such an evaluation framework.</p><p>Even if evaluating ensemble hydro-meteorological forecasts based on a limited number of documented flood events remains dependent on the statistical representativeness of the available data, the evaluation framework proposed herein helps drawing rapid and robust conclusions about the usefulness of newly developed rainfall ensemble forecast approaches and about their limits and improvement possibilities. It also contributes to pull together the community towards better framing post-event evaluations, so that we can put together events and evaluations from different parts of the world to collectively enhance our capacity to forecast, take decisions and increase preparedness for floods.</p>

Flash-flood events can have catastrophic socio-economic consequences. To reduce their impacts, it is of crucial importance to set up efficient warning systems. Although first operational flash-flood warning systems have recently been implemented, some limitations are clearly identified by end-users: non-exhaustive geographic coverage, limited lead times, warnings based on hazard assessment instead of risk. However, the desirable improvements raise real scientific challenges in various domains. In this context, the PICS (Prevision immediate des impacts des crues soudaines -Flash-flood events impacts nowcasting -2018-2022) project gathers French scientific teams with varied skills (meteorologists, hydrologists, hydraulicians, economists, social geographers) and operational stakeholders (civil security, local authorities, insurance companies, managers of hydroelectric facilities and transport network). Funded by the French national research agency (ANR), it aims to develop and evaluate pre-operational forecasting chains able to estimate the potential impacts of flash floods with short anticipation lead times (up to 6 hours). These modelling chains include different components. Distributed hydrological models transform the observed and forecasted rainfall into runoff. Hydraulic models translate this runoff into potential flooded areas. Impact models incorporate these results to evaluate the potential for social and economic impacts. The research and operational partners selected four case studies based on various criteria, including the occurrence of human impacts and damages, the availability of validation data. Validation data include discharges recorded at gauging stations, but also more original information collected after each event, such as peak discharges and maximum water levels estimated from flood marks, insurance claims, damages observed on infrastructure (roads, railway…), victims interviews, casualties,etc. This presentation focuses on the methodology used for the involvement of representative potential end-users, leading to fruitfull dialogues and informative outcomes. Some of the first results of the project are also presented.