Abstract: The development and implementation of a real-time flood forecasting system with a hydro-meteorological operational alert procedure during the MAP-D-PHASE Project is described in this paper. This chain includes both probabilistic and deterministic forecasts. The hydrological model used to generate the runoff simulations is the rainfall-runoff distributed FEST-WB model, developed at Politecnico di Milano. The observed data to run the control simulations were supplied by ARPA-Piemonte. The analysis is focused on Maggiore Lake basin, an Alpine basin between North-West of Italy and Southern Switzerland. Two hindcasts during the D-PHASE period are discussed in order to evaluate certain effects regarding discharge forecasts due to hydro-meteorological sources of uncertainties. In particular, in the June convective event it is analysed how the effect of meteorological model spatial resolution can influence the discharge forecasts over mountain basins, while in the November stratiform event how the effect of the initial conditions of soil moisture can modify meteorological warnings. The study shows how the introduction of alert codes appears to be useful for decision makers to give them a spread of forecasted QDFs with the probability of event occurrence, but also how alert warnings issued on the basis of forecasted precipitation only are not always reliable.
　　Key words: Hydro-meteorological chain, MAP-D-PHASE, quantitative discharge forecasts, ensemble hydrological forecasts.
　　1. Introduction
　　Over the last 20 years severe river floods have occurred in Europe, causing thousands of deaths and billion of Euros in insured economic losses [1]. Weather forecasts, combined with hydrological models, can be used to define early water-system control actions to take preventive measures or reduce problems with floods, droughts or water quality [2]. Therefore, coupling meteorological and hydrological models became a great issue and challenge in the scientific community during the last two decades.
　　To diminish the impact of floods through an early warning system, in 2003 the European Commission started the development of a EFAS (European flood alert system) to provide medium-range flood simulations across Europe [3, 4]. Other international programmes dealing with these topics were HEPEX in 2004, which aimed at fostering the development of probabilistic hydrological forecasting and corresponding decision making tools [5, 6], AMPHORE (Application des Methodologies de Previsions Hydrometeorologiques Orientees aux Risques Environnementaux), a continuation of HYDROPTIMET [7], mainly devoted to the hydro-meteorological modelling study of heavy precipitation episodes resulting in floods event and the optimisation of the existing warning system in the Western Mediterranean Basin [8, 9], RAPHAEL(Runoff and Atmospheric Process for Flood hazard Forecasting and Control) [10], and the European COST Action 731 (propagation of uncertainty in advanced
　　 meteo-hydrological forecast system) [11, 12].
　　The idea of a forecasting cascade was also one of the scientific objectives of the MAP (Mesoscale Alpine Programme) between 1994 and 2005. A unique initiative to improve the understanding of the processes involved in orographically induced precipitation events and the improvements of the high-resolution numerical weather prediction [13, 14].
　　After these encouraging results obtained in the MAP Project and considering that orographic precipitation has often led to disastrous flooding events over the Alps, it was decided to devote the MAP FDP (Forecast Demonstration Project) to show recent improvements in the operational use of an “end-to-end” forecasting system, consisting of atmospheric models, hydrological prediction systems, nowcasting tools and warnings for end users. The project acronym chosen was D-PHASE that stands for Demonstration of Probabilistic Hydrological and Atmospheric Simulation of flood Events in the Alpine region and is a FDP (Forecast Demonstration Project) of the WWRP (World Weather Research Programme of WMO). The MAP FDP has addressed the entire forecasting chain, ranging from limited-area ensemble forecasting, high-resolution atmospheric modelling(km-scale), hydrological modelling and nowcasting to decision making by the end users, i.e., it is foreseen to set up an end-to-end forecasting system. For a complete review see Refs. [15-18].
　　The use of EPS (ensemble prediction systems), instead of single (deterministic) forecasts for flood warning [19], is increasing among the hydrological community in order to quantify better the uncertainty of flood prediction. From the hydrological perspective using EPS as input to a hydrological model it is an important tool to produce river discharge predictions[20-22] and to assess uncertainty involved in forecasting precipitation [23, 24].
　　In this study it presents a hindcast for two different types of precipitation events that occurred during the D-PHASE Operational Period (DOP), from 1 June to 30 November 2007, in the Verbano basin in order to evaluate certain effects regarding discharge forecasts due to hydro-meteorological sources of uncertainties.
　　Two non-hydrostatic meteorological limited area models are used to force the distributed hydrological model (FEST-WB): one with a coarse spatial resolution, supported by the EPS (COSMO-LEPS system based on COSMO model) and the other with a finer grid, but with one deterministic output only(MOLOCH model).
　　The paper is organized as follows: in both events there is a meteorological introduction of the actual weather scenario, followed by hydrological forecast analyses over the watersheds. In the June convective event, it has been studied how the effect of meteorological model spatial resolution can influence the discharge forecasts over mountain basins (and subbasins), while in the November stratiform event how the effect of the initial conditions of soil moisture can modify meteorological warnings.
　　2. Area of Study
　　The subject area is the Verbano basin, also known as Maggiore, a regulated lake at the border between North-West Italy and South Switzerland. The drainage area covers 6,598 km2: 3,229 km2 in Italy and 3,369 km2 in Switzerland. The study focuses on the three main rivers: the Ticino (1,616 km2), the Toce (1,534 km2) and the Maggia (926 km2) (Fig. 1).
　　Nearly 17% of the total area is above 2,000 m a.s.l. Climate conditions are typically humid, characterized by higher precipitations in autumn, spring and summer with a dry season in winter [25]. Snowfall characterizes precipitation in autumn and winter and snow melting in spring gives a significant contribution to runoff. Climatic characteristics, together with morphology and soil texture, induce frequent flood events (1993, 1994, 1996, 2000 and 2002).
　　Three threshold levels were defined in the framework of the MAP-D-PHASE project to issue meteorological and hydrological warnings for the three
　　main river basins: a “yellow-attention” level when 60-day return period was exceeded, an “orange-alert”level when 180-day return period was exceeded and“red-alarm” level when 10-year return period was exceeded. Rainfall thresholds were estimated for each basin for six durations, ranging from 3 to 72 hours on the basis of statistics of daily precipitation over the Alps and scaling assumptions with respect to storm duration and area. Flood peak thresholds were estimated on the basis of at-site statistics of runoff peaks or from the rational formula. Of course, in operational forecasting levels are usually higher (with return periods of 10 year and more), but low thresholds were chosen in this project in order to see some action on the maps during the very short demonstration phase. Rainfall and discharge thresholds for the three studied river basins are reported in Table 1 and Table 2, respectively.
　　Available digital cartographic data include: the DEM(digital elevation model) available in raster format at 100 m × 100 m resolution, CORINE land cover maps updated in the year 2000, and pedologic characteristics for soils available in vector format [26, 27]. From these basic thematic layers, basin parameters required for the application of the hydrological model, have been derived at a spatial resolution of 1,000 m × 1,000 m. These include: Curve Number [28], flow direction, slope and aspect, residual and saturated soil moisture, albedo, pore size distribution index, saturated hydraulic conductivity, wilting point, field capacity and soil depth.
　　3. Hydrologic and Meteorological Data
　　To calibrate and test the hydrological model FEST-WB meteorological and hydrologic ground measured data were collected by the telemetric monitoring system in Italy and Switzerland. Records of rainfall, air temperature, short wave solar radiation and air relative humidity are available at hourly or sub-hourly time steps. Hydrometric observations at 30 minutes time step are available at Candoglia, Solduno and Bellinzona gauging stations (rivers Toce, Maggia and Ticino respectively, see Fig. 1). All these data were accessible for the time period from 1 January 2000 to 31 December 2003; for a complete review about the FEST-WB hydrological model calibration the reader can refer to Ref. [9].
　　During the MAP-DPHASE period, ground measured meteorological forcings were used for the hydrological model initialization, before being forced by forecasted meteorological fields to predict river discharge, for more details, see section 4 “Coupling strategy”.
　　3.1 Meteorological Models
　　The hydro-meteorological chain includes both probabilistic forecasts based on ensemble prediction systems with a lead time of a few days and short-range forecasts based on a high resolution deterministic atmospheric model, in order to predict the QDF(quantitative discharge forecast). The probabilistic forecast was supplied by COSMO-LEPS model(Consortium for Small-scale Modelling–Limited area Ensemble Prediction System), implemented and developed by A.R.P.A. Emilia-Romagna in the framework of COSMO Consortium [29, 30]. The spatial resolution is 10 km (0.09°), while the temporal resolution is 3 h, with 40 vertical levels, 16 ensemble members, nested on ECMWF EPS (European Centre for Medium Range Forecast-Ensemble Prediction System) and 132 h as lead-time; the run starts every day at 12:00 UTC, while the hydrological simulation begins 12 hours later at 00:00 UTC, so 120 hours of hydrological simulation are available.
　　The deterministic forecast was supplied by MOLOCH model (MOdello LOCale in “H” coordinate) of I.S.A.C.-C.N.R. in Bologna, Italy [31]. The model chain comprises the hydrostatic model BOLAM and the non-hydrostatic model MOLOCH, nested in BOLAM. The BOLAM model has a horizontal grid spacing of 0.11° in rotated coordinates (about 12 km), with 50 levels and a parameterization (Kain-Fritsch) of moist convection. On the contrary, the MOLOCH model has a horizontal grid spacing of 0.02°, corresponding to about 2.3 km, with 50 levels; moist deep convection is computed explicitly. The forecasting chain is based on the 18:00 UTC, ECMWF analysis/forecasts at 0.25° resolution; BOLAM run starts at 18:00 UTC, MOLOCH is nested at 00:00 UTC,
　　the same time as the hydrological simulation.
　　The different spatial resolution used by the two weather models over the Maggiore basin is shown in Fig. 2: the COSMO-LEPS model, with a spatial resolution of 10 km, which covers the Toce basin(1,534 km2) with 15 squares, on the contrary, the high-resolution MOLOCH model, with a spatial grid of 2.3 km, which fills the Toce basin with 317 squares.
　　3.2 Hydrological Model
　　Hydrological simulations were performed using the FEST-WB distributed water balance model [9, 32-34].
　　FEST-WB calculates the main processes of the hydrological cycle: evapotranspiration, infiltration, surface runoff, flow routing, subsurface flow and snow dynamics. The computational domain is discretized with a mesh of regular square cells (1 km in this application), in which water fluxes are calculated at hourly time intervals.
　　Five main components can be identified in the FEST-WB model:(1) the flow paths and channel network definition;(2) the spatialization of site measured meteorological forcings;
　　(3) the snow pack dynamics;
　　(4) the runoff calculation;
　　(5) the overland and subsurface flow routing.
　　In the first component the flow path network is automatically derived from the digital elevation model using a least-cost path algorithm [35]. It assigns flow from each pixel to one of its eight neighbours, without the necessity to remove pits in the elevation data. For hillslope and channel network definition the model uses the constant minimum support area concept. It consists of selecting a constant critical support area that defines the minimum drainage area required to initiate a channel [36].
　　In the second component, the spatial interpolation of ground measured meteorological forcings is carried out. The model requires precipitation, air temperature, air relative humidity, and net solar radiation, sum of short wave and long wave components.
　　The observed data at ground stations are interpolated with a regular grid, using the inverse distance weighting (IDW) technique. In order to facilitate integration with meteorological models, the FEST-WB can also accept spatial gridded meteorological as input.
　　The third component deals with snow dynamics simulation. The snow module of the FEST-WB includes snow melt and snow accumulation dynamics.
　　In the FEST-WB model the partitioning of total precipitation, P, in liquid, Pl, and solid, Ps, phase is a function of air temperature, Ta [37]:
　　 where Cm (m·°C-1·s-1) is an empirical coefficient, depending on meteorological conditions and geographic location; generally Cm coefficient ranges from 4.8 × 10-8 to 6.9 × 10-8 m·°C-1·s-1.
　　The predefined temperature Tb fixes a threshold beyond which snow starts melting, and its value is usually assumed to be equal to 0 °C; Tb and Cm are calibrated values of the model.
　　The terrain covered by snow is supposed to be frozen and hence the melted water is prevented from infiltrating the soil. Conversely, the liquid fraction of snow water equivalent, Rs, the sum of melted water and liquid precipitation, is supposed to flow cell by cell through the snow pack with a linear reservoir routing scheme [40] with a celerity of 1.67 × 10-3 m·s-1 [41]. When Rs reaches a cell not covered by snow, it is added to the liquid precipitation of that cell, becoming part of the infiltrable water.
　　In the fourth component, the runoff is computed for each elementary cell, according to a modified SCS-CN method extended for continuous simulation [33].
　　The potential maximum retention, S, is updated cell by cell at the beginning of rainfall as a linear function of the degree of saturation, ??)
　　(7) where R is surface runoff flux, D is drainage flux, ET is evapotranspiration rate and Z is soil depth. Soil moisture in cells covered by snow is assumed not to vary over time.
　　The actual evapotranspiration, ET, is computed as a fraction of the potential rate calibrated by a function that, in turn, depends on soil moisture content [42]. Potential evapotranspiration is calculated with a radiation-based equation [43].
　　The fifth component performs the runoff routing throughout the hillslope and river network. The surface flow routing, calculated for the cells that are not covered by snow, is based on the Muskingum-Cunge method in its non-linear form with time variable celerity [32]. Subsurface flow routing, similar to the method implemented for routing in the snow pack, is calculated with a linear reservoir routing scheme [40] with a celerity calculated as a function of soil saturated conductivity.
　　The FEST-WB model can save state variables on the file system at regular time intervals, permitting the restart of a simulation from a previous condition. This feature is used where a long simulation has to be carried out at a different time, or to initialize the model before carrying out a forecast run.
　　4. Coupling Strategy
　　The hydro-meteorological chain was launched
　　5. Simulation Results over the Three River Basins
　　 5.1 The June 2007 Event: Effect of Model Spatial Resolution
　　The June event (13-15 June 2007) was the most severe and relevant during the DOP on the Maggiore Lake basin. The synoptic analysis over Europe on 15 June 2007 showed a “cold drop” located South-West of the British Isles, triggering moist flow from the Mediterranean Sea towards the Alps and the Po Valley, causing convective cells with associated thunderstorms on the Lake Maggiore basin 95-110 millimetres fell in three days (13, 14, 15 June) over the three basins(Table 3), about 85-95 of them in only 24 hours between 14 and 15 June.
　　This amount of rainfall yielded the following measured peak discharges: 783.2 m3·s-1 observed at Candoglia (orange warning), 941.7 m3·s-1 at Solduno(orange warning) and 761.5 m3·s-1 at Bellinzona(orange warning); however, these discharge values caused no flood damage in the catchment areas.
　　Before analysing the results, it is important to highlight that since the precipitation forecast of the CLEPS and MOLOCH models over the three Maggiore Lake basins showed some under/overestimation errors during the event, better results in hydrological forecasts were obtained with the meteorological run initialized on the 14 June (i.e. 24-48 hours before the main peak discharge on 15 June) as is shown in the Bow-Whisker plots for all three basins(Figs. 3-5). Thus, all the tables and figures shown regarding this June 2007 event, and related to the simulations driven by observed and meteorological model data, refer to the hydrological simulation launched on 14 June 2007.
　　5.1.1 River Toce at Candoglia
　　Over the Toce basin the two meteorological models were characterized by an opposite behaviour in terms
　　 event.
　　The performance was even worse for the forecast initialized two days before the peak event and for the run started on the same day, as shown in Fig. 3.
　　An opposite result was obtained using the MOLOCH model. In fact, with the one day ahead run, the peak discharge was overestimated (1,162.3 m3·s-1), but the magnitude of the event was correctly predicted, issuing an orange warning (Fig. 6).
　　5.1.2 Impact at Subbasin Scale
　　The COSMO-LEPS forecast precipitation error was
　　 overestimation for some ensemble members in the peak flow.
　　If considering the ensembles median of the cumulated precipitation (red dashed line), there is an underestimation of about 40 mm (?P) in the Bogna subbasin (Fig. 8), in comparison with the subbasin mean rainfall (dark solid line); this produced an underestimation of the forecasted ensemble median
　　 simulated amount was forecasted over the Devero subbasin.
　　5.1.3 River Ticino at Bellinzona
　　Moving now our attention to the Ticino basin, the COSMO-LEPS and MOLOCH models significantly underestimated the cumulative precipitation (Table 6). The maximum observed discharge at Bellinzona was 761.5 m3·s-1 (orange warning), while the simulated maximum discharge by FEST-WB driven by observed hydro-meteorological data was 1,210.4 m3·s-1 (Fig. 10).
　　The relevant overestimation by the hydrological FEST-WB model, when it was forced with a meteorological forecast, was compensated by a precipitation underestimation of the COSMO-LEPS and MOLOCH models. This led to small errors in peak discharge prediction (Table 6 and Fig. 4) and to the correct issuing of the hydrological warning level(Table 5).
　　5.1.4 River Maggia at Solduno
　　Last analyses for this convective summer event were carried out on the Maggia, the smallest of the three catchments. As well as for the other two basins, the two
　　 meteorological models have demonstrated a better hydrological simulation 24-48 hours before the observed peak (Fig. 5).
　　The FEST-WB simulation shows a very good performance (956.4 m3·s-1), in accordance with the observed discharge value (941.7 m3·s-1), both in terms of time (Table 7) and peak flow (Table 6). Although cumulative precipitation was underestimated by both weather models, thanks to an overestimation in precipitation rate, especially in the MOLOCH model(Fig. 11), the hydrological forecasts correctly showed that the orange warning has been exceed (Table 5).
　　In this first experiment carried out in 2007, it is found that an overestimation or underestimation of
　　weather models can be enhanced or smoothed out by hydrological model performance. Errors in meteorological forecasts can be a consequence of the different spatial resolution between the two meteorological models with markedly different orographies as shown in Fig. 2 in Section 3.1. But, another aspect to be considered is that both forecasts are fed into a hydrological model on 00:00 UTC as“initialization time”. This means that the hydrological forecast is based on a 12-hour forecast by CLEPS and at the same hour-forecast by MOLOCH.
　　Especially for short lead times and small catchments, forecast error is in general a function of lead time (the longer the lead time, the more uncertain the forecast). This is probably not as important for the 3- and 2-day forecasts, but it could be for the 1-day forecasts. High-resolution models (like MOLOCH) are mainly targeted for the short-range where more spatial detail is required, while the lower-resolution models such as COSMO-LEPS is mainly targeted for ranges from one day onwards. It is known that it may not be recommended to directly compare the forecast qualities for such different lead times of meteorological forecasts, but the idea it is not to put one model against the other, but to show that forecasting hydro-meteorological systems should be supported by both the two models, i.e. probabilistic and deterministic.
　　5.2 The November 2007 Event: Effect of Soil Moisture Conditions
　　The November event over the Maggiore Lake basin was a stratiform event with light but continuous precipitation over the area, but with very small peak stream-flows in all three main basins. The weather analysis over Europe on 23 November shows a typical autumnal pattern with an upper level trough, coming from the Atlantic Ocean, moving eastward, trigging humid air from the South toward the Southern edge of Italian and Swiss Alps.
　　After the long dry period that hit the Southern edge of the Alps from the beginning of October, this precipitation was the first relevant meteorological phenomenon that occurred after 50 days of the dry autumn season of 2007.
　　For the sake of brevity, it is focused the attention over the Toce basin, where a more detailed physiographic characterization is available. The observed amount of precipitation during this stratiform event (21-24 November) was about 80 mm as a mean basin value (Table 8).
　　Figs. 12 and 13 show the millimetres of precipitation forecasted by the MOLOCH model (Fig. 12) and the expected probability in exceeding the meteorological yellow code by the CLEPS model (Fig. 13) with the weather forecasts initialized on 21, 22 and 23 November 2007.
　　According to the D-PHASE threshold (Tables 1 and 2), the CLEPS and MOLOCH models issued a meteorological warning (yellow code) expected on 22 and 23 November, but the measured peak discharge on 23 November at the Candoglia gauging station was only 57.8 m3·s-1, which is a very low value, with no alert all!
　　As it has been seen for the June event, there is not always a univocal relationship between meteorological and hydrological warnings. In this event, due to the dry antecedent soil condition, the FEST-WB hydrological simulations, forced with forecasted meteorological
　　 data, performed well, issuing no warning. As shown in Fig. 14, both the CLEPS and MOLOCH forecast are very far from the yellow threshold.
　　Looking at the soil moisture field, very dry values near to ?res (Eq. (6)) before the event generally over the Toce basin (Fig. 15-left) are found, and even at the end of the rainfall with the soil not totally saturated, as proof of the drought period that hit North-West Italy during the autumn of 2007; values near to the
　　6. Conclusions
　　This study, part of the project MAP-D-PHASE, evaluated a forecasting operational chain based on a distributed hydrological model, coupled with two weather output: one deterministic of the MOLOCH model and one with an ensemble prediction system of the COSMO-LEPS.
　　These experiments were carried out for two events during the D-PHASE period: a convective event in June and a stratiform event in the autumn over the three main basins of Lake Maggiore: the Toce, Ticino and Maggia.
　　The results obtained show how alert warnings issued on the basis of forecasted precipitation are not always reliable. In fact, an underestimation of the warning level was observed in the June event due to an underestimation of cumulative precipitation. To the contrary an overestimation of the meteorological warning level was observed in the November event.
　　But, hydrological alerts are not the exact consequence of meteorological warnings, above all in mountain watersheds where many uncertainties must be considered in hydrological forecasts. Furthermore, an alert issued on the basis of precipitation only cannot take into account the actual state of the river basin, which is crucial in defining transformation into runoff. Therefore, it is necessary to use a hydrological rainfall-runoff simulation and a coupling strategy between meteorological and hydrological models.
　　When forced with meteorological ground observed data, the FEST-WB performance, in terms of error in simulated peak discharge, was different in the three river basins, with an error ranging from -4% at Solduno to 59% at Bellinzona during the June event. Despite this, the magnitude of the flood event, defined by the warning level, was always correctly simulated. When coupled with the meteorological models, some errors are enhanced, while other are compensated, with a general degradation in the performance of the hydrological model.
　　As far as the peak discharge time lag is concerned(Table 7), the MOLOCH model predicted much better the peak flow arrival, in comparison with the CLEPS’, which is subject to a systematic delay for June event and to an advance bias in November’s.
　　The study highlights how it is important in hydrological simulations to use both probabilistic meteorological models with a coarser resolution and deterministic ones with a high spatial resolution, developing what it is now called a multi-model approach [45-47], which gives more information, since it comes from different meteorological models.
　　However, in this case study, it is not the aim to provide general trends, but to show the results relative to the main cases study investigated during the D-PHASE Project. In fact, generalized conclusions can be only drawn after a large number of cases.
　　Acknowledgments
　　The authors are grateful to ARPA Regione Piemonte(Italy), ARPA Regione Lombardia (Italy), Ufficio Federale dell’Ambiente UFAM Berna (Switzerland) and the Federal Office of Meteorology and Climatology Meteoswiss (Switzerland) for providing the ground measured data. The authors would like to thank also ARPA Regione Emilia-Romagna (Italy) for its contribute in providing COSMO-LEPS model data, ISAC-CNR (Italy) for MOLOCH model data, and EPSON METEO Centre (Italy) for the substantial grant received in these years of Ph.D. research project. This work was partially supported by PRE.G.I. Project and Dote Ricerca Applicata, are funded by Regione Lombardia.
　　References
　　[1] Munich Re Topics Geo, Natural catastrophes 2009: Analysis, assessments and positions [Online], 2010, Münchener Rückversicherungs-Gesellschaft, München, Germany, Order number 302-06735, www.Munichre.com.
　　[2] S.J. Van Andel, R.K. Price, A.H. Lobbrecht, F. van Kruiningen, R. Mureau, Ensemble precipitation and water-level forecasts for anticipatory water-system control, Journal of Hydrometeorology 9 (2008) 776-788.
　　[3] J. Thielen, J. Bartholmes, M.H. Ramos, A. de Roo, The European flood alert system, Part 1: Concept and development, Hydrology and Earth System Sciences 13(2009) 125-140.
　　[4] J.C. Bartholmes, J. Thielen, M.H. Ramos, S. Gentilini, The European flood alert system EFAS-Part 2: Statistical skill assessment of probabilistic and deterministic operational forecasts, Hydrology and Earth System Sciences 13 (2009) 141-153.
　　[5] J. Schaake, K. Franz, A. Bradley, R. Buizza, The hydrologic ensemble prediction experiment (HEPEX), Hydrology and Earth System Sciences 3 (2006) 3321-3332.
　　[6] J. Thielen, J. Schaake, R. Hartman, R. Buizza, Aims, challenges and progress of the hydrological ensemble prediction experiment (HEPEX) following the third HEPEX workshop held in Stresa 27 to 29 June 2007, Atmospheric Science Letters 9 (2) (2008) 29-35.
　　[7] D. Rabuffetti, M. Milelli, The hydro-meteorological chain in Piemonte region, North Western Italy―Analysis of the Hydroptimet test cases, Natural Hazards and Earth System Science 5 (6) (2005) 845-852.
　　[8] A. Amengual, T. Diomede, C. Marsigli, A. Martin, A. Morgillo, R. Romero, et al., A hydrometeorological model
　　 intercomparison as a tool to quantify the forecast uncertainty in a medium size basin, Natural Hazards Earth System Science 8 (2008) 819-838.
　　[9] D. Rabuffetti, G. Ravazzani, C. Corbari, M. Mancini, Verification of operational quantitative discharge forecast(QDF) for a regional warning system-The amphore case studies in the upper Po River, Natural Hazards Earth System Science 8 (2008) 161-163.
　　[10] B. Bacchi, R. Ranzi, Hydrological and meteorological aspects of floods in the Alps: An overview, Hydrology and Earth System Science 7 (2003) 785-798.
　　[11] A. Rossa, G. Haase, C. Keil, P. Alberoni, S. Ballard, J. Bech, et al., Propagation of uncertainty from observing systems into NWP: COST-731 working group 1, Atmospheric Science Letters 11 (2010) 145-152.
　　[12] M. Zappa, K.J. Beven, M. Bruen, A. Cofino, K. Kok, E. Martin, et al., Propagation of uncertainty from observing systems and NWP into hydrological models: COST-731 working group 2, Atmospheric Science Letters 2 (2010) 83-91.
　　[13] R. Ranzi, B. Bacchi, G. Grossi, Runoff measurements and hydrological modelling for the estimation of rainfall volumes in an alpine basin, Quarterly Journal of the Royal Meteorological Society 129B (588) (2003) 653-672.
　　[14] R. Ranzi, M. Zappa, B. Bacchi, Hydrological aspects of the Mesoscale Alpine Programme: Findings from field experiments and simulations, Quarterly Journal of the Royal Meteorological Society 133B (625) (2007) 867-880.
　　[15] M. Zappa, M.W. Rotach, M. Arpagaus, M. Dorninger, C. Hegg, A. Montani, et al., MAP D-PHASE: Real-time demonstration of hydrological ensemble prediction systems, Atmospheric Science Letters 2 (2008) 80-87.
　　[16] M. Arpagaus, M. Rotach, P. Ambrosetti, F. Ament, C. Appenzeller, F. Bouttier, et al., MAP D-PHASE: Demonstrating forecast capabilities for flood events in the Alpine region, Ver?ffentlichungen der MeteoSchweiz 78(2009) 75 [Online], http://www.meteoschweiz.admin.ch/web/de/forschung/pu blikationen/alle_publikationen/veroeff_78.Par.0001.Dow nloadFile.tmp/veroeff78.pdf.
　　[17] M.W. Rotach, P. Ambrosetti, F. Ament, C. Appenzeller, M. Arpagaus, H.S. Bauer, et al., MAP D-PHASE: real-time demonstration of weather forecast quality in the Alpine region, Bulletin of the American Meteorological Society 90 (9) (2009) 1321-1336.
　　[18] R. Ranzi, B. Bacchi, A. Ceppi, M. Cislaghi, U. Ehret, S. Jaun, et al., Real-time demonstration of hydrological ensemble forecasts in MAP D-PHASE, La Houille Blanche 5 (2009) 95-104.
　　[19] J. Thielen, K. Bogner, F. Pappenberger, M. Kalas, M. Del Medico, A. de Roo, Monthly-medium and short-range flood warning: Testing the limits of predictability, Meteorological Applications 16 (1) (2009) 77-90.
　　[20] H.L. Cloke, F. Pappenberger, Operational Flood Forecasting: A Review of Ensemble Approaches, Technical Memorandum 578, European Centre for Medium Range Weather Forecasts, Reading, United Kingdom, 2008.
　　[21] R. Krzysztofowicz, The case for probabilistic forecasting in hydrology, Journal of Hydrology 249 (2001) 2-9.
　　[22] F. Pappenberger, K.J. Beven, N.M. Hunter, P.D. Bates, B.T Gouweleeuw, J. Thielen, et al., Cascading model uncertainty from medium range weather forecasts (10 days) through a rainfall-runoff model to flood inundation predictions within the European flood forecasting system(EFFS), Hydrology and Earth System Science 9 (2005) 381-393.
　　[23] M.H. Ramos, J. Bartholmes, J. Thielen, Development of decision support products based on ensemble weather forecasts in the European flood alert system, Atmospheric Science Letters 8 (2007) 113-119.
　　[24] E. Todini, A model conditional processor to assess predictive uncertainty in flood forecasting, International Journal of River Basin Management 6 (2) (2008) 123-137.
　　[25] Hydrological Atlas of Switzerland (HADES), Swiss Federal Office for the Environment, Berne, Switzerland, 2010.
　　[26] CEC, CORINE Land Cover Technical Guide, European Union, Directorate-General Environment, Nuclear Safety and Civil Protection, Luxemburg, 1993.
　　[27] EEA, CORINE Land Cover Technical Guide-Addendum 2000, European Environment Agency, Copenhagen, Denmark, 2000.
　　[28] Soil Conservation Service, Urban Hydrology for Small Watershed, Tech. Rel. No. 55, U.S. Department of Agriculture, Washington D.C., United States, 1986.
　　[29] C. Marsigli, F. Boccanera, A. Montani, T. Paccagnella, The COSMO-LEPS mesoscale ensemble system: Validation of the methodology and verification, Nonlinear Processes in Geophysics 12 (2005) 527-536.
　　[30] A. Montani, D. Cesari, C. Marsigli, T. Paccagnella, Seven years of activity in the field of mesoscale ensemble forecasting by the COSMO-LEPS system: Main achievements and open challenges, Tellus. 63A (2011) 605-624.
　　[31] P. Malguzzi, G. Grossi, A. Buzzi, R. Ranzi, R. Buizza, The 1966 ‘century’ flood in Italy: A meteorological and hydrological revisitation, J. Geophys. Res., 111 (2006) 15.
　　[32] N. Montaldo, G. Ravazzani, M. Mancini, On the prediction of the Toce alpine basin floods with distributed hydrologic models, Hydrological Processess 21 (2007) 608-621.
　　[33] G. Ravazzani, M. Mancini, I. Giudici, P. Amadio, Effects
　　 of soil moisture parameterization on a real-time flood forecasting system based on rainfall thresholds, in: Quantification and Reduction of Predictive Uncertainty for Sustainable Water Resources Management, IAHS Publication, Vol. 313, 2007, pp. 407-416.
　　[34] G. Ravazzani, D. Rabuffetti, C. Corbari, A. Ceppi, M. Mancini, Testing FEST-WB, a continuous distributed model for operational quantitative discharge forecast in the upper Po river, in: Proceedings of the AMHY-FRIEND International Workshop on Hydrological Extremes, University of Calabria, Cosenza Italy, July 10-12, 2008.
　　[35] C.R. Ehlschlaeger, Using the AT search algorithm to develop hydrologic models from digital elevation data, in: Proc. Int. Geo. Inf. Sys. (IGIS) Symposium, Baltimore, MD, United States, 1989, pp. 275-281.
　　[36] D.R. Montgomery, E. Foufoula-Georgiou, Channel network source representation using digital elevation models, Water Resources Research 29 (12) (1993) 3925-3934.
　　[37] D.G. Tarboton, T.G. Chowdhury, T.H. Jackson, A Spatially Distributed Energy Balance Snowmelt Model, Utah Water Research Laboratory, 1994.
　　[38] C. Corbari, G. Ravazzani, J. Martinelli, M. Mancini, Elevation based correction of snow coverage retrieved from satellite images to improve model calibration, Hydrology and Earth System, Science 13 (2009) 639-649.
　　[39] J. Martinec, A. Rango, Parameter values for snowmelt runoff modelling, Journal of Hydrology 84 (1986) 197-219.
　　[40] V.M. Ponce, Engineering Hydrology, Principles and Practices, Prentice Hall, Englewood Cliffs, New Jersey, Unites States, 1989, pp. 260-261.
　　[41] A. Salandin, D. Rabuffetti, S. Barbero, M. Cordola, G. Volontè, M. Mancini, The Epiglacial lake on the Belvedere Glacier: Monitoring and numerical simulation of the phenomenon aimed at forecasting and managing emergency, Neve e Valanghe 51 (2004) 58-65. (in Italian)
　　[42] N. Montaldo, V. Toninelli, J.D. Albertson, M. Mancini, P.A. Troch, The effect of background hydrometeorological conditions on the sensitivity of evapotranspiration to model parameters: Analysis with measurements from an Italian Alpine catchment, Hydrology and Earth System Science 7 (6) (2003) 848-861.
　　[43] C.H.B. Priestley, R.G. Taylor, On the assessment of surface heat flux and evaporation using large scale parameters, Monthly Weather Review 100 (1972) 81-92.
　　[44] D.S. Wilks, Statistical Methods in the Atmospheric Sciences, 2nd ed., Academic Press, Elsevier, 2006.
　　[45] J. Bartholmes, E. Todini, Coupling meteorological and hydrological models for flood forecasting, Hydrology and Earth System Science 9 (2005) 333-346.
　　[46] S. Davolio, M.M. Miglietta, T. Diomede, C. Marsigli, A. Morgillo, A. Moscatello, A meteo-hydrological prediction system based on a multi-model approach for precipitation forecasting, Natural Hazard and Earth System Science 8(2008) 143-159.
　　[47] T. Diomede, S. Davolio, C. Marsigli, , M.M. Miglietta, A. Moscatello, P. Papetti, et al., Discharge prediction based on multi-model precipitation forecasts, Meteorology and Atmospheric Physics 101 (2008) 245-265.