• Austin, J. F., 1980: The blocking of middle latitude westerly winds by planetary scale waves. Quart. J. Roy. Meteor. Soc, 106 , 327350.

  • Bleck, R., , and Mattocks C. , 1984: A preliminary analysis of the role of potential vorticity in Alpine lee cyclogenesis. Contrib. Atmos. Phys, 57 , 357368.

    • Search Google Scholar
    • Export Citation
  • Buzzi, A., , and Tibaldi A. , 1978: Cyclogenesis in the lee of the Alps: A case study. Quart. J. Roy. Meteor. Soc, 104 , 271287.

  • Buzzi, A., , Tartaglione N. , , and Malguzzi P. , 1998: Numerical simulations of the 1994 Piedmont flood: Role of orography and moist processes. Mon. Wea. Rev, 126 , 23692383.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Doswell C. A. III, , , Ramis C. , , Romero R. , , and Alonso S. , 1998: A diagnostic study of three heavy precipitation episodes in the western Mediterranean region. Wea. Forecasting, 13 , 102124.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Egger, J., 1972: Numerical experiments on the cyclogenesis in the Gulf of Genoa. Contrib. Atmos. Phys, 45 , 320346.

  • Frei, C., , and Schär C. , 2001: Detection probability of trends in rare events: Theory and application to heavy precipitation in the Alpine region. J. Climate, 14 , 15681584.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Godfred-Spenning, C. R., , and Simmonds I. , 1996: An analysis of Antarctic sea-ice and extratropical cyclone associations. Int. J. Climatol, 16 , 13251332.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , McIntyre M. E. , , and Robertson A. W. , 1985: On the use and significance of isentropic potential vorticity maps. Quart. J. Roy. Meteor. Soc, 111 , 877946.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jones, D. A., , and Simmonds I. , 1993: A climatology of Southern Hemisphere extratropical cyclones. Climate Dyn, 9 , 131145.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc, 77 , 437471.

  • Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-year reanalysis: Monthly means CD-ROM and documentation. Bull. Amer. Meteor. Soc, 82 , 247267.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koning, W., , Sausen R. , , and Sielmann F. , 1993: Objective identification of cyclones in GCM simulations. J. Climate, 6 , 217231.

  • Krichak, S. O., , Tsidulko M. , , and Alpert P. , 2000: November 2, 1994, severe storms in the southern Mediterranean. Atmos. Res, 53 , 4562.

  • Lambert, S. J., 1988: A cyclone climatology of the Canadian Climate Centre general circulation model. J. Climate, 1 , 109110.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Leienäs, H., , and Økland H. , 1983: Characteristics of Northern Hemisphere blocking as determined from long time series of observational data. Tellus, 35A , 350362.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Le Treut, H., , and Kalnay E. , 1990: Comparison of observed and simulated cyclone frequency distribution as determined by an objective method. Atmósfera, 3 , 5771.

    • Search Google Scholar
    • Export Citation
  • Llasat, M. C., , and Puigcerver M. , 1994: Meteorological factors associated with floods in the north-eastern part of the Iberian Peninsula. Nat. Hazards, 9 , 8193.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mamassis, N., , and Koutsoyiannis D. , 1996: Influence of atmospheric circulation types on space and time distribution of intense rainfall. J. Geophys. Res, 101 (D21) 2626726276.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Massacand, A. C., , Wernli H. , , and Davies H. C. , 1998: Heavy precipitation on the Alpine southside: An upper-level precursor. Geophys. Res. Lett, 25 , 14351438.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McGinley, J., 1982: A diagnosis of Alpine lee cyclogenesis. Mon. Wea. Rev, 110 , 12711287.

  • Meneguzzo, F., , Pasqui M. , , Messeri G. , , and Rossi M. , 2000: High-resolution simulation of a deep Mediterranean cyclone using RAMS model. Proc. Second EGS Plinius Conf. on Mediterranean Storms, Siena, Italy, EGS, 225–235.

    • Search Google Scholar
    • Export Citation
  • Murray, R. J., , and Simmonds I. , 1991: A numerical scheme for tracking cyclone centres from digital data. Part I: Development and operation of the scheme. Aust. Meteor. Mag, 39 , 155166.

    • Search Google Scholar
    • Export Citation
  • Murray, R. J., , and Simmonds I. , 1995: Responses of climate and cyclones to reductions in Arctic winter sea ice. J. Geophys. Res, 100 , 47914806.

  • Obled, C., , Bontron G. , , and Garçon R. , 2002: Quantitative precipitation forecasts: A statistical adaptation of model outputs through an analogues sorting approach. Atmos. Res, 63 , 303324.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Palutikof, J. P., , Goodess C. M. , , and Guo X. , 1994: Development of daily rainfall scenarios for south-west Spain using a circulation-type approach to downscaling. Int. J. Climatol, 14 , 853896.

    • Search Google Scholar
    • Export Citation
  • Pinto, J. G., , Klawa M. , , Ulbrich U. , , Rudari R. , , and Speth P. , 2001: Extreme precipitation events over north-west Italy and their relationship with tropical–extratropical interactions over the Atlantic. Proc. Third EGS Plinius Conf. on Mediterranean Storms, Baja Sardinia, Italy, EGS, 321–332.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , and Atlas R. , 2001: Tropical cyclone–like vortices in the extratropics: Observational evidence and synoptic analysis. Wea. Forecasting, 16 , 734.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reale, O., , Feudale L. , , and Turato B. , 2001: Evaporative moisture sources during a sequence of floods in the Mediterranean region. Geophys. Res. Lett, 28 , 20852088.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rice, J., 1982: The derivation of a computer-based synoptic climatology of Southern Hemisphere extratropical cyclones. M.S. thesis, Dept. of Meteorology, The University of Melbourne, 113 pp.

    • Search Google Scholar
    • Export Citation
  • Romero, R., , Ramis C. , , Alonso S. , , Doswell C. A. III, , and Stensrud D. J. , 1998: Mesoscale model simulations of three heavy precipitation episodes in the western Mediterranean region. Mon. Wea. Rev, 126 , 18591881.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rudari, R., 2001: European climate predictability and the synoptic control on regional intense rainfall events. Ph.D. thesis, Università degli Studi di Padova, 248 pp.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., , Box J. E. , , Barry R. G. , , and Walsh J. E. , 1993: Characteristics of Arctic synoptic activity, 1952–89. Meteor. Atmos. Phys, 51 , 147164.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stein, O., 2000: The variability of Atlantic–European blocking as derived from long SLP time series. Tellus, 52A , 225236.

  • Tafferner, A., 1990: Lee cyclogenesis resulting from the combined outbreak of cold air and potential vorticity against the Alps. Meteor. Atmos. Phys, 43 , 3147.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tibaldi, S., , and Molteni F. , 1990: On the operational predictability of blocking. Tellus, 42A , 343365.

  • Tibaldi, S., , and Cacciamani C. , 2001: ACCORD: Atmospheric Circulation Classification and Regional Downscaling. Climate Res, 17 , 126165.

    • Search Google Scholar
    • Export Citation
  • Treidl, R. A., , Birch E. C. , , and Sajekcki P. , 1981: Blocking action in the Northern Hemisphere: A climatological study. Atmos.– Ocean, 19 , 123.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) Orographic map with annotations showing the geographic areas of interest. Events with quantiles over 0.95: (b) Seasonal distribution for all the stations in the domain shows autumn as the season that presents the highest percentage of extreme precipitation events over the entire domain. (c) Conditional probability maps of extreme events [Xx′: FX(x′) = 0.95] for a station, baricentric with respect to the Tuscany cluster, in the SON season. (d) Cluster of stations selected for the combined study. Sample size = 36; x′ = 39 mm

  • View in gallery

    (a) Classification of synoptic conditions bringing severe weather in the Tuscany region. The sketches represent characteristic conditions based on the large statistical sample of cases identified in this study. Solid line arrows represent the midtroposphere flow while dashed arrows represent the low-level flow. Arrow's thickness distinguishes relative strength. The relevant orography is schematized in gray shading. (b) Mean condition for the moisture flux, computed as a vertical integral from 1000 to 300 hPa. The mean condition is computed on the basis of the reference time series identified from the precipitation records of the different stations selected on the basis of the conditional probability maps. For an immediate comprehension of the wettest areas, colors represent the module of the vectors schematized by the arrows

  • View in gallery

    Conditional probability maps at (a) lag 0 and (b) lag +1 day for a station in the southern part of Tuscany in the autumn season. The reference station is indicated with a cross. The sample size is 55. The first (conditioning) reference threshold in terms of quantile is 0.92 and corresponds to 31 mm cumulated in 1 day; the second (conditioned) threshold in terms of quantile is 0.82

  • View in gallery

    Climatology for the autumn season (years 1958–86) for (a) 500-hPa geopotential and (b) sea level pressure. Anomaly fields computed as the mean of the time series—selected on the basis of the cluster quantiles dates—minus the climatology for the autumn season (years 1958–86) for (c) 500-hPa geopotential and (d) sea level pressure. Std dev anomaly fields computed as the 4-day std dev of the time series— selected on the basis of the cluster quantiles dates—minus the climatological value for the autumn season (years 1958–86) for (e) 500-hPa geopotential and (f) sea level pressure

  • View in gallery

    Mean condition for the low-level wind. The mean condition is computed on the basis of the reference time series identified from the precipitation records of the different stations selected on the basis of the conditional probability maps

  • View in gallery

    Composite maps for geopotential height at 500 hPa: (a) type a—the sample size used for this composite represents about 65% of the total sample, and (b) type b—the sample size used for this composite represents about 35% of the total sample. The contour lines are labeled in meters. Composite maps for sea level pressure: (c) type a—the sample size used for this composite represents about 65% of the total sample, and (d) type b—the sample size used for this composite represents about 35% of the total sample. The contour lines are labeled in hectopascals. The lows are marked as L and the highs as H

  • View in gallery

    Std dev anomaly fields computed as the 4-day standard deviation of the time series—selected on the basis of the type b dates—minus the climatologic value for the autumn season (years 1958–86) for (a) 500-hPa geopotential and (b) sea level pressure. (c) The low-level wind for the type b dates

  • View in gallery

    Blocking frequency for all the longitudes. (a) The type a distribution (continuous line) is compared with climatology (dotted line). (b) The type b distribution (continuous line) is compared with climatology (dotted line).

  • View in gallery

    Reproduction of the Deutscher Wetterdienst weather charts at 0000 UTC 25 Oct 1980. (a) Geopotential height at 500 hPa. The contour lines are labeled in dekameters. The synoptic trough low is centered over the Scandinavian peninsula. (b) Sea level pressure. The contour lines are labeled in hectopascals. The extended primary low is over the south part of Scandinavia, and its cold front impinges the Alpine barrier from northwest. A lee cyclone is visible in the Gulf of Genoa and the associated frontal system is over Italy. In both maps lows are marked as T and highs as H. The maps show one of the best examples of type a conditions.

  • View in gallery

    Cyclone tracks for the 25 Oct 1980 event. The triangles represent open systems; the circles represent closed systems. Each mark is separated from the following one by a 6-h time interval. The size of the mark represents the strength of the cyclone. The cyclone labeled 1731 is the primary system, while the one labeled 1876 represents the secondary system influencing directly the heavy precipitation associated with the event in the target region. The gray dot is the position of the primary cyclone when the secondary one appears. The black dot represents the position where the secondary cyclone arises and the position of the cyclone most favorable for generating heavy precipitation in the target region

  • View in gallery

    Reproduction of the Deutscher Wetterdienst weather charts at 0000 UTC 29 Oct 1966. (a) Geopotential height at 500 hPa. The contour lines are labeled in dekameters. The map shows a clear cutoff low condition (type b). The cutoff low has moved from the Gulf of Biscay to the Pyrenees while the disturbance that generated the trough is located north of Scandinavia. (b) Sea level pressure. The contour lines are labeled in hectopascals. The map shows the secondary cyclone in the Gulf of Genoa with its associated front. In both maps lows are marked as T and highs as H. The maps show one of the best examples of type b conditions

  • View in gallery

    Cyclone tracks for the 29 Oct 1966 event. The triangles represent open systems; the circles represent closed systems. Each mark is separated from the following one by a 6-h time interval. The size of the mark represents the strength of the cyclone. The cyclone labeled 2158 is the primary system; the one labeled 2333 is the secondary triggered system. The light gray triangle is the position of the primary cyclone when the secondary one appears. The dark gray dot represents the position where the secondary cyclone arises, and the black dot is the position of the cyclone most favorable for generating heavy precipitation in the target region

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 47 47 14
PDF Downloads 26 26 10

Terrain and Multiple-Scale Interactions as Factors in Generating Extreme Precipitation Events

View More View Less
  • 1 Dipartimento di Ingegneria Ambientale, Università degli Studi di Genova, Genoa, and Centro di Ricerca Interuniversitario in Monitoraggio Ambientale, Università degli Studi di Genova e della Basilicata, Savona, Italy, and Massachusetts Institute of Technology, Cambridge, Massachusetts
  • | 2 Massachusetts Institute of Technology, Cambridge, Massachusetts
  • | 3 Dipartimento di Ingegneria Ambientale, Università degli Studi di Genova, Genoa, and Centro di Ricerca Interuniversitario in Monitoraggio Ambientale, Università degli Studi di Genova e della Basilicata, Savona, Italy
© Get Permissions
Full access

Abstract

The Mediterranean region is often affected by flooding and landslides due to heavy precipitation events. These events have been the subject of specific interest because they represent complex interaction of synoptic-scale upper-level steering flows and local topographic barriers. In the present work, data from a dense network of surface precipitation gauges over northern Italy and a global atmospheric analysis at a coarser scale are combined to develop a multiscale diagnostic model of the phenomenon.

Composite maps are formed based on departures from climatology and standard deviation of sea level pressure, 500-hPa geopotential, wind, and water vapor flux. A diagnostic model is built based on the evidence that shows the spawning of secondary mesoscale features in the steering synoptic flow. The mesoscale features draw moisture and energy from local sources and cause extreme precipitation events over adjoining areas.

The primary trough system steering the flow often originates in the North Sea and extends over middle Europe. When this system flows across the Alps, a secondary cyclone develops in the western Mediterranean, more frequently over the Gulf of Genoa on the lee side of the Alps barrier. The mesoscale cyclone is evident in the weather charts, and its signature is identifiable in the dense surface gauge network data, but it is not evident in the atmospheric analyses maps owing to their coarse resolution.

Further analysis of the ensemble of such cases is made by manual inspection of daily all-Europe weather charts. Two precursor conditions for this mesoscale cyclogenesis are identified. It is shown that longitudinal blocking frequency over a larger region strongly differs from climatology (especially for the second one of these precursors). A low pressure center tracking algorithm is used to follow the evolution of some events. Two cases are presented as illustrative of the patterns identified by the ensemble composites.

Corresponding author address: Dr. Roberto Rudari, Centro di Ricerca Interuniversitaro in Monitoraggio Ambientale, Università degli Studi di Genova, Via Cadorna 7, Savona (SV) 17100, Italy. Email: rr@cima.unige.it

Abstract

The Mediterranean region is often affected by flooding and landslides due to heavy precipitation events. These events have been the subject of specific interest because they represent complex interaction of synoptic-scale upper-level steering flows and local topographic barriers. In the present work, data from a dense network of surface precipitation gauges over northern Italy and a global atmospheric analysis at a coarser scale are combined to develop a multiscale diagnostic model of the phenomenon.

Composite maps are formed based on departures from climatology and standard deviation of sea level pressure, 500-hPa geopotential, wind, and water vapor flux. A diagnostic model is built based on the evidence that shows the spawning of secondary mesoscale features in the steering synoptic flow. The mesoscale features draw moisture and energy from local sources and cause extreme precipitation events over adjoining areas.

The primary trough system steering the flow often originates in the North Sea and extends over middle Europe. When this system flows across the Alps, a secondary cyclone develops in the western Mediterranean, more frequently over the Gulf of Genoa on the lee side of the Alps barrier. The mesoscale cyclone is evident in the weather charts, and its signature is identifiable in the dense surface gauge network data, but it is not evident in the atmospheric analyses maps owing to their coarse resolution.

Further analysis of the ensemble of such cases is made by manual inspection of daily all-Europe weather charts. Two precursor conditions for this mesoscale cyclogenesis are identified. It is shown that longitudinal blocking frequency over a larger region strongly differs from climatology (especially for the second one of these precursors). A low pressure center tracking algorithm is used to follow the evolution of some events. Two cases are presented as illustrative of the patterns identified by the ensemble composites.

Corresponding author address: Dr. Roberto Rudari, Centro di Ricerca Interuniversitaro in Monitoraggio Ambientale, Università degli Studi di Genova, Via Cadorna 7, Savona (SV) 17100, Italy. Email: rr@cima.unige.it

1. Introduction

Identification of the precursor and large-scale synoptic conditions that result in extreme precipitation events in the Mediterranean has been the subject of specific interest in the last few years (Doswell et al. 1998; Frei and Schär 2001; Krichak et al. 2000; Llasat and Puigcerver 1994; Pauliticof et al. 1994; Romero et al. 1998). The reason for this interest is twofold: 1) the region is devastated by costly flooding and landslide hazards that need to be forecast with better skill, and 2) the diagnosis of the complex interaction of steering atmospheric flows and topographic barriers, that is, the essence of this hydrometeorological phenomenon, will advance the understanding of multiscale atmospheric events.

The literature to date has mainly focused on case studies of case events. Each study has produced a different perspective on similar synoptic and mesoscale mechanisms that are thought to be responsible for extreme precipitation events over the western Mediterranean and northern Italy specifically (Buzzi et al. 1998; Massacand et al. 1998; Pinto et al. 2001; Reale et al. 2001). Few attempts have been made in this geographic area to link a larger population of extreme events to the principal synoptic feature in a rigorous statistical framework. Recent investigations in this geographic area mainly concern climatic links among surface patterns and large-scale weather regimes (Tibaldi and Cacciamani 2001). Some attempts to address the extreme events at the weather scale follow empirical techniques based on analogies (Obled et al. 2002).

This study introduces a basic method that combines a dense network of surface precipitation stations and coarse-grid atmospheric reanalysis data on atmospheric state variables to identify and analyze a large sample of characteristic events. The data from the dense surface gauge precipitation network and coarse-grid atmospheric analyses are combined to produce composite maps that discriminate the distinct synoptic patterns inducing localized extreme precipitation. Even though the local areas experiencing heavy precipitation may be in close proximity, they are characterized by unique and different patterns of synoptic weather (Rudari 2001). Local topography is demonstrated to be a key factor in the development and evolution of the hydrometeorological features.

In this work a portion of Tuscany, Italy (area ∼ 1000 km2), is selected through a statistical approach, and the heavy precipitation events over this region are related to the large-scale flow fields in order to improve the understanding of these phenomena. The proposed approach is capable of identifying a large statistical sample for the study and adds to the previous understanding that has been built mostly on individual case studies. The large-scale fields need to go as far back as the North Sea to be fully described. A primary and secondary system appears in the ensemble of 172 cases with heavy precipitation over the local area. The upper-level steering flow is characterized by a low that drifts downstream across Europe but spawns a secondary cyclone in the lee of the western Mediterranean topographic barriers. The secondary system is mesoscale and feeds low-level moisture and energy derived from the western Mediterranean Sea to the local area.

The 172 ensemble of cases are verified by detailed inspection of daily weather charts. Of these, 152 independent events are identified that serve to further distinguish the lee cyclogenesis phenomenon into two categories. Composite maps for these categories are also presented. To further demonstrate the multiscale atmospheric phenomena, some specific cases are analyzed using a low pressure center tracking algorithm. Two of the study cases are presented to demonstrate the formation of lee cyclones as suggested by the composites.

2. Synoptic regimes associated with heavy rainfall in Tuscany

Precipitation gauge station records from different offices of the Italian National Hydrological Service (Servizio Idrografico e Mareografico Nazionale) and from other minor sources, covering most of northwestern Italy, are used as a surface database. Additional quality-control measures are applied to the daily records. Range and spatial consistency tests were performed to flag and remove extreme daily accumulations that were not spatially coherent with respect to neighboring stations. The quality-control procedure selected 447 stations (out of 1563 original) that matched the requirements for this study of extreme precipitation events in northwestern Italy with 1958–86 as the observation period. The data coverage over the domain is exceptional (typical station spacing is 10 km), although some areas of the western and central Po Valley exhibit a lower station density.

Traditionally, extreme precipitation events were identified using rainfall thresholds over an extended area where rain gauge measurements were available (e.g., Mamassis and Koutsoyiannis 1996). In a complex orography environment, such as the one under study, differences in elevation of 200 m can be found in rain gauges that are only a few kilometers apart (Fig. 1a shows the orography of the analyzed domain with annotation of the geographic areas of interest). This can lead to dramatic changes in rainfall distribution due to the interaction of topography and atmospheric flows as well as the varying location exposures to the dominant moisture flux. It is therefore necessary to define the extreme events at each station based on exceedence probability instead of total daily accumulation. Days with extreme rainfall at each station are defined as days with accumulations exceeding the 95th percentile of the station record value.

There is significant seasonality in the region. The choice of season is therefore the first concern. From the analysis of the surface data the autumn season [September–November (SON)] shows the highest frequency of extremes when the 95th percentile for each station is based on the entire station record. The chart in Fig. 1b shows the percentage of extremes occurring in each season for the Tuscany group of stations. SON contains almost 50% of the extremes. The three remaining seasons share almost the same number of events, with winter slightly more populated than summer and spring. Moreover, historically many of the devastating floods in Tuscany (e.g., the great flood of Florence, 4–6 November 1966) have occurred in autumn.

In this study the conditional probability of exceeding given thresholds at the stations (on the same day) is used to develop a group of stations showing similar conditional probability patterns. The computed quantity is essentially Prob(Xn > xn|xm > Xm > xm), where Xn is the event quantile of the nth station; Xm is the event quantile of the mth station with nm; xn, xm, and xm are the event quantile thresholds on the basis of which the probability maps are computed. The quantile thresholds were both set to 0.95, and a drop of the conditional probability value to 0.4 is chosen as threshold value for the station selection.

This grouping of stations allows developing a large sample of dates for each localized region when extreme rainfall has occurred in any member station. There are no overlaps in dates for each grouping. Furthermore the method is nonparametric and is an effective way of isolating local hydrometeorological extreme events (Rudari 2001).

In Fig. 1c the autumn conditional probability map that forms the basis for the definition of the Tuscany station group is presented. The index station that forms the conditional state for the probability values at each station is marked by a cross. Spatial clustering is evident. The coherence of the stations that qualify for inclusion in the Tuscany group with topography and orographic factors is shown in Fig. 1d. There are 36 stations in the Tuscany group. The number of stations belonging to the group allows the setting of a higher threshold for the extreme definition (0.97), maintaining a good sample size of extremes. The result is a large sample with 171 dates, and at least in one of the stations the 0.97 quantile is exceeded.

Much larger-scale synoptic patterns (covering the northern Atlantic, Europe, and western Asia— from 20° to 70°N and from 30°W to 60°E) associated with extreme rainfall in the region of interest are identified by forming composites of atmospheric 500-hPa geopotential, winds, pressure, and vapor flux. The atmospheric fields are drawn from the coarse-scale (2.5° latitude and longitude spacing) data from the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) reanalysis (Kalnay et al. 1996; Kistler et al. 2001). Each composite is the conditional mean of the atmospheric variable—the conditioning state is the set of dates with extreme surface rainfall for the selected stations. The composite fields are represented as departures from the climatology of the autumn season for the 29-yr record.

Figure 2b shows the composite map for the column-integrated moisture flux. A weak stream of moisture flux along the baroclinic trough starting from the Atlantic is present, although the main moisture transport is localized in the Mediterranean. The direction of the vectors in proximity of the Italian coast points toward the target region, but the highest values of moisture flux are shifted southward with respect to Tuscany. The synoptic-scale features are not sufficient to exhaustively explain the moisture sources for the localized region extreme rainfall events. The weak large-scale moisture transport, the more-southern-than-expected position of the moisture flux, and its derivative maxima suggests that mesoscale mechanisms embedded in the large-scale flow may be important factors. However, such mesoscale mechanisms cannot be identified with the coarse-resolution NCEP–NCAR data.

The evidence for the presence of the mesoscale feature may be diagnosed in the dense network of surface precipitation station data. Another set of conditional probability station maps are produced in this study that represent the probability of rainfall percentile exceedance (Fig. 3) at the stations on the day following an extreme event in the index station (marked as a cross) in the center of the Tuscany group. The pattern of time lag evident in Fig. 3 shows a residual correlation in the stations located northwest of the station taken as reference. This generally indicates a translation of the hydrometeorological event over the period. This pattern is consistent with all the others produced by the stations belonging to the group. The daily temporal resolution of the rainfall data does not allow a direct interpretation of these patterns with regard to the movement of the precipitating clusters.

The minimum time interval needed to track such precipitating systems is several hours. It is nonetheless possible to draw important indirect information from these maps. The lagged probability maps show the contribution of all the events that appear to move northwest creating the extreme rainfall cumulates in those areas over a 24-h period.

The direction indicated by the map is consistent with a mesoscale cyclonic circulation. Case studies in the literature suggest that this type of mesoscale circulation feeds moisture from local Mediterranean Sea and farther upstream sources to the Tuscany region (e.g., Meneguzzo et al. 2000).

Other important fields for the composite study are presented as anomalies in Figs. 4c–f. Figures 4a and 4b show the climatology for the 500-hPa geopotential and sea level pressure (SLP) for the SON season in the period from 1958 to 1986. Figure 4c shows the 500-hPa geopotential field anomaly. A systematic pattern, distinct from the season climatology, is clearly evident on days when extreme rainfall (greater than 0.97-percentile accumulation) events occur over Tuscany. A deep negative anomaly centered over France extends eastward to the Gulf of Genoa. The sea level pressure anomaly shown in Fig. 4d has the negative anomaly centered southeast of the 500-hPa geopotential anomaly and extends over the Gulf of Genoa with similar or enhanced intensity. This extension in the lower levels of the atmosphere may be linked to increased cyclogenesis over the anomaly areas.

Figure 4e and 4f maps show the standard deviation anomalies of the 500-hPa geopotential and sea level pressure fields. The standard deviation anomaly fields are computed as the average standard deviation of the 4 days preceding the cluster quantile dates including the day of the event minus the climatological standard deviation of the same quantity for the autumn season (years 1958–86).

Although the mean composite fields show a strong signal, the standard deviation fields on days with extreme events over Tuscany depart from the climatology of the SON season less clearly. This could imply that the synoptic steering mechanism is present but the real triggering factors for the extreme events are mesoscale in nature. They cannot be identified with data at the resolution of the NCEP–NCAR reanalysis.

Nevertheless, both the 500-hPa geopotential and the sea level pressure standard deviation anomaly fields highlight the basic features of the synoptic conditions. The 500-hPa geopotential standard deviation anomaly field (Fig. 4e) shows a higher-than-normal activity centered on the Mediterranean area. The sea level pressure standard deviation anomaly field (Fig. 4f) identifies two areas of stronger activity: one localized in the North Sea and the other in the Gulf of Genoa.

The patterns evident in Fig. 4 show the signature of lee cyclogenesis in the western Mediterranean area. This phenomenon has been identified as an important factor in the hydrometeorology of the region (Buzzi and Tibaldi 1978). Numerous case studies of specific extreme events have been diagnosed to be associated with western Mediterranean lee cyclogenesis. In this study, the methodology of Rudari (2001) is applied to study this important synoptic mode for a large sample. The large sample allows the further discrimination of the factors that influence this synoptic mode.

The diagnostics presented so far are summarized in the schematic in Fig. 2a. The conceptual model shows the mean flow trajectories obtained by the composite analysis and the position of the 500-hPa jet, the low-level flow conditions, and the orographic barriers that play an active role leveling the hydrometeorology of extreme mesoscale events in the region. A strong 500-hPa steering jet is commonly present when extreme events occur over Tuscany (see Fig. 4c). The high-level jet enters the Mediterranean area from the north-northwest direction and becomes essentially zonal close to the target region. The baroclinic trough position is not anomalous but the geopotential values are lower than the average. The low-level flow enters the Mediterranean with a direction that resembles the one followed by the high-level winds. Near the target region it impinges the western flank of the Alps. As a result the low-level flow cold advection is retarded with respect to the high levels and the flow itself is deflected, creating flow-around or flow-over conditions, or both. The flow then bends sharply and becomes meridional near Tuscany.

These large-scale conditions are typical preconditions for what in literature is known as cyclogenesis in the lee of the Alps (Buzzi and Tibaldi 1978). The composite low-level wind conditioned on heavy rainfall in the Tuscany region (Fig. 5) shows a rapid change in low-level wind direction that possibly signifies the presence of a mesoscale cyclonic feature over the Gulf of Genoa.

The synoptic setting of Alpine lee cyclogenesis is characterized by the progression of an upper-level trough and its associated cold front toward the Alps (Buzzi and Tibaldi 1978; Bleck and Mattocks 1984). Lee cyclogenesis takes place in two phases (Egger 1972; Buzzi and Tibaldi 1978; McGinley 1982). The first phase is frontal retardation, a cold-air outbreak into the western Mediterranean, and the rapid formation of a shallow vortex over the Gulf of Genoa. During the second phase the growth rate drops to baroclinic values, and the structure of the growing cyclone approaches that of typical extratropical low pressure systems. The primary effect of the mountain is thus the first-phase formation of a low-level vortex (or amplification of a preexisting vortex), while the second phase can be interpreted as a mere modification of classical cyclogenesis (Hoskins et al. 1985; Tafferner 1990).

Any low-level vortex structure able to contribute to the second phase must be associated with a surface thermal anomaly or a low-level potential vorticity anomaly (Buzzi et al. 1998). A well-documented mechanism for the generation of surface thermal anomaly events is the retardation of an approaching cold front. This results in the generation of an Alpine-scale wake characterized by a warm surface anomaly. This anomaly constitutes by itself the driving agent of the second phase, although the orographic retardation also implies other effects that contribute to the lee cyclogenetic process (McGinley 1982; Bleck and Mattocks 1984; Tafferner 1990).

The coarse resolution of the NCEP–NCAR reanalysis data does not allow the diagnosis of the development of lee cyclones. Augmentation with the dense station network and combined composite maps, as done in this study, are required to identify the conditions associated with Alpine lee cyclogenesis for a large sample (not a single case) of events. In order to confirm the patterns identified from the sample days with extreme events over Tuscany, the conditions during each of the days are studied independently in order to confirm the development of Alpine lee cyclogenesis.

3. Identification of cases

A more detailed look at the individual cases that together constitute the sample of dates with extreme rainfall events over Tuscany supports the role of Alpine lee cyclogenesis for extreme events over the region. European-region meteorological analysis charts complied by Deutscher Wetterdienst (German Meteorological Office) and archived on paper at the location are used for the analysis.

The 171 dates with extreme rainfall contain 152 independent events. The remainder are too close in date (i.e., less than 3 days) to constitute independent events. The meteorological charts for each of these dates offer a more detailed description of the fields. Because they are analysis charts based on the dense network of stations across Europe, their resolution (distinct from posting) is considerably finer when compared to the NCEP– NCAR reanalysis data that is posted on 2.5° latitude and longitude.

The dates when extreme rainfall events occurred over Tuscany (together with the preceding and following dates) are manually screened in this study. A summary of the analyses is given as follows:

  • More than 85% of the events clearly present cyclogenesis in the western Mediterranean as evident in secondary systems developing out of the main steering primary system after it interacts with the distinctive morphology of the Mediterranean basin;
  • The primary system interacts with the western Mediterranean morphology following two principal synoptic types, and just an insignificant number (less than 20) of events that produced severe weather in the area depart from these two types.

On the basis of the manual screening of the European weather charts from Deutscher Wetter Dienst, a classification of the 152 events into types a and b are made. The main discriminating features for the classification are the path followed by the primary system and the position of this primary cyclone when the secondary cyclone appears.

In type a the primary cyclone originates west of the British Isles and then moves toward the north of Europe following a west–east direction, reaching the North Sea or the Scandinavian peninsula. Then the system often moves northeast, sometimes reaching the polar circle. When the cyclone is positioned near the Scandinavian peninsula the secondary system appears in the Gulf of Lyon or, more often, in the Gulf of Genoa.

In type b the primary cyclone forms east of the coast of Iceland and then moves toward Europe following a northerly direction. The system reaches central Europe through the North Sea, moving eventually south to France, north of Spain, and, more frequently, to the Gulf of Biscay. The primary system fades when it reaches one of these destinations, and a secondary cyclone forms in the western Mediterranean.

The two types show consistent synoptic features, apart from the primary cyclone track, which are reported below. Type a is the dominant type, accounting for more than 60% of the cases. The synoptic conditions associated with type a do not differ distinctly from the climatology of the autumn season, especially when the 500-hPa geopotential fields are considered. The mean 500-hPa geopotential and sea level pressure composite maps for the dates when type a synoptic regime has been identified are given in Figs. 6a and 6c, respectively. The upper-air condition resembles the synoptic condition described in the conceptual model sketched in Fig. 2a. A weak trough is centered north of the British Isles. The upper-level air masses enter into the Mediterranean from north-northwest over the Iberian Peninsula. The geopotential field bends and become almost zonal over Italy. This upper-air structure favors the development of the surface low positioned over the North Sea to the west of the Scandinavian peninsula. This primary system steers a cold front that crosses Europe and enters the Mediterranean.

The result of the frontal system interaction with the complex orography of the western Mediterranean basin is a localized cyclogenesis that varies its position with respect to the frontal system location. Figure 6c shows a gradient in the sea level pressure field from the Balearic Island to the Gulf of Genoa. This feature may be caused by these small-scale secondary systems. The western high is localized in front of the Moroccan Atlantic coast while the eastern high is placed over the Caspian Sea.

The type b cases, however, present clear distinction from the autumn climatology. Type b composite maps for 500-hPa geopotential and the sea level pressure fields are presented in Figures 6b and 5d, respectively. The upper-air condition noticeably differs from the one identified for type a. The synoptic trough has its axis in the southwest–northeast direction elongating from Scandinavia over central Europe. The positive geopotential anomaly is stronger over the British Isles and forces the trough over Europe, where a cutoff low becomes evident. This cutoff system, on average present over France, is usually triggered and sustained by a cyclonic anomaly at the tropopause, as evident from the 200-hPa geopotential fields (not shown). The sea level pressure field presents a marked low in the Mediterranean, a sign of the intense cyclogenetic activity in this area. The western high is localized in the region of Portugal, and the eastern high is placed north of the Caspian Sea. Their value is higher in comparison with the ones shown for type a, and they are considerably more extended.

The strength of the cyclogenesis signal is such that even the standard deviation field shows noticeable differences with respect to climatology. Specifically, the type b standard deviation anomaly of both the 500-hPa geopotential and the sea level pressure fields show a structure that puts in evidence the cutoff low conditions over the Gulf of Genoa, in the lee of the Alps (Fig. 7). Figure 7c, showing the surface winds for type b, presents an evident cyclonic flow in support of this argument.

It is important to test whether or not there is evidence for linking the occurrence of these weather patterns (types a and b) to atmospheric blocking conditions. Evidence of blocking cannot be discerned (Figs. 6a and 6c) because composite averaging obscures any pattern that may have variations in its center location in every occurrence. They can, however, be detected if each single ensemble member is analyzed.

Atmospheric blocking has long been recognized as a physical process with profound effects on local weather and its extremes (Tibaldi and Molteni 1990). In fact, blocking across Europe has been related to heavy precipitation in the region of interest for this study (Stein 2000). In this study, blocking diagnostic measures are used to further distinguish between the two identified hydrometeorological patterns.

There are several different definitions of blocking measure. Austin (1980) and Triedl et al. (1981) show that a typical European blocking pattern is characterized by a high pressure area at 60°N and a low at 40°N in the midtroposphere. Using this relation, Lejenäs and Økland (1983) applied geopotential height difference between 40° and 60°N to create a longitudinal blocking index.

The definition of the blocking measure used in this study is the one originally established by Lejenäs and Økland (1983) and later modified by Tibaldi and Molteni (1990). The geopotential height gradient (GHG) is computed for each longitude point of the NCEP–NCAR reanalysis grid as
i1525-7541-5-3-390-eq1
where
i1525-7541-5-3-390-eq2
with Δ = −5°, 0°, and +5°. A given longitude is then defined as “blocked” at a specific time if the condition GHG > 0 is satisfied for at least one value of Δ.

The frequency of this blocking measure is estimated at each longitude location. These frequencies are computed both for all days in SON seasons for the period considered in the study (climatology) and for the composites corresponding to days with types a and b patterns. The results are shown in Fig. 8.

The frequency distributions of both categories differ significantly from the autumn-season climatology that has a dampen distribution with a small peak on 10°E longitude. For the composite maps of the two categories the blocking measure follows the negative anomaly position of geopotential height at 500 hPa shown in Figs. 6a and 6b.

As expected, type b is associated more often with block conditions when compared to type a. The percentage of events showing at least one blocked longitude is around 40% for type a and 74% for type b, confirming a substantial difference between the two synoptic patterns. In general, 47% of the events causing heavy precipitation over Tuscany show at least one blocked longitude. Therefore, blocking does not represent a necessary condition for extreme precipitation over the target area, but the frequency distributions of blocked longitudes show significant differences from the autumn climatology.

4. Cyclone tracking

To confirm that the steering type identified by the composite analyses is indeed a primary factor that triggers secondary features after it interacts with the significant topographic barriers of the region, a cyclone-tracking algorithm is applied to analyze the time evolution of over 30 study cases. In the following sections specific cases that describe the two identified types are presented.

A cyclone-tracking algorithm is used to study the time evolution of several events that contribute to the composites. Several schemes have been proposed to locate and track pressure lows in gridded data (Rice 1982; Lambert 1988; Le Treut and Kalnay 1990; Koning et al. 1993; Serreze et al. 1993). In this study the tracking scheme by Murray and Simmonds (1991) is adopted. This scheme is distinct from the others cited. It searches for lows on a spline-fitted surface built from the sea level pressure fields. The advantage is that it allows both closed and open depression contours to be tracked (Jones and Simmonds 1993; Murray and Simmonds 1995; Godfred-Spenning and Simmonds 1996).

The Murray and Simmonds (1991) scheme starts with transforming data posted on a latitude–longitude grid into a polar stereographic array using bicubic spline interpolation. The low is found by searching this array for a gridpoint maximum of the Laplacian of pressure, ∇2p. A maximum of ∇2p identifies a maximum of relative cyclonic geostrophic vorticity ξ ≈ (1/ρf)∇2p, where ρ is the air density and f is the Coriolis parameter. The position is then matched, by iterative approximation, to the center of an ellipsoid in the array data. If a closed center cannot be found or does not lie near, the routine searches for an open depression. The Laplacian of the pressure in the vicinity of the center is taken to be a measure of strength of the system. Systems that fail to reach a specified minimum strength are excluded.

Tracking is accomplished in a three-stage process in which 1) a subsequent position is predicted for each cyclone based on a weighted mean of the previous displacement and of the displacement based on mean cyclone velocities, 2) a probability of an identification between the projected cyclone and each cyclone present at the new time is reconciled, and 3) a matching is made that maximizes the calculated probability of association between the projection and new positions.

Two particular cases are selected out of the 152 independent events to illustrate type a and b conditions. These are meant to illustrate typical evolution of events for each mechanism. Whereas the composites are statistical maps and not realizations of atmospheric states, these case studies serve to show that the types identified in the study are not statistical artifacts, and indeed such cases are replicates of samples whose statistical characteristics are contained in the composites.

a. The 25 October 1980 event

This event represents a clear example of type a events. The 500-hPa geopotential field (Fig. 9a) resembles the composite map pattern with a trough showing a cutoff structure over Scandinavia. The 500-hPa contour lines enter the Mediterranean area from north-northwest over the Pyrenees and then bend into a zonal formation over Italy. The associated low extends from the North Sea to the southern part of the Scandinavian peninsula (Fig. 9b), and its frontal system impinges the Alps from the west. These conditions trigger a lee cyclone in the Gulf of Genoa, which brings a cold front against the Italian coast.

The tracking algorithm shows that the primary system starts west of Ireland at the longitude of Iceland (50°N, 20°W) and then moves toward the British Isles. Each mark (Fig. 10) is separated from the following one by a 6-h time interval. The system crosses the North Sea and finally dissipates near the pole past western Siberia. The cyclone first appears on 21 October and passes across the North Sea on 24 October. On 25 October its position is south of Scandinavia (gray dot in Fig. 10). At this time (0000 UTC 25 October) an associated lee cyclone appears in the Gulf of Genoa (black dot in Fig. 10). This secondary system steers a cold front over Italy. The event produces heavy precipitation in Tuscany. The system then migrates south and dissipates on 28 October near the African coast.

b. The 29 October 1966 event

The 29 October event is a typical cutoff low type b event. It demonstrates interaction between a primary or steering system and a secondary lee cyclone system. The evolution of the system is complex and begins on 25 October. Figure 11 shows the final part of the evolution in both the 500-hPa geopotential and sea level pressure fields. The complete tracking of the system is summarized in Fig. 12. Each mark in Fig. 12 is separated from the following one by a 6-h time interval.

The system begins as a trough moving from Iceland toward the Scandinavian peninsula. An upper-air cyclonic anomaly starts early on 25 October and moves south toward central Europe. This low geopotential anomaly is then squeezed by a deep intrusion of high geopotential localized west of the British Isles. This ridge pushes the low even farther south and helps the development of a cutoff low that centers over the Pyrenees on 28–29 October. At the surface a cyclone moves from the North Sea into the European region. The weather charts (Fig. 11) show an open depression near the Gulf of Biscay. A secondary cyclone develops on the lee side of the Alps and the front moves across Italy during 28–29 October. The secondary system and associated front are clearly evident in Fig. 11b.

The primary system appears over the eastern coasts of Greenland at 1200 UTC 24 October and rapidly moves into the North Sea on 26 October. In the following days the system is forced into Europe where it slowly weakens. At 0600 UTC on 28 October, the cyclone has developed into an open system near the Gulf of Biscay and it remains there until 1800 UTC of the same day. During the same period a secondary cyclone appears in the lee side of the Alps and deepens until the end of 29 October. Throughout this period it remains confined to the Gulf of Genoa (see Fig. 12) and feeds the extreme precipitation event over Tuscany.

5. Conclusions

Large-scale synoptic conditions that dominate during heavy precipitation events in a flood- and landslide-prone region in northern Italy are diagnosed by the combined use of a dense surface precipitation gauge network and a coarse-scale atmospheric analysis dataset. The key advantages of the approach are that 1) it identifies an ensemble of cases (numbering over 150 here) that represent characteristic conditions, 2) it is nonparametric in that the conditional probability and conditional average (i.e., composite) do not assume any distributions and are unbiased, and 3) through analysis of an ensemble of events, the generic preconditions and conditions for extreme events are identified and may guide operational hazardous weather watches.

The composite maps diagnose the favorable conditions for heavy precipitation in the southern part of Tuscany. Based on the composite maps of low-level flow (sea level pressure fields, wind), upper-level steering flow (500-hPa geopotential), and water vapor flux, a conceptual model of the large-scale meteorological conditions that often lead to extreme events in Tuscany is introduced. Mean departure and standard deviation maps are used to address the uniqueness of the synoptic patterns with respect to seasonal climatology.

The results indicate that a steering synoptic flow is necessary and that the local extreme events themselves are maintained by related local features that originate from within the large-scale flow. Furthermore, the synoptic conditions that lead to extreme events over the target area are shown to be characterized by distinct blocking patterns that extend well downstream. Atmospheric blocking is often considered an important factor in the eventual generation of extreme precipitation. It is shown to be a contributing factor in this study as well but it cannot be declared a necessary condition for extreme precipitation in Tuscany.

The steering flow is identifiable as far away as the North Sea and extends over central Europe. When this system interacts with the considerable topography of the region, a cyclonic system develops more frequently over the Gulf of Genoa on the lee side of the Alps barrier. The mesoscale cyclone is evident in the lagged precipitation probability map based on the dense surface gauge network, but it is not visible in the atmospheric analyses maps owing to their coarse resolution. The lee cyclone draws moisture and energy from the adjacent western Mediterranean Sea and forces a low-level moist airflow that enters the target region in a zonal flow pattern (Reale and Atlas 2001).

Last, a study of 171 European daily weather analyses is performed to demonstrate that the cases contributing to the composite maps do indeed follow the general conceptual model. In the study of the weather maps, 152 independent cases of the phenomenon are identified. The cases are separated into two sets that better discriminate the finding by composite analyses. These cases confirm that western Mediterranean cyclogenesis due to orographic interference is a fundamental condition for heavy rainfall in Tuscany; they also provide a more clear insight into the triggers of these secondary cyclones.

Acknowledgments

The authors acknowledge the important help of Dr. Pinto in the development of this work and of the Mitteilungen aus dem Institut fuer Geophysik und Meteorologie der Universitaet zu Koeln for making available its resources to the first author. The research was partially supported by the Italian National Research Council through GNDCI, National Group for Protection from Hydrologic Hazards. This work is part of the research carried out by the first author in partial fulfillment of his Ph.D. program requirements.

REFERENCES

  • Austin, J. F., 1980: The blocking of middle latitude westerly winds by planetary scale waves. Quart. J. Roy. Meteor. Soc, 106 , 327350.

  • Bleck, R., , and Mattocks C. , 1984: A preliminary analysis of the role of potential vorticity in Alpine lee cyclogenesis. Contrib. Atmos. Phys, 57 , 357368.

    • Search Google Scholar
    • Export Citation
  • Buzzi, A., , and Tibaldi A. , 1978: Cyclogenesis in the lee of the Alps: A case study. Quart. J. Roy. Meteor. Soc, 104 , 271287.

  • Buzzi, A., , Tartaglione N. , , and Malguzzi P. , 1998: Numerical simulations of the 1994 Piedmont flood: Role of orography and moist processes. Mon. Wea. Rev, 126 , 23692383.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Doswell C. A. III, , , Ramis C. , , Romero R. , , and Alonso S. , 1998: A diagnostic study of three heavy precipitation episodes in the western Mediterranean region. Wea. Forecasting, 13 , 102124.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Egger, J., 1972: Numerical experiments on the cyclogenesis in the Gulf of Genoa. Contrib. Atmos. Phys, 45 , 320346.

  • Frei, C., , and Schär C. , 2001: Detection probability of trends in rare events: Theory and application to heavy precipitation in the Alpine region. J. Climate, 14 , 15681584.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Godfred-Spenning, C. R., , and Simmonds I. , 1996: An analysis of Antarctic sea-ice and extratropical cyclone associations. Int. J. Climatol, 16 , 13251332.

    • Search Google Scholar
    • Export Citation
  • Hoskins, B. J., , McIntyre M. E. , , and Robertson A. W. , 1985: On the use and significance of isentropic potential vorticity maps. Quart. J. Roy. Meteor. Soc, 111 , 877946.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jones, D. A., , and Simmonds I. , 1993: A climatology of Southern Hemisphere extratropical cyclones. Climate Dyn, 9 , 131145.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc, 77 , 437471.

  • Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-year reanalysis: Monthly means CD-ROM and documentation. Bull. Amer. Meteor. Soc, 82 , 247267.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Koning, W., , Sausen R. , , and Sielmann F. , 1993: Objective identification of cyclones in GCM simulations. J. Climate, 6 , 217231.

  • Krichak, S. O., , Tsidulko M. , , and Alpert P. , 2000: November 2, 1994, severe storms in the southern Mediterranean. Atmos. Res, 53 , 4562.

  • Lambert, S. J., 1988: A cyclone climatology of the Canadian Climate Centre general circulation model. J. Climate, 1 , 109110.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Leienäs, H., , and Økland H. , 1983: Characteristics of Northern Hemisphere blocking as determined from long time series of observational data. Tellus, 35A , 350362.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Le Treut, H., , and Kalnay E. , 1990: Comparison of observed and simulated cyclone frequency distribution as determined by an objective method. Atmósfera, 3 , 5771.

    • Search Google Scholar
    • Export Citation
  • Llasat, M. C., , and Puigcerver M. , 1994: Meteorological factors associated with floods in the north-eastern part of the Iberian Peninsula. Nat. Hazards, 9 , 8193.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mamassis, N., , and Koutsoyiannis D. , 1996: Influence of atmospheric circulation types on space and time distribution of intense rainfall. J. Geophys. Res, 101 (D21) 2626726276.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Massacand, A. C., , Wernli H. , , and Davies H. C. , 1998: Heavy precipitation on the Alpine southside: An upper-level precursor. Geophys. Res. Lett, 25 , 14351438.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McGinley, J., 1982: A diagnosis of Alpine lee cyclogenesis. Mon. Wea. Rev, 110 , 12711287.

  • Meneguzzo, F., , Pasqui M. , , Messeri G. , , and Rossi M. , 2000: High-resolution simulation of a deep Mediterranean cyclone using RAMS model. Proc. Second EGS Plinius Conf. on Mediterranean Storms, Siena, Italy, EGS, 225–235.

    • Search Google Scholar
    • Export Citation
  • Murray, R. J., , and Simmonds I. , 1991: A numerical scheme for tracking cyclone centres from digital data. Part I: Development and operation of the scheme. Aust. Meteor. Mag, 39 , 155166.

    • Search Google Scholar
    • Export Citation
  • Murray, R. J., , and Simmonds I. , 1995: Responses of climate and cyclones to reductions in Arctic winter sea ice. J. Geophys. Res, 100 , 47914806.

  • Obled, C., , Bontron G. , , and Garçon R. , 2002: Quantitative precipitation forecasts: A statistical adaptation of model outputs through an analogues sorting approach. Atmos. Res, 63 , 303324.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Palutikof, J. P., , Goodess C. M. , , and Guo X. , 1994: Development of daily rainfall scenarios for south-west Spain using a circulation-type approach to downscaling. Int. J. Climatol, 14 , 853896.

    • Search Google Scholar
    • Export Citation
  • Pinto, J. G., , Klawa M. , , Ulbrich U. , , Rudari R. , , and Speth P. , 2001: Extreme precipitation events over north-west Italy and their relationship with tropical–extratropical interactions over the Atlantic. Proc. Third EGS Plinius Conf. on Mediterranean Storms, Baja Sardinia, Italy, EGS, 321–332.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , and Atlas R. , 2001: Tropical cyclone–like vortices in the extratropics: Observational evidence and synoptic analysis. Wea. Forecasting, 16 , 734.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reale, O., , Feudale L. , , and Turato B. , 2001: Evaporative moisture sources during a sequence of floods in the Mediterranean region. Geophys. Res. Lett, 28 , 20852088.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rice, J., 1982: The derivation of a computer-based synoptic climatology of Southern Hemisphere extratropical cyclones. M.S. thesis, Dept. of Meteorology, The University of Melbourne, 113 pp.

    • Search Google Scholar
    • Export Citation
  • Romero, R., , Ramis C. , , Alonso S. , , Doswell C. A. III, , and Stensrud D. J. , 1998: Mesoscale model simulations of three heavy precipitation episodes in the western Mediterranean region. Mon. Wea. Rev, 126 , 18591881.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rudari, R., 2001: European climate predictability and the synoptic control on regional intense rainfall events. Ph.D. thesis, Università degli Studi di Padova, 248 pp.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., , Box J. E. , , Barry R. G. , , and Walsh J. E. , 1993: Characteristics of Arctic synoptic activity, 1952–89. Meteor. Atmos. Phys, 51 , 147164.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stein, O., 2000: The variability of Atlantic–European blocking as derived from long SLP time series. Tellus, 52A , 225236.

  • Tafferner, A., 1990: Lee cyclogenesis resulting from the combined outbreak of cold air and potential vorticity against the Alps. Meteor. Atmos. Phys, 43 , 3147.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tibaldi, S., , and Molteni F. , 1990: On the operational predictability of blocking. Tellus, 42A , 343365.

  • Tibaldi, S., , and Cacciamani C. , 2001: ACCORD: Atmospheric Circulation Classification and Regional Downscaling. Climate Res, 17 , 126165.

    • Search Google Scholar
    • Export Citation
  • Treidl, R. A., , Birch E. C. , , and Sajekcki P. , 1981: Blocking action in the Northern Hemisphere: A climatological study. Atmos.– Ocean, 19 , 123.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

(a) Orographic map with annotations showing the geographic areas of interest. Events with quantiles over 0.95: (b) Seasonal distribution for all the stations in the domain shows autumn as the season that presents the highest percentage of extreme precipitation events over the entire domain. (c) Conditional probability maps of extreme events [Xx′: FX(x′) = 0.95] for a station, baricentric with respect to the Tuscany cluster, in the SON season. (d) Cluster of stations selected for the combined study. Sample size = 36; x′ = 39 mm

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 2.
Fig. 2.

(a) Classification of synoptic conditions bringing severe weather in the Tuscany region. The sketches represent characteristic conditions based on the large statistical sample of cases identified in this study. Solid line arrows represent the midtroposphere flow while dashed arrows represent the low-level flow. Arrow's thickness distinguishes relative strength. The relevant orography is schematized in gray shading. (b) Mean condition for the moisture flux, computed as a vertical integral from 1000 to 300 hPa. The mean condition is computed on the basis of the reference time series identified from the precipitation records of the different stations selected on the basis of the conditional probability maps. For an immediate comprehension of the wettest areas, colors represent the module of the vectors schematized by the arrows

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 3.
Fig. 3.

Conditional probability maps at (a) lag 0 and (b) lag +1 day for a station in the southern part of Tuscany in the autumn season. The reference station is indicated with a cross. The sample size is 55. The first (conditioning) reference threshold in terms of quantile is 0.92 and corresponds to 31 mm cumulated in 1 day; the second (conditioned) threshold in terms of quantile is 0.82

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 4.
Fig. 4.

Climatology for the autumn season (years 1958–86) for (a) 500-hPa geopotential and (b) sea level pressure. Anomaly fields computed as the mean of the time series—selected on the basis of the cluster quantiles dates—minus the climatology for the autumn season (years 1958–86) for (c) 500-hPa geopotential and (d) sea level pressure. Std dev anomaly fields computed as the 4-day std dev of the time series— selected on the basis of the cluster quantiles dates—minus the climatological value for the autumn season (years 1958–86) for (e) 500-hPa geopotential and (f) sea level pressure

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 5.
Fig. 5.

Mean condition for the low-level wind. The mean condition is computed on the basis of the reference time series identified from the precipitation records of the different stations selected on the basis of the conditional probability maps

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 6.
Fig. 6.

Composite maps for geopotential height at 500 hPa: (a) type a—the sample size used for this composite represents about 65% of the total sample, and (b) type b—the sample size used for this composite represents about 35% of the total sample. The contour lines are labeled in meters. Composite maps for sea level pressure: (c) type a—the sample size used for this composite represents about 65% of the total sample, and (d) type b—the sample size used for this composite represents about 35% of the total sample. The contour lines are labeled in hectopascals. The lows are marked as L and the highs as H

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 7.
Fig. 7.

Std dev anomaly fields computed as the 4-day standard deviation of the time series—selected on the basis of the type b dates—minus the climatologic value for the autumn season (years 1958–86) for (a) 500-hPa geopotential and (b) sea level pressure. (c) The low-level wind for the type b dates

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 8.
Fig. 8.

Blocking frequency for all the longitudes. (a) The type a distribution (continuous line) is compared with climatology (dotted line). (b) The type b distribution (continuous line) is compared with climatology (dotted line).

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 9.
Fig. 9.

Reproduction of the Deutscher Wetterdienst weather charts at 0000 UTC 25 Oct 1980. (a) Geopotential height at 500 hPa. The contour lines are labeled in dekameters. The synoptic trough low is centered over the Scandinavian peninsula. (b) Sea level pressure. The contour lines are labeled in hectopascals. The extended primary low is over the south part of Scandinavia, and its cold front impinges the Alpine barrier from northwest. A lee cyclone is visible in the Gulf of Genoa and the associated frontal system is over Italy. In both maps lows are marked as T and highs as H. The maps show one of the best examples of type a conditions.

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 10.
Fig. 10.

Cyclone tracks for the 25 Oct 1980 event. The triangles represent open systems; the circles represent closed systems. Each mark is separated from the following one by a 6-h time interval. The size of the mark represents the strength of the cyclone. The cyclone labeled 1731 is the primary system, while the one labeled 1876 represents the secondary system influencing directly the heavy precipitation associated with the event in the target region. The gray dot is the position of the primary cyclone when the secondary one appears. The black dot represents the position where the secondary cyclone arises and the position of the cyclone most favorable for generating heavy precipitation in the target region

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 11.
Fig. 11.

Reproduction of the Deutscher Wetterdienst weather charts at 0000 UTC 29 Oct 1966. (a) Geopotential height at 500 hPa. The contour lines are labeled in dekameters. The map shows a clear cutoff low condition (type b). The cutoff low has moved from the Gulf of Biscay to the Pyrenees while the disturbance that generated the trough is located north of Scandinavia. (b) Sea level pressure. The contour lines are labeled in hectopascals. The map shows the secondary cyclone in the Gulf of Genoa with its associated front. In both maps lows are marked as T and highs as H. The maps show one of the best examples of type b conditions

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Fig. 12.
Fig. 12.

Cyclone tracks for the 29 Oct 1966 event. The triangles represent open systems; the circles represent closed systems. Each mark is separated from the following one by a 6-h time interval. The size of the mark represents the strength of the cyclone. The cyclone labeled 2158 is the primary system; the one labeled 2333 is the secondary triggered system. The light gray triangle is the position of the primary cyclone when the secondary one appears. The dark gray dot represents the position where the secondary cyclone arises, and the black dot is the position of the cyclone most favorable for generating heavy precipitation in the target region

Citation: Journal of Hydrometeorology 5, 3; 10.1175/1525-7541(2004)005<0390:TAMIAF>2.0.CO;2

Save