• Agència Catalana de l’Aigua, 2001: Delimiting of flood plain for the draft of the INUNCAT: Internal Basins of Catalonia: Hydrological calculations and determination of the flood wave discharge (in Catalan) Vol. 2, Department de Medi Ambient, Generalitat de Catalunya, 88 pp. [Available online at http://mediambient.gencat.net/aca/documents/ca/planificacio/inuncat/conquesinternes/calculhidro_vii.pdf.].

  • Agència Catalana de I’Aigua, 2003: Technical recommendations for flood studies at local area (in Catalan). Department de Medi Ambient, Generalitat de Catalunya, 106 pp. [Available online at http://mediambient.gencat.net/aca/documents/ca/planificacio/criteris_tec-nics/recomanacions_tecniques_estudis_inundabilitat.pdf.].

  • Alpert, P., and Coauthors, 2002: The paradoxical increase of Mediterranean extreme daily data in spite of decrease in total values. Geophys. Res. Lett., 29 .1536, doi:10.1029/2001GL013554.

    • Search Google Scholar
    • Export Citation
  • Anderson, M. L., , Chen Z. Q. , , Kavvas M. L. , , and Feldman A. , 2002: Coupling HEC-HMS with atmospheric models for prediction of watershed runoff. J. Hydrol. Eng., 7 , 312318.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Antolik, M. S., 2000: An overview of the National Weather Service’s centralized statistical quantitative precipitation forecast. J. Hydrol., 239 , 306337.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bacchi, B., , Buzzi A. , , Grossi G. , , and Ranzi R. , 2002: Flood forecasting in a midsize catchment in the southern Alps: Recent experiences on the use of couple meteorological and hydrological model. Proc. Third EGS Plinius Conf., Baja Sardinia, Italy, Consiglio Nazionale delle Ricerche, 201–208.

  • Bellon, A., , and Zawadzki I. , 1994: Forecasting of hourly accumulations of precipitation by optimal extrapolation of radar maps. J. Hydrol., 157 , 211233.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., , and Seaman N. L. , 1985: A simple scheme for improved objective analysis in curved flow. Mon. Wea. Rev., 113 , 11841198.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Betts, A. K., , and Miller M. J. , 1986: A new convective adjustment scheme. Part II: Single column using GATE wave, BOMEX, ATEX and Artie air-mass data sets. Quart. J. Roy. Meteor. Soc., 112 , 693709.

    • Search Google Scholar
    • Export Citation
  • Blackadar, A. K., 1979: High resolution models of the planetary boundary layer. Adv. Environ. Sci. Eng., 1 , 5085.

  • Brémaud, P. J., , and Pointin Y. , 1993: Forecasting heavy rainfall from rain cell motion using radar data. J. Hydrol., 142 , 373389.

  • Castelli, F., 1995: Atmosphere modelling and hydrology prediction uncertainty. Proc. U.S.-Italy Research Workshop on Hydrometeorology: Impacts and Management of Extreme Floods, Perugia, Italy, Water Resources Research and Documentation Center and Colorado State University, 13–17.

  • Chow, V. T., , Maidment D. R. , , and Mays L. W. , 1988: Applied Hydrology. McGraw-Hill, 572 pp.

  • Clark, M. P., , and Hay L. E. , 2004: Use of medium-range numerical weather prediction model output to produce forecasts of stream-flow. J. Hydrometeor., 5 , 243262.

    • Search Google Scholar
    • Export Citation
  • Conway, B. J., , and Browning K. A. , 1988: Weather forecasting by interactive analysis of radar and satellite imagery. Philos. Trans. Roy. Soc., 324 , 299315.

    • Search Google Scholar
    • Export Citation
  • Davis, C. A., , and Emanuel K. A. , 1991: Potential vorticity diagnostics of cyclogenesis. Mon. Wea. Rev., 119 , 19291953.

  • Deidda, R., 2000: Rainfall downscaling in a space-time multifractal framework. Water Resour. Res., 36 , 17791794.

  • Deidda, R., , Benzi R. , , and Siccardi F. , 1999: Multifractal modelling of anomalous scaling laws in rainfall. Water Resour. Res., 35 , 18531867.

  • Dolciné, L., , Andrieu H. , , Sempere-Torres D. , , and Creutin D. , 2001: Flash flood forecasting with coupled precipitation model in mountainous Mediterranean basin. J. Hydrol. Eng., 6 , 19.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1993: A nonhydrostatic version of the Penn State/NCAR Mesoscale Model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 14931513.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ferraris, L., , Rudari R. , , and Siccardi F. , 2002: The uncertainty in the prediction of flash floods in the northern Mediterranean environment. J. Hydrometeor., 3 , 714727.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gómez, M., , and López R. , 1998: Rehabilitation of sewer networks in member states: Sant Boi sewer network. Research Rep., European Project SPRINT SP-98, 112 pp.

  • Grell, G. A., , Dudhia J. , , and Stauffer D. R. , 1995: A description of the fifth-generation of the Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 122 pp.

  • Groisman, P. Y., and Coauthors, 1999: Changes in the probability of heavy precipitation: Important indicators of climatic change. Climatic Change, 42 , 243283.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hardaker, P. J., , Collier C. G. , , and Pierce C. E. , 1994: The GANDOLF system: A look to the future for operational flood forecasting using satellite and radar measurements. British Hydrological Society Occasional Paper No. 5: Hydrological Uses of Weather Radar, K. Tilford, Ed., 39–63.

  • Hewitson, B. C., , and Crane R. G. , 1992: Large-scale atmospheric controls on local precipitation in tropical Mexico. Geophys. Res. Lett., 19 , 18351838.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Homar, V., , Romero R. , , Ramis C. , , and Alonso S. , 2002: Numerical study of the October 2000 torrential precipitation event over eastern Spain: Analysis of the synoptic-scale stationarity. Ann. Geophys., 20 , 20472066.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and Pan H-L. , 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jasper, K., , and Kaufmann P. , 2003: Coupled runoff simulations as validation tools for atmospheric models at the regional scale. Quart. J. Roy. Meteor. Soc., 129 , 673692.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43 , 170181.

  • Llasat, M. C., , Rigo T. , , and Barriendos M. , 2003: The “Montserrat-2000” flash flood event: A comparison with the floods that have occurred in the northeast Iberian Peninsula since the 14th century. Int. J. Climatol., 23 , 453469.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Martín-Vide, J. P., , Nierola D. , , Bateman A. , , Navarro A. , , and Velasco E. , 1999: Runoff and sediment transport in a torrential ephemeral stream of the Mediterranean coast. J. Hydrol., 225 , 118129.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Menéndez, M., 1998: Design discharge calculations and flood plain management. CEDEX, Madrid, Research Rep., European Project Flood Aware: Prevention and Forecast of Floods, 29 pp. [Available online at http://www.lyon.cemagref.fr/projects/floodaware/report/03cedex.pdf.].

  • Mlawer, E. J., , Taubman S. J. , , Brown P. D. , , Iacono M. J. , , and Clough S. A. , 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Rev., 102 , D14. 1666316682.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Murphy, A. H., 1993: What is a good forecast? An essay on the nature of goodness in weather forecasting. Wea. Forecasting, 8 , 281293.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nash, J. E., , and Sutcliffe J. V. , 1970: River flow forecasting through conceptual models. Part I: A discussion of principles. J. Hydrol., 10 , 3. 282290.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ranzi, R., , Bacchi B. , , and Grossi G. , 2003: Runoff measurements and hydrological modelling for the estimation of rainfall volumes in an alpine basin. Quart. J. Roy. Meteor. Soc., 129 , 653672.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reisner, J., , Rasmussen R. M. , , and Bruintjes R. T. , 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 Mesoscale Model. Quart. J. Roy. Meteor. Soc., 124B , 10711107.

    • Search Google Scholar
    • Export Citation
  • Romero, R., 2001: Sensitivity of a heavy rain producing western Mediterranean cyclone to embedded potential vorticity anomalies. Quart. J. Roy. Meteor. Soc., 127 , 25592597.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Romero, R., , Martín A. , , Homar V. , , Alonso S. , , and Ramis C. , 2006: Predictability of prototype flash flood events in the western Mediterranean under uncertainties of the precursor upper-level disturbance. Adv. Geosci., 7 , 5563.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Siccardi, F., 1996: Rainstorm hazards and related disasters in the western Mediterranean region. Remote Sens. Rev., 14 , 521.

  • Singh, V. P., 1997: Effect of spatial and temporal variability in rainfall and watershed characteristics on stream flow hydrograph. Hydrol. Processes, 11 , 16491669.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • USACE-HEC, 1998: HEC-HMS Hydro-logic Modeling system user’s manual. U.S. Army Corps of Engineers Hydrologic Engineering Center, Davis, CA, 188 pp. [Available online at http://www.hec.usace.army.mil.].

  • USACE-HEC, 2000: Hydrologic Modeling system HEC-HMS. Tech. reference manual, U.S. Army Corps of Engineers Hydrologic Engineering Center, Davis, CA, 157 pp. [Available online at http://www.hec.usace.army.mil.].

  • U.S. Department of Agriculture, 1986: Urban hydrology for small watersheds. Tech. Release 55, Nature Resources Conservation Service, Washington, DC, 164 pp.

  • von Storch, H., , and Zwiers F. W. , 1999: Statistical Analysis in Climate Research. Cambridge University Press, 494 pp.

  • Wilks, D., 1999: Multisite downscaling of daily rainfall with a stochastic weather generator. Climate Res., 11 , 125136.

  • Zhang, D. L., , and Anthes R. A. , 1982: A high resolution model of the planetary boundary layer: Sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteor., 21 , 15941609.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, D. L., , and Fritsch J. M. , 1986: Numerical simulation of the meso-β scale structure and evolution of the 1977 Johnstown flood. Part I: Model description and verification. J. Atmos. Sci., 43 , 19131943.

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

    (a) Geographical location of the IBC where the Montserrat flash-flood event was produced. Several catchments (shaded) and locations affected by the episode are indicated. (b) The Catalan topography with a depiction of the main mountainous systems and rivers (Montserrat Mountain and Llobregat River are indicated).

  • View in gallery

    Rain gauge distribution in the IBC from the SAIH. It includes a total of 126 automatic rainfall stations distributed over an area of 16 000 km2.

  • View in gallery

    Digital terrain model of Llobregat River basin. It has a cell size of 50 m and displays the basin division (numbered), tributaries, stream gauges (circles), and reservoirs (triangles) mentioned in the text.

  • View in gallery

    (a) CAPPI reflectivity at 1.2-km altitude recorded by the Barcelona radar at 0400 LT 10 Jun 2000. (b) SAIH-derived analysis of accumulated rainfall over the IBC during the Montserrat episode.

  • View in gallery

    Observed, SAIH rain gauge driven, and MM5–NCEP simulation driven runoff discharge at (a) Súria, (b) Sant Sadurní, (c) Castellbell, (d) Abrera, and (e) Sant Joan Despí.

  • View in gallery

    SAIH rain gauge driven runoff discharge at Sant Joan Despí for the different (a) spatial and (b) temporal discretizations.

  • View in gallery

    Configuration of the four computational domains used for the MM5 numerical simulations (horizontal resolutions are 54, 18, 6, and 2 km, respectively) and MM5–NCEP simulation initial state, showing geopotential height at 500 hPa (continuous line, in gpm), temperature at 500 hPa (dashed line, in °C), and isentropic PV on the 330-K surface (shaded, according to scale) at 0000 UTC 9 Jun 2000.

  • View in gallery

    Spatial distribution of accumulated rainfall during the Montserrat event in the Llobregat basin, from (a) SAIH rain gauges, (b) MM5–NCEP simulation, (c) MM5–NCEP–4D simulation, and (d) MM5–ECMWF simulation. Contour interval (CI) is 20 mm starting at 20 mm.

  • View in gallery

    Accumulated volume during the Montserrat event, per subcatchment of the Llobregat basin, from SAIH rain gauge, MM5–NCEP, MM5–NCEP–4D, and MM5–ECMWF simulations. See Fig. 3 for subcatchment numbering.

  • View in gallery

    Temporal sequence, at l-h time steps, of accumulated volume in the Llobregat basin during the Montserrat event, from SAIH rain gauge, MM5–NCEP, MM5–NCEP–4D, and MM5–ECMWF simulations.

  • View in gallery

    Spatial distribution of accumulated rainfall during the Montserrat event in the Llobregat basin, from (a) −5% PV, (b) +5% PV, (c) WEST, and (d) EAST simulations. CI is 20 mm starting at 20 mm.

  • View in gallery

    Accumulated volume during the Montserrat event, per subcatchment of the Llobregat basin, from (a) SAIH rain gauge, −5% PV, and +5% PV simulations, and (b) SAIH rain gauge, WEST, and EAST simulations.

  • View in gallery

    Temporal sequence, at l-h time steps, of accumulated volume in the Llobregat basin during the Montserrat event, from SAIH rain gauge, −5% PV, +5% PV, WEST, and EAST simulations.

  • View in gallery

    Observed, −5% PV simulation driven, +5% PV simulation driven, WEST simulation driven, and EAST simulation driven runoff discharge at (a) Castellbell, (b) Abrera, and (c) Sant Joan Despí.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 123 123 18
PDF Downloads 88 88 9

A Hydrometeorological Modeling Study of a Flash-Flood Event over Catalonia, Spain

View More View Less
  • 1 Grup de Meteorologia, Departament de Física, Universitat de les Illes Balears, Palma de Mallorca, Spain
  • | 2 Secció d’Enginyeria Hidràulica i Hidrològica, Departament d’Enginyeria Hidràulica, Marítima i Ambiental, Universitat Politècnica de Catalunya, Barcelona, Spain
  • | 3 Grup de Meteorologia, Departament de Física, Universitat de les Illes Balears, Palma de Mallorca, Spain
© Get Permissions
Full access

Abstract

During the early morning of 10 June 2000, the Catalonia region was affected by a hazardous convective rainfall episode that produced a large increase on flow regimes in many internal catchments of the region. The present modeling study is focused upon the Llobregat basin, the biggest internal catchment with a drainage area of 5040 km2. The first objective of the study is the characterization of the watershed hydrological response to this flash-flood event based on rain gauge data and the Hydrologic Engineering Center’s Hydrological Modeling System (HEC-HMS) runoff model. The HEC-HMS model has been calibrated using five episodes of similar torrential characteristics, and the effects of the spatial segmentation of the basin and of the temporal scale of the input rainfall field have been examined. These kinds of episodes present short recurrence intervals in Mediterranean Spain, and the use of mesoscale forecast driven runoff simulation systems for increasing the lead times of the emergency management procedures is a valuable issue to explore. The second objective uses NCEP and ECMWF analyses to initialize the nonhydrostatic fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5) in order to simulate the 10 June 2000 flash-flood episode with appropriate space and time scales to force the runoff model. The final objective analyzes the sensitivity of the catchment’s response to the spatial and temporal uncertainty of the rainfall pattern based on an ensemble of perturbed MM5 simulations. MM5 perturbations are introduced through small shifts and changes in intensity of the precursor upper-level synoptic-scale trough. Main results indicate that 1) an optimum configuration of the runoff model can be clearly defined that best adjusts the simulated basin’s hydrological response to observed peak discharges, their timing, and total volume; 2) the MM5-control driven runoff simulation shows a reasonable reproduction of the observed discharge at the basin’s outlet and appears to be a suitable tool for the hydrometeorological forecasting of flash floods in the Llobregat basin as a whole; and 3) the ensemble of perturbed runoff simulations does not exhibit any relevant degradation of the forecast skill, and some of the members even outperform the control experiment at different stream gauge locations. That is, the catchment is relatively insensitive to rainfall forecast errors of a few tenths of kilometers and no more than 1–2 h.

Corresponding author address: Arnau Amengual, Dept. de Fisica, Universitat de les Illes Balears, 07122 Palma de Mallorca, Spain. Email: arnau.amengual@uib.es

Abstract

During the early morning of 10 June 2000, the Catalonia region was affected by a hazardous convective rainfall episode that produced a large increase on flow regimes in many internal catchments of the region. The present modeling study is focused upon the Llobregat basin, the biggest internal catchment with a drainage area of 5040 km2. The first objective of the study is the characterization of the watershed hydrological response to this flash-flood event based on rain gauge data and the Hydrologic Engineering Center’s Hydrological Modeling System (HEC-HMS) runoff model. The HEC-HMS model has been calibrated using five episodes of similar torrential characteristics, and the effects of the spatial segmentation of the basin and of the temporal scale of the input rainfall field have been examined. These kinds of episodes present short recurrence intervals in Mediterranean Spain, and the use of mesoscale forecast driven runoff simulation systems for increasing the lead times of the emergency management procedures is a valuable issue to explore. The second objective uses NCEP and ECMWF analyses to initialize the nonhydrostatic fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5) in order to simulate the 10 June 2000 flash-flood episode with appropriate space and time scales to force the runoff model. The final objective analyzes the sensitivity of the catchment’s response to the spatial and temporal uncertainty of the rainfall pattern based on an ensemble of perturbed MM5 simulations. MM5 perturbations are introduced through small shifts and changes in intensity of the precursor upper-level synoptic-scale trough. Main results indicate that 1) an optimum configuration of the runoff model can be clearly defined that best adjusts the simulated basin’s hydrological response to observed peak discharges, their timing, and total volume; 2) the MM5-control driven runoff simulation shows a reasonable reproduction of the observed discharge at the basin’s outlet and appears to be a suitable tool for the hydrometeorological forecasting of flash floods in the Llobregat basin as a whole; and 3) the ensemble of perturbed runoff simulations does not exhibit any relevant degradation of the forecast skill, and some of the members even outperform the control experiment at different stream gauge locations. That is, the catchment is relatively insensitive to rainfall forecast errors of a few tenths of kilometers and no more than 1–2 h.

Corresponding author address: Arnau Amengual, Dept. de Fisica, Universitat de les Illes Balears, 07122 Palma de Mallorca, Spain. Email: arnau.amengual@uib.es

1. Introduction

The topography of the Spanish Mediterranean area makes it especially prone to flash-flood events. Mountain systems near the coast usually act as natural barriers to the warm moist Mediterranean air, inducing the generation of intense rainfall rates that show high variability in space and time. Serious damage can occur when intense convective rainfall events combine with short hydrological response times, characteristic of steep streams and increasing urbanization rates in coastal areas. Furthermore, in this semiarid environment many small and medium steep streams are ephemeral, which can cause unexpected and extensive flood damage. Increased flows over short periods, high flow velocities, and large volumes of sediment constitute threats to property and human life (Martín-Vide et al. 1999). Given all these factors, it is very important to obtain accurate forecasts of the rising flows before these kinds of episodes.

During flooding events the amount of time available for the implementation of emergency management procedures is limited. The possibility of increasing the lead times associated with the runoff forecasting must be examined, since traditional warning systems based on rainfall observations do not provide the timely predictions required to implement the expected precautionary civil protection measures (Siccardi 1996). In the western Mediterranean, recent efforts in this issue have focused on quantifying rainfall amounts from radar (Dolciné et al. 2001), since from these estimations, rainfall–runoff models can provide forecasts with lead times ranging from <l h (Brémaud and Pointin 1993) to a few hours depending on the rainfall types (Conway and Browning 1988; Bellon and Zawadzki 1994; Hardaker et al. 1995). However, the only way to gain additional lead time in runoff forecasting is to have rainfall information ahead of its occurrence. One way in which this can be accomplished is by translating mesoscale model rainfall forecasts into runoff forecasts (Anderson et al. 2002; Ferraris et al. 2002). The feasibility of real-time forecasting over the Spanish Mediterranean area using the capabilities of a hydrological model forced by a mesoscale atmospheric model becomes one of the main issues of the present study.

The 10 June 2000 flash-flood event in Catalonia (northeast Spain) is a good example of the catastrophic effects of a rapid and sudden flow increase in a short time period. The region was affected by a convective rainfall episode that affected many internal catchments of the region and produced serious material and human damages. The feasibility of hydrometeorological forecasting model strategies will be examined for this event. Specifically, the study is centered on the Llobregat basin, with a drainage area of 5040 km2. The first objective of the paper is to reproduce the hydrological response to the flash-flood event using the Hydrologic Engineering Center’s Hydrological Modeling System (HEC-HMS) runoff model. An independent sample of events is used to calibrate the HEC-HMS in terms of soil behavior (losses and imperviousness), which exerts a fundamental role over the runoff volume for the episode, and flood wave celerity in the main channels of the catchment, an important factor owing to the high flow velocities. To optimize the rainfall–runoff model setup, the effects of diverse spatial and temporal scales of the rainfall field on the simulated basin response have been quantified.

The second objective of the paper is to test the appropriateness of the aforementioned atmospheric driven runoff simulations as a methodology for obtaining 12–48-h forecasts of these extreme events, which would greatly expand the time necessary for emergency management procedures. In particular, the HEC-HMS runoff model is forced with mesoscale rainfall forecasts derived from fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) simulations initialized with meteorological grid analyses from the National Centers for Environmental Prediction (NCEP) and the European Centre for Medium-Range Weather Forecasts (ECMWF). The third objective consists of assessing the sensitivity of the Llobregat basin to the inherent uncertainties of some aspects of the hydrometeorological forecasting chain (Ferraris et al. 2002; Castelli 1995; Murphy 1993). An ensemble of MM5 simulations with small shifts and variations of intensity of the precursor upper-level synoptic-scale trough is designed for this purpose. With this method, it is possible to assess the effects on the hydrological response due to relatively moderate spatial and temporal errors of the forecast rainfall field.

The rest of the paper is structured as follows: section 2 consists of a brief description of the study area; section 3 describes the hydrometeorological episode; section 4 describes the hydrological tools used for the basin characterization, the methodology followed for the calibration, and evaluation of the rainfall–runoff model and the analysis of the sensitivity to the spatiotemporal scales of the input rainfall field; section 5 contains the meteorological tools applied to forecast the rainfall event and to design the ensemble of perturbed mesoscale simulations; section 6 presents and discusses the results; and section 7 provides an assessment of the methodology, which includes future directions for later development of the system.

2. The study area

a. Overview of the Llobregat basin

The Llobregat basin is the most important of the internal hydrographic catchments in Catalonia (Fig. la) in terms of size, river length, mean flow and population living inside. It is composed of the Llobregat River and its main tributaries, the Anoia and the Cardener. Llobregat basin extends from the Pyrenees, with heights over 3000 m, through the Pre-Pyrenees, constituted by a band of folded Mesozoic materials, and crossing the central depression, formed by tertiary materials more or less eroded, with a height transition from 750 m in the Pre-Pyrenees to 200 m in the precoastal range. The last section of the river crosses the Mediterranean orographic systems, formed by two mountainous alignments almost parallel to the coast line: the precoastal range formed by varied morphological mounts [e.g., Montseny (1712 m), Montserrat (1236 m), and Serra del Cardó (942 m)] and the coastal range, consisting of small-altitude mountains [e.g., Montnegre (759 m), Collserola (512 m), and Garraf (660 m)] (Fig. lb). The basin has a drainage area of 5040 km2 and a maximum length close to 170 km.

Furthermore, the hydrographic catchment is divided into a wide range of climatic areas owing to the diversity of the pluviometric records depending on the altitude. Annual accumulated rainfall in the Llobregat basin can range from quantities exceeding 1000 mm in the Pyrenees (over 1000 m), 700 mm over Pre-Pyrenees (with elevations between 600 and 1000 m), and 500 mm for altitudes below 500 m. The rainfall regime is typical of the Mediterranean areas, with most heavy-rainfall episodes occurring mainly in autumn, with occasional episodes in the spring and the summer. These daily rainfall episodes can represent a large fraction of the annual amounts.

b. The rain and stream gauge networks

On 10 June 2000, heavy rainfall took place over the northeastern part of Spain and the most intense episode affected the whole of the internal basins of Catalonia (IBC; Fig. 1a). An analysis of the pluviometric evolution of the episode used 5-min rainfall data recorded at 126 stations inside the IBC and distributed over an area of 16 000 km2 (Fig. 2). These stations belong to the Automatic Hydrological Information System (SAIH) network of the Catalan Agency of Water (ACA). Out of the 126 stations, about 40–50 lie inside the Llobregat basin or near its boundaries.

Runoff in the Llobregat basin was recorded in five flow gauges (Fig. 3) located in (i) Súria town, on the Cardener River, with a dranaige area of 940 km2 and elevation from 250 m at gauge level to 2350 m in the Pyrenees; (ii) Sant Sadurní d’Anoia city, on the Anoia River, with a drainage area of 736 km2 and elevations from 125 m at gauge level to 850 m at headwater; and (iii) Castellbell (3340 km2), (iv) Abrera (3587 km2), and (v) Sant Joan Despí (4915 km2) towns along the Llobregat River. During the episode, 5-min runoff measurements were collected, jointly with the rainfall records, for the SAIH database.

3. Brief description of the Montserrat flash-flood episode

Heavy rainfall on 10 June 2000 lasted about 6 h, from 0200 to 0800 local time (LT corresponds to UTC + 2 h). The extraordinary rise of the Catalonia internal river basin flow regimes (e.g., El Llobregat, El Besós, El Francolí, and La Riera del Bisbal; see Fig. 1a for locations) produced serious damages. Some of the most notable disasters consisted of the partial destruction of the infrastructure of Montserrat’s Monastery (720 m) and some roads connecting with this mountainous area; the collapse of some bridges and sections in the plain roadway; and the flooding of residential zones with the attendant destruction of some dwellings, especially in the tourist municipality of El Vendrell (Fig. 1a). As a consequence, about 500 people had to be evacuated from the monastery and the episode caused five fatalities and material losses estimated at about 65 million euros.

The most remarkable hydrometeorological feature of this case, known as the “Montserrat” flash-flood event for its impact upon Montserrat Mountain, was the high intensity of the sustained rainfall, which accumulated hourly quantities above 100 mm and a 6-h maximum up to 200 mm. Figures 4a and 4b depict, respectively, the radar image of the lowest constant altitude plan position indicator (CAPPI) at 0400 LT and the cumulative rainfall distribution in the internal catchments from 2300 LT 9 June to 2300 LT 10 June. The maximum amounts were observed in the basin of the Llobregat River, with 224 mm in the town of Rajadell. Up to 134 mm were observed at Bisbal del Penedés town, in the basin of the Riera de Bisbal, of which above 100 mm occurred in less than 2 h. Values exceeding 100 mm were also observed in the basins of the Francolí, Gaià, and Foix (Llasat et al. 2003; see Fig. 1a for locations).

Focusing on the Llobregat basin (Fig. 3), the maximum flow discharge observed at Súria was 260 m3 s−1 at 1225 LT with a time to peak of 6 h (Fig. 5a, black solid line). In the Anoia affluent, the maximum observed flow stage was close to 2.7 m, with an associated peak discharge of 270 m3 s−1 at 0645 LT, and a time to peak of about 2 h (Fig. 5b). This was the first river that received the consequences of the event from 0130 until 0600 LT 10 June. Around 0300 LT, the rainfall extended to the entire Llobregat River basin, lasting for 4 h. As a consequence, 5-min intensities exceeded 120 mm per hour from 0300 to 0530 LT, with the highest values over the Llobregat basin occurring between 0400 and 0700 LT (Fig. 4a). In Castellbell town, an increase on the flow stage was observed above 4.5 m with several peak discharges, the maximum of these reaching 1000 m3 s−1 at 0800 LT with an associated time to peak close to 1 h and 40 min (Fig. 5c). In Abrera, a town sited approximately 15 km downstream, the maximum peak discharge was close to 1200 m3 s−1 (at 1250 LT; Fig. 5d). Finally, at Sant Joan Despí city, near the Llobregat River mouth where the last river gauge is installed, the maximum peak discharge was up to 1400 m3 s−1 at 1015 LT with a timing close to 2 h and 20 min (Fig. 5e). These short response times shown by the hydrographs (see Fig. 5) indicate substantial flow velocities in the subbasins induced by the high rainfall rates, and discharge that propagated very rapidly downstream (9 km h−1 on average).

The Spanish Center for Studies and Experimentation on Public Works (CEDEX) has issued, in the framework of a report on flood plain management, the return periods corresponding to certain runoff thresholds for several national catchments. For the Llobregat basin, the associated return period for an outflow of 1025 m3 s−1 is 10 yr, whereas for a peak discharge of 1600 m3 s−1 the recurrence interval is 20 yr (Menéndez 1998). These estimations emphasize the notable magnitude of the Montserrat event (1400 m3 s−1). The probability of suffering a similar catastrophic episode in the Llobregat basin is relatively low, but it must be emphasized that several hazardous episodes of different magnitudes and spatial scales are produced every year over the Spanish Mediterranean regions. In addition, future climate change scenarios and their possible impact on these types of events have to be taken into account. Some authors have indicated an increase in the probability of heavy-rainfall episodes in several parts of the world (Groisman et al. 1999), and a paradoxical increase of extreme daily rainfall in spite of a decrease in total values has been observed already in the Mediterranean basin (Alpert et al. 2002).

4. Hydrological tools

a. Rainfall–runoff model and input data

This study is carried out using the HEC-HMS rainfall–runoff model, developed by the U.S. Army Corps of Engineers (USACE-HEC 1998). HEC-HMS utilizes a graphical interface to build a semidistributed watershed model and to set up rainfall and control variables for the simulation. Figure 3 depicts the digital terrain model for the Llobregat basin—with a cell resolution of 50 m—together with the main watercourses and its tributaries, the considered division in subbasins and the location of the available river gauges. After the analysis presented in section 4c, the basin is divided in 39 subwatersheds with an average size of 126 km2 and an extension of 4915 km2 upstream from Sant Joan Despí, where the last flow gauge is installed.

HEC-HMS is forced using a single hyetograph for each subbasin. Rainfall spatial distributions were first generated from 30-min, 1-h, and 3-h accumulated values at SAIH rain gauges (see section 4c) using the kriging interpolation method with a horizontal grid resolution of 1 km. Then, temporal rainfall series were calculated for each subbasin as the areal average of the gridded rainfall within the subcatchment. The same methodology is used to assimilate forecast rainfall fields in HEC-HMS (section 6), except that atmospheric model gridpoint values are used instead of SAIH observations.

b. Theoretical background and basin calibration

The hydrologic model calculates runoff volume by subtracting from rainfall the water volume that is lost through interception, infiltration, storage, evaporation, and transpiration. The loss rate is calculated using the Soil Conservation Service curve number (SCS-CN; see, e.g., U.S. Department of Agriculture 1986). This method assumes the storm runoff volumes to be proportional to the rainfall volumes exceeding an initial abstraction threshold, through the ratio of the accumulated infiltration to a storage capacity. With this assumption and the continuity principle the cumulative volume of storm flow becomes nonlinearly related to the excess rainfall volume, which is a function of cumulative rainfall, soil cover, land use, and antecedent moisture (Chow et al. 1988; Bacchi et al. 2002). The SCS-CN model has been tested on several experimental areas and river basins worldwide and, in Catalonia, has been adopted by ACA in their technical studies (Agència Catalana de l’Aigua 2001, 2003). The SCS-CN model has the advantage that with a single parameter (the storage capacity) it reproduces two phenomena that are systematically observed during floods: an initial loss of rain and an increase in the efficiency of the basin in producing runoff as a response to the rainfall input (Ranzi et al. 2003). A synthetic unit hydrograph (UH) provided by SCS is used to convert rainfall excess into direct runoff on a watershed. The SCS-UH relates the peak discharge with the time to the UH peak through the subbasin area and a conversion constant. The flood hydrograph is routed using the Muskingum method (Chow et al. 1988; USACE-HEC 2000).

The Llobregat basin contains two reservoirs located in the upstream areas of the Cardener affluent and the Llobregat River (Fig. 3). Therefore, these watercourses cannot be modeled under the natural regime since the dams have an important hydrograph diffusion effect in the flood wave. The technical characteristics of both reservoirs—storage capacity, maximum outflow, maximum elevation, and initial level—have been obtained from the aforementioned technical reports by ACA. The detention ponds are modeled in HEC-HMS introducing a reservoir element that follows the elevation–storage–outflow relationship series, which depends on the characteristics of the dam, the outlet, and the spillway, besides the initial elevation of the water level.

The calibration of the rainfall–runoff model is carried out using five episodes of similar extraordinary characteristics to our Montserrat case of study, selected from the period between the deployment of the SAIH system (in 1996) and 2004 (Table 1). Owing to the malfunction of the flow gauge network for some of these episodes, the stream gauges at Abrera and Sant Joan Despí are not available for all the cases (Table 1), limiting to some extent the calibration of the lower Llobregat basin. Calibration of the infiltration parameters for each independent episode combines a manual procedure, where the SCS curve numbers are derived from field measurements and normal antecedent moisture conditions (Agència Catalana de l’Aigua 2001), and an automatic procedure, using as an objective function the peak-weighted root-mean-square error and applying the univariate-gradient search algorithm method (USACE-HEC 2000). This objective function, Z, is defined as
i1525-7541-8-3-282-eq1
where NQ is the number of computed hydrograph ordinates; qo(i) the observed flow at time i; qs(i) the simulated flow at time i, computed with a selected set of model parameters; and qo(mean) the mean of observed flows. Later, the CN values derived from each of the five episodes are averaged per subbasin. In addition and following the same methodology, the flood wave celerity for the main streams is also considered as a calibration index—by means of K parameter—owing to the nature of these kinds of episodes characterized by very high flow velocities. With the intention of capturing as well as possible the flow wave celerities involved in the Montserrat extreme episode, the maximum propagation velocities obtained among the previous calibration episodes were used. The calibrated parameters were then used to run HEC-HMS for the Montserrat case during a 96-h simulation, from 9 June 2000 at 0000 LT to 12 June 2000 at 2400 LT, with a 10-min time step. This period completely encompasses the flood event and the subsequent hydrograph tail.

The previous calibration process and subsequent rain gauge driven runoff simulations have been repeated for three spatial disaggregations of the catchment (21, 39, and 60 subbasins) with 1-h accumulated rainfall discretization and varying temporal resolutions of the incoming rainfall data (30 min, 1 h, and 3 h) with a 39-subbasin segmentation, in order to explore the sensitivities of the Llobregat basin and find an optimum configuration of the modeling system. The next section is fully devoted to this issue.

c. Sensitivity analysis to the spatial and temporal rainfall scales

To study the effects of the spatial scales of the rainfall field on the total basin response, the sensitivity of the catchment to three different spatial segmentations was evaluated: the basin was broken down into 21, 39, and 60 subbasins and the rainfall–runoff model was forced with hourly accumulated rainfall. The skill of the resulting runoff simulations is expressed in terms of the Nash–Sutcliffe efficiency criterion (NSE; Nash and Sutcliffe 1970), a “goodness-of-fit” measure widely used in hydrological model validation (Jasper and Kaufmann 2003; Dolciné et al. 2001). The NSE values can range from −∞ to 1, with higher values indicating a better agreement of the model results with the observations. NSE is defined as
i1525-7541-8-3-282-eq2
where xi and yi are the observed and model-simulated discharged values at flow gauge site at time i, respectively, and x is the mean observed value. This same index will be used in following sections to evaluate other spatial and temporal series.
The performance of the runoff simulations is also checked by means of the relative error of total volume at flow gauge sites, expressed as percentage (%EV):
i1525-7541-8-3-282-eq3
where Vo and Vs are the observed and simulated runoff volumes, respectively. Therefore, %EV > 0 and %EV < 0 would indicate an over- and underestimation of the volume by the model, respectively.

Table 2 shows the skill indices for the five calibration episodes. The results suggest a choice of 39 subbasins. For the Montserrat flash flood (Table 3) the optimum evaluation configuration in terms of model performance corresponds to 39 subbasins, particularly for the smallest watersheds, at Súria and Sant Sadurní gauges (∼1000 km2). For the largest basins, with areas exceeding 3000 km2, the distinction is not so clear, and in Castellbell the 60-subbasin subdivision appears to be superior. The last two downstream gauges (Abrera and Castellbell) present similar statistical scores among the three discretizations, though the 39-subbasin configuration is slightly superior (see Fig. 6a for basin outlet). In general, then, the rainfall–runoff model reproduces better the Montserrat event by dividing the Llobregat basin into 39 subbasins.

This result appears to be related to the number of rain gauges lying inside the whole basin (36), implying an average area per station of 136.5 km2. This area can be compared with the mean size of the 21, 39, and 60 subcatchments: 241.7, 126.0, and 81.9 km2, respectively. Therefore, for 21 subbasins the model hyetograph tends to overlap information of several rain gauges per subbasin, smoothing out detailed information of the spatial structure of the rainfall field that the rain gauge network is able to resolve. On the contrary, for 60 subbasins the rainfall–runoff model does not acquire reliable information of the rainfall field for ungauged catchments. The configuration using 39 subbasins seems to optimize the performance of the simulated basin response, since it represents more adequately the truly resolved spatial variabilities of the rainfall field. It is worth noting that the differences in the outflow characteristics at the flow gauges among the three watershed discretizations diminishes at larger scales (Table 3).

To study the effects of the temporal scales of the rainfall field on the total basin response, the sensitivity of the catchment using a 39-subbasin segmentation together with 30-min, l-h, and 3-h accumulated rainfall discretizations have been analyzed for the calibration and Montserrat episodes (Tables 4 and 5). Table 4 displays weak differences among the three temporal discretizations at the flow gauges indicated in Table 1. The NSE and %EV skill scores results in Table 5 indicate that the hourly discretization optimizes the simulation of the Llobregat basin response to the Montserrat event, since it presents the best performance in three of the five flow sites and a notable reproduction of the observed flow at the remaining gauges. Nevertheless, slightly better accuracy at the basin outlet is exhibited by the 3-h rainfall field discretization experiment (Fig. 6b). With the exception of Sant Joan, the hydrographs computed at the different flow gauges (not depicted) show greater peak discharges for the 30-min evaluation experiment and faster response times for the 3-h discretization when compared with observed. This result agrees with the notion that the higher the temporal variability of rainfall the greater the peak discharges, and also that a 3-h temporal discretization may be inappropriate for strong storms and/or watersheds with fast response times (Singh 1997).

From the set of the evaluation experiments analyzed, it seems that the most appropriate coherence between the spatial and temporal scales of the flash-flood event that the rain gauge network is able to resolve is reached for 39 subbasins combined with 1-h input rainfall data in the hydrological model (Tables 3 and 5). This is the configuration of the model that will be used for the mesoscale model driven runoff simulations. Tables 6 and 7 report the main hydrological model parameters: curve numbers, initial abstractions, times of concentration, and routing parameters.

5. Meteorological tools

The nonhydrostatic MM5 numerical model is used to perform the meteorological simulations. It is a high-resolution short-range weather forecast model developed by the PSU and NCAR (Dudhia 1993; Grell et al. 1995). Simulations are designed using 24 vertical σ levels and three spatial domains with 82 × 82 grid points (Fig. 7). Their respective horizontal resolutions are 54, 18, and 6 km, with integration time steps of 162, 54, and 18 s. The domains are centered in northeast Spain where the convective episode developed. In particular, the finest domain spans the entire Catalan territory and contiguous land and oceanic areas, and is used to supply the high-resolution rainfall fields to drive the hydrologic simulations. The interaction between the domains follows a two-way nesting strategy (Zhang and Fritsch 1986).

To initialize the model and to provide the time-dependent boundary conditions, NCEP and ECMWF meteorological grid analyses are used. MM5–NCEP simulation uses the analysis from the Global American Center for Environmental Prediction for the large domain, and are updated every 12 h with a 2.5° spatial resolution. MM5–ECMWF simulation uses the analysis of the European Centre for Medium-Range Weather Forecasts, with a spatial resolution of 0.3° and an update frequency of 6 h. In both cases the first-guess fields interpolated from the analyses on the MM5 model grid are improved using surface and upper-air observations with a successive-correction objective analysis technique (Benjamin and Seaman 1985). The tendencies along the model coarse-domain boundaries, specified by differences of the fields between the 12- and 6-h apart analyses, respectively, are applied using a Newtonian relaxation approach (Grell et al. 1995).

To parameterize moist convective effects the Betts–Miller cumulus scheme (Betts and Miller 1986) is used in the large domain and the Kain–Fritsch parameterization scheme (Kain 2004) in the intermediate domain. No convection scheme is in principle used in the inner one owing to the high horizontal resolution. Explicit microphysics is represented in all domains with prediction equations for cloud and rainwater fields, cloud ice, and snow allowing for slow melting of snow, supercooled water, graupel, and ice number concentration (Reisner et al. 1998). The planetary boundary layer physics is formulated using a modified version of the Hong and Pan scheme (Hong and Pan 1996). Surface temperature over land is calculated using a force–restore slab model (Blackadar 1979; Zhang and Anthes 1982) and over sea it remains constant during the simulations. Finally, long- and shortwave radiative processes are formulated using the Rapid Radiative Transfer Model (RRTM) scheme (Mlawer et al. 1997).

Furthermore, since it is debatable whether a 6-km-resolution domain can resolve convection appropriately without a convection scheme, an additional experiment has been designed. This simulation, labeled as MM5–NCEP–4D, coincides with MM5–NCEP except that it applies the Kain–Frisch scheme for the third domain. It also incorporates a fourth domain of 2-km horizontal resolution forced in two-way mode, in which convection is fully explicit. The possible benefits of enhanced horizontal resolution in this complex orographic region can thus be assessed.

With the purpose of generating the ensemble of perturbed simulations, the invertibility principle of Ertel potential vorticity (PV; Hoskins et al. 1985) is applied. In particular, we are interested in studying the sensitivity of the Montserrat hydrometeorological event to uncertainties in the precise representation of the upper-level precursor trough (shown in Fig. 7), being aware that small-scale aspects of the circulation are propitious to analysis or forecast errors. The piecewise PV inversion scheme described in Davis and Emanuel (1991) is then used as a clean approach to manipulate the upper-level synoptic trough in the model initial conditions. This requires a simple identification of the PV signature of the trough (shown as shaded in Fig. 7) and then the balanced mass and wind fields associated with that PV element can be used to alter the meteorological fields in a physically consistent way (effectively, a change in the structure or position of the trough). This method has already shown its value for assessing the predictability of flash-flood events in the western Mediterranean area (e.g., Romero 2001; Homar et al. 2002; Romero et al. 2006).

Using the NCEP-derived initial conditions, the upper-level trough intensity is perturbed ±5% (simulations −5% PV and +5% PV) and its position is displaced ±54 km along the zonal direction (experiments WEST and EAST). This short ensemble of simulations is a first approximation to the problem of incorporating the spatiotemporal uncertainty of the rainfall forecast into a medium-size catchment like the Llobregat basin. The whole set of MM5 simulations comprises a 36-h period, from 9 June 2000 at 0000 UTC to 10 June 2000 at 1200 UTC, after the end of the rainfall event in Catalonia.

6. Results and discussion

a. SAIH rain gauge driven runoff simulation

SAIH rain gauge derived rainfall of the “Montserrat event” is used to drive the calibrated HEC-HMS model in a single evaluation runoff simulation according to the methodology described in section 4. Figure 8a displays the spatial distribution of the accumulated rainfall upon the entire watershed, and Figs. 9 and 10 show, respectively, the accumulated volume per subbasin and the temporal sequence of accumulated volume over the entire basin at hourly time steps. These distributions will be compared against the simulated ones in next sections.

As a general overview, Table 8 and Fig. 5 show a good HEC-HMS skill for the characterization of the Llobregat basin response to the Montserrat event. NSE exceeds 0.65 in the set of flow gauges, and particularly at Abrera it exceeds 0.9. Relative errors in volume are reasonably small and only at Castellbell the error is close to 15%, though in all the stream gauges the volume is overestimated. Therefore, the results indicate a reasonable goodness of fit for the main peak discharges, their timing, and the volume estimations at the flow gauges. For small-scale features, however, the rain gauge driven run shows some inaccuracies: at Súria, multiple peaks are simulated instead of a single one (Fig. 5a); at Sadurní, Castellbell, and Abrera, the opposite case occurs and the model only simulates an envelope of the higher-frequency peaks (Figs. 5b, 5c and 5d); at Sant Joan Despí, a significant delay occurs in the time to peak (Fig. 5e).

b. MM5–NCEP, MM5–NCEP–4D, and MM5–ECMWF driven runoff simulations

To assess the skill of the MM5 mesoscale runs, the spatial and temporal distributions of the simulated rainfall volume are compared against the rain gauge derived volume pattern. The spatial comparison is done using the 39 subbasins as accumulation units for the whole episode, and the temporal comparison uses hourly accumulations for the whole basin. The degree of agreement between model and observed rainfall distributions is quantified using the NSE and root-mean-square error (RMSE) statistical indices (Table 9). With regard to the spatial distribution, the best skill scores are obtained by the MM5–ECMWF and MM5–NCEP–4D experiments. The MM5–NCEP experiment shows a moderately worse behavior, but on the contrary, it is the best for the temporal distribution. From a hydrological point of view, the MM5–NCEP simulation is the most suitable, attending to total precipitated water over the Llobregat basin, discharged volume at the basin outlet (Table 10), and the amount of the maximum hourly precipitated volume in the basin (Table 11). The superior behavior in these aspects of the MM5–NCEP rainfall simulation can be appreciated in Figs. 5, 9 and 10 and Table 8. MM5–ECMWF driven runoff simulation is not shown owing to its low skill on reproducing the episode, with corresponding statistical scores at the Llobregat basin outlet of NSE = −0.19 and %EV = −86.3. Figure 8d reveals that the MM5–ECMWF rainfall field is very deficient for this particular case study (cf. with Fig. 8a). In addition, the MM5–NCEP–4D driven runoff simulation exhibits a remarkable underestimation of the peak discharges and volumes at the different stream gauges, except at Súria (Fig. 5; Table 8). The MM5–NCEP–4D rainfall field contains very fine spatial features owing to the inclusion of the 2-km resolution forcing in the simulation (Fig. 8c), but the quantitative and spatial rainfall forecast is not better than the MM5–NCEP result (Fig. 8b). The inclusion of a convective scheme in the third domain appears to have a negative impact on the simulation. Therefore, MM5–NCEP simulation is chosen as the control simulation for this investigation.

The most remarkable deficiency of the control simulation is the northeastward shift of the rainfall pattern toward higher terrain and a more elongated shape with regard to the observed distribution, although with similar amounts (Fig. 8). It seems reasonable to argue that the catchment’s complex orography, dominated by the Pyrenees, the precoastal, and coastal ranges, is a determinant factor for the mesoscale model to produce that spatial distribution. Nevertheless, the simulated heavy rainfall lies within the Llobregat basin and the simulated timing of the rainfall episode is remarkably good (Fig. 10), in benefit of the MM5–NCEP driven runoff simulation. Since the MM5 control simulation tends to concentrate the maximum rainfall toward the upper part of the basin, where the two reservoirs are located, then a significant effect of hydrograph diffusion in the runoff would be expected. For the reservoir located in the Llobregat River, with an initial volume of 95 hm3 and an inflow volume of 14.0 hm3, the resulting outflow volume is 12.2 hm3. The peak discharge diminishes from 560.8 to 145.4 m3 s−1 with an attending delay close to 8 h. The diffusion effect by the reservoir located in the Cardener River is smaller: a decrease from 96.6 to 71.6 m3 s−1 with a delay of about 1.5 h.

The MM5–NCEP driven runoff simulation at the Súria site displays a good agreement with the observed peak discharge but not with its timing (Fig. 5a; Table 8). At the Sant Sadurní site the simulation is very deficient and no runoff is produced (Fig. 5b; Table 8): the mesoscale model widely underestimates the rainfall amounts in the Anoia watershed (cf. Figs. 8a and 8b; see Fig. 9). At the Castellbell and Abrera sites, runoff is widely overestimated producing a large error in the peak estimation and, consequently, making these results less suitable for use in emergency management directives (Figs. 5c and 5d). As the hydrograph is routed downstream, the overestimation of the runoff volume decreases owing to the deficit of the simulated rainfall in the southwestern subbasins, which contribute to the inflow. Another characteristic feature of the simulated runoff hydrographs along the Cardener and Llobregat Rivers is a lag time of around 3 h with respect to the observed flows, which is consistently routed downstream toward the basin outlet (Fig. 5e). This is due to several factors: the aforementioned hydrograph diffusion by the basin’s reservoirs, the fact that the core of the simulated heavy rainfall occurs farther upstream and with a certain delay compared with the observations, and the exceptional flood wave propagation for this particular event.

c. Ensemble of MM5-perturbed driven runoff simulations

Following the PV inversion method described in section 5, four additional mesoscale runs (−5% PV, +5% PV, WEST, and EAST) are performed in order to produce MM5-perturbed driven runoff simulations. These simulations, together with the previously referenced experiments MM5–NCEP, MM5–NCEP–4D, and MM5–ECMWF, become a useful experimental dataset to investigate the effects of the uncertainty of the mesoscale model initial conditions on the hydrometeorological chain. It is well known (e.g., Ferraris et al. 2002) that even slight spatial and temporal errors of the rainfall pattern can have a significant impact on the response of small catchments (up to hundreds of kilometers squared). However, the spatiotemporal gap between operational meteorological model outputs, and the required hydrological model inputs, should be considerably smoothed for a basin of medium size (thousands of kilometers squared). The results in the last section showed that the Llobregat basin was reasonably capable of filtering the forecast rainfall errors as long as the main rainfall nuclei lie within the catchment (Tables 8 and 9). Our hypothesis is that the basin should be relatively insensitive to realistic perturbations of the rainfall field introduced through the PV inversion method, and therefore the predictability of flash-flood events should be appreciable in this medium-size catchment. The use of ensemble strategies like the one tested here should provide a very useful probabilistic approach to the problem in the context of real-time operations.

Figures 11 and 12 display the spatial distributions of accumulated rainfall volume for the perturbed experiments. The −5% PV and WEST simulations (Figs. 11a and c) are fairly similar to the observed rainfall pattern (Fig. 8a), such that the spatial goodness-of-fit statistical indices of forecast rainfall outperform the results of the reference MM5–NCEP simulation. On the contrary, the spatial errors of +5% PV and EAST simulations are greater than in the reference experiment (Table 9). It seems, then, that a weaker or more distant upper-level precursor trough benefits the rainfall forecast of the Montserrat event. Presumably, the resulting slower-moving surface cyclone is more representative of the actual disturbance.

Furthermore, the whole ensemble of perturbed experiments slightly underestimates the total water collected over the Llobregat basin as it occurred with MM5–NCEP, although the −5% PV slightly improves the control simulation, with only 11.6 hm3 below the observed value (Table 10). The underestimation of precipitated volume is particularly severe in the Anoia subcatchment, where only the −5% PV run is able to produce appreciable values of rainfall (Fig. 11). Even so, the runoff simulation at Sant Sadurní gauge is rather poor, albeit for the rest of the ensemble dataset, runoff is not produced at all (Table 12). In addition, Tables 9 and 11 and Fig. 13 exhibit a certain uniformity in the temporal distributions of the rainfall volume for the perturbed experiments. Nevertheless, none of these prove to be superior in the temporal evolution to the control simulation. It is interesting to note that the ensemble of simulated rainfall fields exhibits a larger heterogeneity in space than in time (cf. the respective NSE indices; Table 9).

Finally, Table 12 summarizes the statistical indices at the five stream gauges for the ensemble of perturbed runoff simulations. At small basin scales, the skill is rather low owing to the lack of coherence among the meteorological and hydrological spatiotemporal scales (figures not shown). But at larger scales, the skill of the ensemble to forecast the discharge is considerably improved (Figs. 14a, 14b and 14c), to the extent that different members of the ensemble outperform the control simulation at different stream gauges (e.g., Castellbell and Abrera). These results demonstrate the value of an ensemble strategy in order to obtain a higher confidence interval in mesoscale model driven rainfall–runoff forecasts and to enact the appropriate emergency directives.

Essentially, the full set of driven runoff simulations does not exhibit any strong degradation of the forecast skill, not accounting for the ECMWF analysis driven simulation. It appears, then, that this catchment as a whole is relatively insensitive to typical errors of the forecast rainfall, like spatial shifts of a few tenths of kilometers and temporal shifts of not more than 1–2 h. The relative insensitivity of the Llobregat basin is surely a consequence of its medium size, and it is only lost for the smallest subbasins or when the heavy rainfall affects external hydrographic areas as for the ECMWF experiment. The filtering behavior of rainfall uncertainty found for the Llobregat basin in this case could also be raised by the moderate urbanization density and the relatively high predictability of the responsible mesoscale convective system. For smaller basins intercepting significant urban areas or with very local thunderstorms the capability of filtering the rainfall uncertainty is generally not found (Gómez and López 1998).

7. Conclusions and further remarks

This work has analyzed the feasibility of runoff simulations driven by numerical weather prediction mesoscale models over the Llobregat basin, characteristic of the Spanish Mediterranean environment, in an attempt to understand the sensitivity of the basin response to forecast errors, and help to gain additional lead times for warning and emergency procedures before flash-flood situations. The effects of different spatial and temporal rainfall field scales on the basin response has been studied by breaking down the basin in three different segmentations and by considering three temporal scales in a set of six experiments. A configuration considering a 39-subbasin division together with hourly temporal rainfall field discretization optimizes the basin response for the “Montserrat” event. It appears that this result is particularly related to the current density of rain gauges available within or very near the catchment. Similar tests and a recalibration of the runoff model should be applied using a long sample of mesoscale rainfall forecasts rather than rain gauge information in order to properly optimize the numerical system for operational purposes, but this task is beyond the objectives and capabilities of the present study.

Hazardous events present short recurrence periods in Mediterranean Spain as a whole, and the Montserrat event analyzed in this study is a forceful proof of their possible consequences. Using NCEP and ECMWF analyses to initialize the hydrometeorological chain, it was possible to obtain, at least at the basin outlet, reasonable runoff forecasts with up to 12–48-h lead times in the first case. These control runs were complemented by an ensemble of driven rainfall–runoff simulations that were shown to be useful to derive conclusions in depth. With the ensemble of MM5–NCEP perturbed simulations, it was possible to reduce the biases at some sites, as Castellbell and Abrera, where the control simulation would have not produced enough accurate runoff forecasts.

The set of perturbed mesoscale simulations was also introduced to address the effects of the meteorological external scale uncertainty. This source of uncertainty was reflected on the spatial and temporal distribution of the rainfall pattern in the Llobregat basin, with shifts of tenths of kilometers in the position of the heavy-rainfall cores and changes of about 1–2 h in their timing, in some cases outperforming the control run. Interestingly, the basin rainfall–runoff mechanisms were shown to smooth to a high degree the above spatial and temporal differences, thus enhancing, at least for this case, the predictability of flash floods in the Llobregat basin considering the entire catchment and the typical magnitude of mesoscale rainfall output errors. Nevertheless, one of the simulations of the ensemble—the MM5–ECMWF run—exhibited very poor results, and used in a deterministic hydrometeorological system, would have missed completely the hazardous event and inhibit any standard emergency procedure. This is a good example of where a simple multianalysis ensemble prediction system (EPS) accounting for the forecast variance associated to the initial conditions uncertainty would have been found of great value to trigger special flood warnings. However, to further extend the derived results, rainfall forecast errors found in existing mesoscale models should be examined for their typical magnitude and variability in space and time. Obviously, the higher performance of the NCEP-based simulations is simply a particularity of the Montserrat meteorological situation, and not an inherent aspect of this analysis dataset. It is reasonable to expect that the high resolution of ECMWF analyses would generally benefit nested mesoscale numerical forecasts in the region.

The methodology presented in this paper can be automated to obtain short-range runoff forecasts driven by high-resolution mesoscale predictions currently available in real time (see, e.g., http://mm5forecasts.uib.es). We believe that the enhanced predictability shown here in the Llobregat basin as a whole would also apply to many other hazardous episodes, as well as to other Mediterranean catchments of similar size and physical characteristics. In addition, it must be noted that runoff predictions for use in emergency management directives may not need to match exactly the peak discharges or their timing. These predictions must simply reach suitable thresholds so as to cause the appropriate directives to be enacted (Anderson et al. 2002).

The precise hydrological response of a catchment to rainfall events, in terms of the induced runoff, is strongly determined by the spatial and temporal variability of the soil properties. The infiltration mechanism acts as a highly nonlinear filter in the rainfall–runoff transformation and it has been modeled as an integrated process over each subbasin at discrete time steps. The model parameters related with this mechanism and with the flood wave routing have been calibrated using five events. These events are characterized by important discharges and high velocities of the associated flood waves, but the lack of flow data at some flow gauges for some of these events has posed difficulties in the basin calibration. To improve the reliability and skill of the rainfall–runoff model before such hazardous episodes, it would be desirable to get more information of other flash-flood events affecting the Llobregat basin. The expected future increase of the number of recorded cases in the SAIH database, and a larger number of stream gauges operating in the basin, will then permit an improvement of the basin configuration and the forecast and alert schemes.

In general, the set of driven rainfall–runoff simulations showed the lowest skill at the gauges covering small scales of the basin. None of the members of the ensemble, for example, were able to adequately reproduce the flow of the Anoia River for this episode, as an illustration of the forecasting limitations arisen as the hydrological scales of interest are decreased. Numerous techniques exist that attempt to mitigate this lack of coherence between the spatiotemporal scales of meteorological and hydrological models, particularly different applications of statistical downscaling (e.g., Clark and Hay 2004; Hewitson and Crane 1992; von Storch and Zwiers 1999; Wilks 1999; Antolik 2000) or disaggregation techniques (Deidda et al. 1999; Deidda 2000; Ferraris et al. 2002). These research lines appear as of the maximum interest for future studies in order to develop the most suitable hydrometeorological chain forecasting system upon the Spanish Mediterranean area.

Acknowledgments

Three anonymous reviewers are deeply acknowledged for their efforts to improve the quality of this manuscript. The authors would also like to express their appreciation to Maurici Ruiz and all his team from the Geographical Information Systems Laboratory of UIB (SIGUIB), for the valuable help and computer resources provided for this research. The Catalan Agency of Water (ACA) of the Department of Environment of the Government of Catalonia is acknowledged for providing the SAIH rainfall and flow discharge data. Figures 1 and 4b are courtesy of Dra. Carmen Llasat from Universitat de Barcelona. This work has been sponsored by INTERREG IIIB-MEDOCC 2003-03-4.3-I-079 (AMPHORE) and EVK1-2001-00042 (MEDIS) European projects, and by CGL 2005-03918/CLI (PRECIOSO) Spanish project.

REFERENCES

  • Agència Catalana de l’Aigua, 2001: Delimiting of flood plain for the draft of the INUNCAT: Internal Basins of Catalonia: Hydrological calculations and determination of the flood wave discharge (in Catalan) Vol. 2, Department de Medi Ambient, Generalitat de Catalunya, 88 pp. [Available online at http://mediambient.gencat.net/aca/documents/ca/planificacio/inuncat/conquesinternes/calculhidro_vii.pdf.].

  • Agència Catalana de I’Aigua, 2003: Technical recommendations for flood studies at local area (in Catalan). Department de Medi Ambient, Generalitat de Catalunya, 106 pp. [Available online at http://mediambient.gencat.net/aca/documents/ca/planificacio/criteris_tec-nics/recomanacions_tecniques_estudis_inundabilitat.pdf.].

  • Alpert, P., and Coauthors, 2002: The paradoxical increase of Mediterranean extreme daily data in spite of decrease in total values. Geophys. Res. Lett., 29 .1536, doi:10.1029/2001GL013554.

    • Search Google Scholar
    • Export Citation
  • Anderson, M. L., , Chen Z. Q. , , Kavvas M. L. , , and Feldman A. , 2002: Coupling HEC-HMS with atmospheric models for prediction of watershed runoff. J. Hydrol. Eng., 7 , 312318.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Antolik, M. S., 2000: An overview of the National Weather Service’s centralized statistical quantitative precipitation forecast. J. Hydrol., 239 , 306337.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bacchi, B., , Buzzi A. , , Grossi G. , , and Ranzi R. , 2002: Flood forecasting in a midsize catchment in the southern Alps: Recent experiences on the use of couple meteorological and hydrological model. Proc. Third EGS Plinius Conf., Baja Sardinia, Italy, Consiglio Nazionale delle Ricerche, 201–208.

  • Bellon, A., , and Zawadzki I. , 1994: Forecasting of hourly accumulations of precipitation by optimal extrapolation of radar maps. J. Hydrol., 157 , 211233.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., , and Seaman N. L. , 1985: A simple scheme for improved objective analysis in curved flow. Mon. Wea. Rev., 113 , 11841198.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Betts, A. K., , and Miller M. J. , 1986: A new convective adjustment scheme. Part II: Single column using GATE wave, BOMEX, ATEX and Artie air-mass data sets. Quart. J. Roy. Meteor. Soc., 112 , 693709.

    • Search Google Scholar
    • Export Citation
  • Blackadar, A. K., 1979: High resolution models of the planetary boundary layer. Adv. Environ. Sci. Eng., 1 , 5085.

  • Brémaud, P. J., , and Pointin Y. , 1993: Forecasting heavy rainfall from rain cell motion using radar data. J. Hydrol., 142 , 373389.

  • Castelli, F., 1995: Atmosphere modelling and hydrology prediction uncertainty. Proc. U.S.-Italy Research Workshop on Hydrometeorology: Impacts and Management of Extreme Floods, Perugia, Italy, Water Resources Research and Documentation Center and Colorado State University, 13–17.

  • Chow, V. T., , Maidment D. R. , , and Mays L. W. , 1988: Applied Hydrology. McGraw-Hill, 572 pp.

  • Clark, M. P., , and Hay L. E. , 2004: Use of medium-range numerical weather prediction model output to produce forecasts of stream-flow. J. Hydrometeor., 5 , 243262.

    • Search Google Scholar
    • Export Citation
  • Conway, B. J., , and Browning K. A. , 1988: Weather forecasting by interactive analysis of radar and satellite imagery. Philos. Trans. Roy. Soc., 324 , 299315.

    • Search Google Scholar
    • Export Citation
  • Davis, C. A., , and Emanuel K. A. , 1991: Potential vorticity diagnostics of cyclogenesis. Mon. Wea. Rev., 119 , 19291953.

  • Deidda, R., 2000: Rainfall downscaling in a space-time multifractal framework. Water Resour. Res., 36 , 17791794.

  • Deidda, R., , Benzi R. , , and Siccardi F. , 1999: Multifractal modelling of anomalous scaling laws in rainfall. Water Resour. Res., 35 , 18531867.

  • Dolciné, L., , Andrieu H. , , Sempere-Torres D. , , and Creutin D. , 2001: Flash flood forecasting with coupled precipitation model in mountainous Mediterranean basin. J. Hydrol. Eng., 6 , 19.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1993: A nonhydrostatic version of the Penn State/NCAR Mesoscale Model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 14931513.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ferraris, L., , Rudari R. , , and Siccardi F. , 2002: The uncertainty in the prediction of flash floods in the northern Mediterranean environment. J. Hydrometeor., 3 , 714727.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gómez, M., , and López R. , 1998: Rehabilitation of sewer networks in member states: Sant Boi sewer network. Research Rep., European Project SPRINT SP-98, 112 pp.

  • Grell, G. A., , Dudhia J. , , and Stauffer D. R. , 1995: A description of the fifth-generation of the Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 122 pp.

  • Groisman, P. Y., and Coauthors, 1999: Changes in the probability of heavy precipitation: Important indicators of climatic change. Climatic Change, 42 , 243283.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hardaker, P. J., , Collier C. G. , , and Pierce C. E. , 1994: The GANDOLF system: A look to the future for operational flood forecasting using satellite and radar measurements. British Hydrological Society Occasional Paper No. 5: Hydrological Uses of Weather Radar, K. Tilford, Ed., 39–63.

  • Hewitson, B. C., , and Crane R. G. , 1992: Large-scale atmospheric controls on local precipitation in tropical Mexico. Geophys. Res. Lett., 19 , 18351838.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Homar, V., , Romero R. , , Ramis C. , , and Alonso S. , 2002: Numerical study of the October 2000 torrential precipitation event over eastern Spain: Analysis of the synoptic-scale stationarity. Ann. Geophys., 20 , 20472066.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and Pan H-L. , 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jasper, K., , and Kaufmann P. , 2003: Coupled runoff simulations as validation tools for atmospheric models at the regional scale. Quart. J. Roy. Meteor. Soc., 129 , 673692.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43 , 170181.

  • Llasat, M. C., , Rigo T. , , and Barriendos M. , 2003: The “Montserrat-2000” flash flood event: A comparison with the floods that have occurred in the northeast Iberian Peninsula since the 14th century. Int. J. Climatol., 23 , 453469.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Martín-Vide, J. P., , Nierola D. , , Bateman A. , , Navarro A. , , and Velasco E. , 1999: Runoff and sediment transport in a torrential ephemeral stream of the Mediterranean coast. J. Hydrol., 225 , 118129.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Menéndez, M., 1998: Design discharge calculations and flood plain management. CEDEX, Madrid, Research Rep., European Project Flood Aware: Prevention and Forecast of Floods, 29 pp. [Available online at http://www.lyon.cemagref.fr/projects/floodaware/report/03cedex.pdf.].

  • Mlawer, E. J., , Taubman S. J. , , Brown P. D. , , Iacono M. J. , , and Clough S. A. , 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Rev., 102 , D14. 1666316682.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Murphy, A. H., 1993: What is a good forecast? An essay on the nature of goodness in weather forecasting. Wea. Forecasting, 8 , 281293.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nash, J. E., , and Sutcliffe J. V. , 1970: River flow forecasting through conceptual models. Part I: A discussion of principles. J. Hydrol., 10 , 3. 282290.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ranzi, R., , Bacchi B. , , and Grossi G. , 2003: Runoff measurements and hydrological modelling for the estimation of rainfall volumes in an alpine basin. Quart. J. Roy. Meteor. Soc., 129 , 653672.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reisner, J., , Rasmussen R. M. , , and Bruintjes R. T. , 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 Mesoscale Model. Quart. J. Roy. Meteor. Soc., 124B , 10711107.

    • Search Google Scholar
    • Export Citation
  • Romero, R., 2001: Sensitivity of a heavy rain producing western Mediterranean cyclone to embedded potential vorticity anomalies. Quart. J. Roy. Meteor. Soc., 127 , 25592597.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Romero, R., , Martín A. , , Homar V. , , Alonso S. , , and Ramis C. , 2006: Predictability of prototype flash flood events in the western Mediterranean under uncertainties of the precursor upper-level disturbance. Adv. Geosci., 7 , 5563.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Siccardi, F., 1996: Rainstorm hazards and related disasters in the western Mediterranean region. Remote Sens. Rev., 14 , 521.

  • Singh, V. P., 1997: Effect of spatial and temporal variability in rainfall and watershed characteristics on stream flow hydrograph. Hydrol. Processes, 11 , 16491669.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • USACE-HEC, 1998: HEC-HMS Hydro-logic Modeling system user’s manual. U.S. Army Corps of Engineers Hydrologic Engineering Center, Davis, CA, 188 pp. [Available online at http://www.hec.usace.army.mil.].

  • USACE-HEC, 2000: Hydrologic Modeling system HEC-HMS. Tech. reference manual, U.S. Army Corps of Engineers Hydrologic Engineering Center, Davis, CA, 157 pp. [Available online at http://www.hec.usace.army.mil.].

  • U.S. Department of Agriculture, 1986: Urban hydrology for small watersheds. Tech. Release 55, Nature Resources Conservation Service, Washington, DC, 164 pp.

  • von Storch, H., , and Zwiers F. W. , 1999: Statistical Analysis in Climate Research. Cambridge University Press, 494 pp.

  • Wilks, D., 1999: Multisite downscaling of daily rainfall with a stochastic weather generator. Climate Res., 11 , 125136.

  • Zhang, D. L., , and Anthes R. A. , 1982: A high resolution model of the planetary boundary layer: Sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteor., 21 , 15941609.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, D. L., , and Fritsch J. M. , 1986: Numerical simulation of the meso-β scale structure and evolution of the 1977 Johnstown flood. Part I: Model description and verification. J. Atmos. Sci., 43 , 19131943.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

(a) Geographical location of the IBC where the Montserrat flash-flood event was produced. Several catchments (shaded) and locations affected by the episode are indicated. (b) The Catalan topography with a depiction of the main mountainous systems and rivers (Montserrat Mountain and Llobregat River are indicated).

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 2.
Fig. 2.

Rain gauge distribution in the IBC from the SAIH. It includes a total of 126 automatic rainfall stations distributed over an area of 16 000 km2.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 3.
Fig. 3.

Digital terrain model of Llobregat River basin. It has a cell size of 50 m and displays the basin division (numbered), tributaries, stream gauges (circles), and reservoirs (triangles) mentioned in the text.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 4.
Fig. 4.

(a) CAPPI reflectivity at 1.2-km altitude recorded by the Barcelona radar at 0400 LT 10 Jun 2000. (b) SAIH-derived analysis of accumulated rainfall over the IBC during the Montserrat episode.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 5.
Fig. 5.

Observed, SAIH rain gauge driven, and MM5–NCEP simulation driven runoff discharge at (a) Súria, (b) Sant Sadurní, (c) Castellbell, (d) Abrera, and (e) Sant Joan Despí.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 6.
Fig. 6.

SAIH rain gauge driven runoff discharge at Sant Joan Despí for the different (a) spatial and (b) temporal discretizations.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 7.
Fig. 7.

Configuration of the four computational domains used for the MM5 numerical simulations (horizontal resolutions are 54, 18, 6, and 2 km, respectively) and MM5–NCEP simulation initial state, showing geopotential height at 500 hPa (continuous line, in gpm), temperature at 500 hPa (dashed line, in °C), and isentropic PV on the 330-K surface (shaded, according to scale) at 0000 UTC 9 Jun 2000.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 8.
Fig. 8.

Spatial distribution of accumulated rainfall during the Montserrat event in the Llobregat basin, from (a) SAIH rain gauges, (b) MM5–NCEP simulation, (c) MM5–NCEP–4D simulation, and (d) MM5–ECMWF simulation. Contour interval (CI) is 20 mm starting at 20 mm.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 9.
Fig. 9.

Accumulated volume during the Montserrat event, per subcatchment of the Llobregat basin, from SAIH rain gauge, MM5–NCEP, MM5–NCEP–4D, and MM5–ECMWF simulations. See Fig. 3 for subcatchment numbering.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 10.
Fig. 10.

Temporal sequence, at l-h time steps, of accumulated volume in the Llobregat basin during the Montserrat event, from SAIH rain gauge, MM5–NCEP, MM5–NCEP–4D, and MM5–ECMWF simulations.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 11.
Fig. 11.

Spatial distribution of accumulated rainfall during the Montserrat event in the Llobregat basin, from (a) −5% PV, (b) +5% PV, (c) WEST, and (d) EAST simulations. CI is 20 mm starting at 20 mm.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 12.
Fig. 12.

Accumulated volume during the Montserrat event, per subcatchment of the Llobregat basin, from (a) SAIH rain gauge, −5% PV, and +5% PV simulations, and (b) SAIH rain gauge, WEST, and EAST simulations.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 13.
Fig. 13.

Temporal sequence, at l-h time steps, of accumulated volume in the Llobregat basin during the Montserrat event, from SAIH rain gauge, −5% PV, +5% PV, WEST, and EAST simulations.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Fig. 14.
Fig. 14.

Observed, −5% PV simulation driven, +5% PV simulation driven, WEST simulation driven, and EAST simulation driven runoff discharge at (a) Castellbell, (b) Abrera, and (c) Sant Joan Despí.

Citation: Journal of Hydrometeorology 8, 3; 10.1175/JHM577.1

Table 1.

Summary of the episodes used for the calibration of the hydrologic model. Note that the observed flow at the basin outlet in Sant Joan Despi is not available for some of the cases.

Table 1.
Table 2.

NSE efficiency criterion and percentage of error in volume (%EV) for the calibration episodes at the stream gauges indicated in Table 1. Three different basin configurations [21, 39, and 60 subbasins (sb)] and hourly accumulated rainfall are used.

Table 2.
Table 3.

NSE efficiency criterion and percentage of error in volume (%EV) for the Montserrat evaluation event. The SAIH rain gauge driven simulations are carried out with three different basin segmentations (21, 39, and 60 subbasins) at the five stream gauges indicated. Hourly accumulated rainfall is used in all cases.

Table 3.
Table 4.

NSE efficiency criterion and percentage of error in volume (%EV) for the calibration episodes at the stream gauges indicated in Table 1. Three different temporal discretizations (30 min, 1 h, and 3 h) and 39 subbasins segmentation are used.

Table 4.
Table 5.

NSE efficiency criterion and percentage of error in volume (%EV) for the Montserrat evaluation event. The SAIH rain gauge driven simulations are carried out with three different time-scale discretizations (30 min, 1 h, and 3 h) at the five stream gauges indicated. Thirty-nine-subbasin segmentation is used in all cases.

Table 5.
Table 6.

Curve numbers, initial abstractions, and times of concentration for the selected basin configuration (displayed in Fig. 3). Tc (h) is time of concentration in hours.

Table 6.
Table 7.

Muskingum parameters for the selected basin configuration. Numeration of the river reaches follows the upstream direction (see Fig. 3).

Table 7.
Table 8.

NSE efficiency criterion and percentage of error in volume (%EV) at the five stream gauges for the SAIH rain gauge driven and MM5–NCEP driven runoff simulations with three (NCEP) and four domains (NCEP–4D).

Table 8.
Table 9.

NSE efficiency criterion and RMSE (hm3) of the spatial and temporal rainfall volume distributions yielded by the set of mesoscale numerical simulations.

Table 9.
Table 10.

Total precipitated volume (hm3) in Llobregat basin, and discharged volume (hm3) at Sant Joan Despí outlet, from SAIH rain gauges and the set of mesoscale numerical simulations. Observed discharged volume by the SAIH stream gauge was 73.2 hm3.

Table 10.
Table 11.

Maximum l-h accumulated volume for the whole Llobregat basin (hm3) and its corresponding local time (on 10 Jun 2000), from SAIH rain gauges and the set of mesoscale numerical simulations.

Table 11.
Table 12.

NSE efficiency criterion and percentage of error in volume (% EV) at the five stream gauges, for the set of MM5-perturbed driven runoff simulations.

Table 12.
Save