EGU Leonardo Conference Series on the Hydrological Cycle

FLOODS IN 3D: PROCESSES, PATTERNS, PREDICTIONS

EGU Leonardo Conference 23-25 November 2011 Bratislava, Slovakia

Edited by: J. Szolgay, M. Danáčová, K. Hlavčová, S. Kohnová, V. Pišteková Published by: Slovak University of Technology in Bratislava

© Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Radlinského 11, 813 68 Bratislava, Slovakia

Number of prints: 120 copies Number of pages: 110

ISBN 978-80-227-3596-4

ii

EGU Leonardo Topical Conference Series on the Hydrological Cycle

FLOODS IN 3D: PROCESSES, PATTERNS, PREDICTIONS

ORGANISED BY: Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology in Bratislava.

INTERNATIONAL SCIENTIFIC COMMITTEE Alberto Montanari Università di Bologna, Bologna, Italy Attilio Castellarin Università di Bologna, Bologna, Italy Bob Moore Centre for Ecology & Hydrology, Wallingford, Oxfordshire, UK Bruno Merz The Helmholtz Research Centre for Geosciences GFZ, Potsdam, Germany G. di Baldassarre Institute for Water Education, Delft, The Netherlands G.T. Aronica University of Messina, Messina, Italy Henrik Madsen DHI, Hørsholm, Denmark Marco Borga University of Padua, Padova, Italy Pierluigi Claps Politecnico di Torino, Torino, Italy Ralf Merz The Helmholtz Centre for Environmental Research, Leipzig, Germany Roger Moussa Institut National de la Recherche Agronomique, Montpellier, France Thomas Kjeldsen Centre for Ecology & Hydrology, Wallingford, Oxfordshire, UK

CONFERENCE CHAIR Jan Szolgay Slovak University of Technology in Bratislava, Bratislava, Slovakia

LOCAL ORGANIZING COMITTEE AT FACULTY OF CIVIL ENGINEERING, SLOVAK UNIVERSITY OF TECHNOLOGY Kamila Hlavčová Silvia Kohnová Michaela Danáčová Jana Daneková Katarína Juričeková Silvia Stasselová Viera Pišteková Gabriela Babiaková

Editorial This book of abstracts contains 100 contributions submitted by more than 90 registered authors from 19 countries. The authors take full responsibility for all content.

iii

iv

PREFACE

The frequency of disastrous floods, number of fatalities, as well as the economic damage, caused by floods is globally increasing. In this respect environmental changes (such as land- use change, inappropriate river management, and climate variability) have raised increasing concern. Defining reliable and robust flood-risk assessment strategies for emerging and developing countries is also a pressing need because of the lack of data and inappropriate land-use and developments. For these reasons, predicting current and future flood risks continues to be a major challenge for society and science when considering strategies for mitigating and coping with the impact of flooding. As natural phenomena caused by meteorological factors, floods are governed by various landscape properties and modified by human influence. Successful strategies for living with and coping with floods must therefore be based on a sound understanding of the main generating processes of extreme flows, including their spatial and temporal patterns. Incorporating this kind of understanding into predictive models is needed for the reduction of flood risks, and for addressing the potential impacts of global changes on hydrological extremes, including climate and land-use changes. The 2011 Leonardo Conference therefore had the ambition of looking at Floods in 3D: • Processes • Patterns • Predictions To assess and compare developments and impacts, a better understanding of flood processes is essential. This is not an easy task as human activities have disrupted the natural hydrological regimes. Nowadays most river basins cannot be considered pristine. The scientific literature clearly points to the importance of detecting and understanding of hydrological change, especially through improving process-understanding and separating of anthropogenic from natural variations, possibly avoiding an overemphasis on trends. Spatial patterns of flood processes exist on different scales, from small-scale patterns resulting from different runoff production mechanisms operating within a single catchment, to large-scale patterns, caused by the regional variability in hydrometeorology. There has been an increasing need to assess uncertainties when studying spatial and temporal patterns of floods and analysing their possible impact on flood management. The analysis of spatial patterns is central to any regional flood frequency analysis procedure. The analysis of temporal patterns of floods is fundamental to the understanding and assessment of variations in flood frequency. The complex nature of flood risk challenges has established risk assessment methodologies along with their modelling components, such as hydrological and hydraulic simulation methods. Predicting current and future flood risk continues to be a major challenge for climatology, hydrology and water resources engineering. While we have a good qualitative understanding of factors which may lead to changes in flood frequency and flood risk (such as climatic change, land-use effects and human interference in river channels and on flood plains), our quantitative understanding of changes in flood hazard and flood risk is still in need of improvement. These three dimensions of floods were addressed at the 2011 EGU Leonardo Conference, which offered a forum within which the most recent advances for investigating flood processes, patterns and predictions were debated.

v

vi

TABLE OF CONTENTS LEONARDO LECTURE

LOOKING AT FLOODS IN 3D - PROCESSES, PATTERNS, PREDICTIONS: Blöschl, G...... 1

Session: PROCESSES

WHY IT CAN BE WORTH USING A DIFFUSIVE MODEL INSTEAD OF A FULLY DYNAMIC ONE: Aricò, C. - Sinagra, M. - Tucciarelli, T...... 5

INFLUENCE OF MICROTOPOGRAHY FOR "UNCERTAIN" FLOOD HAZARD ESTIMATION IN URBAN AREAS: Aronica, G.T. - Neal, J. - Candela, A. - Bates, P.D...... 6

APPLICABILITY OF THE NRCS CURVE NUMBER METHOD TO SMALL LOWLAND CATCHMENT IN POLAND: Banasik, K. - Woodward, D...... 7

HYDROMETEOROLOGICAL, HYDROLOGICAL AND HYDRAULIC ANALYSIS FOR THE NOV 1, 2010 FLOOD EVENT IN THE VENETO REGION OF ITALY: Borga, M. - Zoccatelli, D. - Marra, F. - Penna, D. - Marchi, L. - Frank, E...... 8

(FLASH) FLOODS ON 27 AUGUST 2010 IN LOWLAND CATCHMENTS IN THE NETHERLANDS AND GERMANY: Brauer, C.C. - Teuling, A.J. - Overeem, A. - Van der Velde, Y. - Hazenberg, P. - Warmerdam, P.M.M - Uijlenhoet, R. - Hobbelt, L.G...... 9

ASSIMILATION OF ASCAT, AMSR-E AND ECMWF DERIVED SOIL MOISTURE PRODUCT FOR FLOOD FORECASTING: Brocca, L. - Melone, F. - Moramarco, T. - Zucco, G. - Wagner, W. - Hasenauer, S. - De Rosnay, P. - Albergel, C. - Matgen, P. - Martin, C. .... 10

DEFINITION OF SOIL MOISTURE INITIAL CONDITION INTO A DISTRIBUTED HYDROLOGICAL MODEL THROUGH LAND SURFACE TEMPERATURE FROM REMOTE SENSING IN THE UPPER YANGTZE RIVER BASIN: Corbari, C. - Mancini, M. - Li, J. - Bob Su ...... 11

CASE STUDY OF USING HYDROLOGIC FLOW ROUTING FOR ESTIMATING THE RELATIONSHIP BETWEEN THE TRAVEL TIME OF FLOOD PEAKS AND DISCHARGE: Danáčová, M. - Šúrek, P. - Szolgay, J...... 12

ESTIMATION OF FUTURE FLOOD RISK FOR FLOOD WARNING SYSTEMS IN TORYSA RIVER BASIN: Daneková, J. - Výleta, R. - Hlavčová, K...... 13

EVALUATING THE PERFORMANCE OF METEOROLOGICAL FORECASTS AND RE- ANALYSIS USING GLOBAL RIVER ROUTING: Dutra, E. - Dai Yamazaki - Wetterhall, F. - Pappenberger, F. - Balsamo, G...... 14

A NEW MODELLING TOOL FOR DETERMINING THE CATCHMENT HYDROLOGICAL RESPONSE (CASE STUDY: ISONZO RIVER): Ferri, M. - Monego, M. - Norbiato, D. - Bauruffi, F...... 15

ANALYSES OF THE DEPENDENCE STRUCTURE OF SIMULATED FLOOD PEAK- VOLUME DATA USING COPULAS: Gaál, L. – Bacigál, T. – Viglione, A. – Szolgay, J...... 16

APPLICATION OF BAYESIAN MCMC METHODS FOR INCLUSION OF HISTORICAL FLOODS IN A STATISTICAL ESTIMATION OF DESIGN MAXIMUM FLOODS: Gaál, L. - Kohnová, S. - Szolgay, J...... 17

vii

WHAT HAS BEEN LEARNED FROM THE POST FLASH FLOOD SURVEYS RECENTLY CONDUCTED IN EUROPE?: Gaume, E. - Borga, M. - Marchi, L...... 18

RETROSPECTIVE CASE STUDIES OF FLOOD SIMULATIONS TO ADDRESS THE SOCIOPOLITICAL ISSUES: Hoes, O. - Tariq, M.A.U.R ...... 19

TOWARDS ADVANCED SOLUTIONS FOR THE CALIBRATION OF HEAVY METAL DISPERSION CONTAMINATION MODELLING IN FLOODPLAIN: Hostache, R. - Matgen, P. - Hissler, C. - Kies, T. - Tosheva, Z. - Stille, P. - David, E. - Bates, P...... 20

OPERATIVE EVALUATION OF PEAK OUTFLOW DURING FLASH FLOOD USING ARTIFICIAL INTELLIGENCE METHODS: Janál, P. - Stary, M...... 21

URBAN FOOTPRINTS ON CATCHMENT STORM RUNOFF: Kjeldsen, T.R. - Miller, J.D. - Packman, J.C...... 22

ESTIMATING CATCHMENT SCALE RETENTION CAPACITY FROM SCATTEROMETER SOIL MOISTURE DATA: Komma, J. - Merz, R. - Blöschl, G. - Wagner, W...... 23

3-DIMENSIONAL MODELING OF RUNOFF GENERATION OF A MOUNTAINOUS RIVER: Kuchment, L.S. - Demidov, V.N...... 24

ON THE VALUE OF MULTIOBJECTIVE OPTIMISATION FOR PROFITING FROM SOFT INFORMATION IN UNGAUGED BASINS: Lombardi, L. - Montosi, E. - Toth, E. - Castellarin, A. - Montanari, A...... 25

RECONSTRUCTION OF THE HYDRAULIC CONDITIONS OF THE 1951 INUNDATION OF THE POLESINE REGION, ITALY: Masoero, A. - Claps, P. - Di Baldassarre, G. - Asselman, N...... 26

2D HYDRODYNAMIC FLOOD MODELLING IN A GRAVEL BED BRAIDED RIVER: Molnar, P.- Boehringer, D. - Wanner, P. - Burlando, P...... 27

EVALUATION AND MODELLING OF THE IMPACT OF LAND USE ON DISCHARGE, NITRATE AND SUSPENDED SOLIDS CONCENTRATION BY FLOOD EVENTS IN SMALL CATCHMENTS: Moravcova, J. - Pavlicek, T. - Ondr, P. - Koupilova, M. - Kvitek, T. .... 28

ON THE USE OF THE DIFFUSIVE WAVE MODEL FOR MODELLING EXTREME FLOOD EVENTS IN THE FLOODPLAIN: NATURAL vs ANTHROPOGENIC CATCHMENTS: Moussa, R. - Ghesquière, J. - Bocquillon, C...... 29

A COMPARISON OF DIFFERENT APPROACHES TO DESIGN FLOOD ESTIMATION IN ALPINE CATCHMENTS: Rogger, M. - Kohl, B. - Merz, R. - Viglione, A. - Blöschl, G. .... 30

WOODY DEBRIS TRANSPORT DURING FLOODS: 2D HYDRODYNAMIC MODELLING APPROACH: Ruiz Villanueva, V. - Bladé Castellet, E. - Sánchez Juny, M. - Bodoque del Pozo, J.M. - Díez Herrero, A...... 31

DISCHARGE HYDROGRAPH ESTIMATION USING 2D DIFFUSIVE MODEL AND SYNCRONOUS WATER LEVEL MEASUREMENTS: Sinagra, M. - Nasello, C. - Moramarco, T. - Tucciarelli, T...... 32

SNOWMELT RUNOFF PREDICTION IN THE UPPER EUPHRATES BASIN, TURKEY: Sorman, A.A. - Gozel, E. - Sensoy, A...... 33

viii

INFORMATION EXPANSION FOR UNDERSTANDING TRANS-BASIN FLOODS: Uhlemann, S. - Thieken, A. - Merz, B...... 34

FLOOD INUNDATION MODELLING IN LARGE RIVERS UNDER UNCERTAINTY USING GLOBALLY AND FREELY AVAILABLE REMOTE SENSING DATA: Yan, K. - Di Baldassarre, G. - Solomatine, D. - Dottori, F...... 35

Session: PATTERNS

DESIGN FLOOD GENERATION FOR LOWLAND RIVER SECTIONS ACCOUNTING THE UNCERTAINTY GENERATED BY SUPERPOSITION AND COINCIDENCE OF FLOODS: Bálint, G. - Lipták, G. - Szilágyi, J. - Zempléni, A...... 39

INDIRECT ESTIMATION OF DESIGN VALUES OF HEAVY RAINFALL IN THE ALPINE REGION OF SLOVAKIA: A CASE STUDY: Bara, M. - Látečková, J. - Kohnová, S. - Gaál, L. - Szolgay, J. - Hlavčová, K...... 40

FLOOD POTENTIAL IN SMALL BASINS OF SLOVAKIA: Barka, I. - Minár, J. - Trizna, M. - Bonk, R...... 41

THE SURFACE WATER OCEAN TOPOGRAPHY SATELLITE MISSION: PROSPECTS FOR FLOOD SCIENCE: Bates, P...... 42

EUROPEAN FLOOD DATA AVAILABILITY AND PARENT DISTRIBUTION OF ANNUAL FLOODS: Castellarin, A. - Kohnova, S. - ES0901 – WG2 Team ...... 43

ACCOUNTING FOR EXTREME FLOODS OCCURRED AT UNGAUGED SITES IN REGIONAL FLOOD FREQUENCY STUDIES: Cong, N.C. - Payrastre, O. - Gaume, E. .... 44

INFLUENCE OF OCEANIC PHENOMENA ON FLOODS: Diez, J.J. - Esteban, M.D. - López- Gutiérrez, J.S. - Negro, V...... 45

INVESTIGATING RUNOFF GENERATION MECHANISMS IN A SWISS PRE ALPINE CATCHMENT BY SPATIAL INTERCOMPARISON: Fischer, Benjamin, M.C. - Seibert, J. .. 46

SCENARIOS OF HEAVY, FLOOD-TRIGGERING PRECIPITATION IN THE CARPATHIAN AREA, ON THE BASIS OF HIGH RESOLUTION REGIONAL CLIMATE MODELS: Gaál, L. - Beranová, R. - Kyselý, J. - Hlavčová, K...... 47

CLIMATOLOGICAL AND GEOLOGICAL CONTROLS OF REGIONAL PATTERNS OF FLOOD TIME SCALES IN AUSTRIA: Gaál, L. - Szolgay, J. - Kohnová, S. - Parajka, J. - Merz, R. - Blöschl, G...... 48

USING PROXIMITY ALONG THE RIVER NETWORK TO IMPROVE DESIGN FLOOD ESTIMATION AT UNGAUGED SITES: Ganora, D. - Laio, F. - Claps, P...... 49

IMPROVING SIMULATIONS OF INUNDATION PATTERNS BY AIRBORNE LASER SCANNING ANALYSES - A CASE STUDY IN A DANUBE WETLAND: Gschöpf C. - Vetter M. - Blöschl, G...... 50

CAN WE ATTRIBUTE FLOOD TRENDS?: Merz, B. - Delgado, J. - Hundecha, Y. - Nguyen, D. - Uhlemann, S. - Vorogushyn, S...... 51

CROSS-CALIBRATION AT REGIONAL SCALE FOR RAINFALL-RUNOFF MODELING IN UNGAUGED CATCHMENTS: Montosi, E. - Montanari , A. - Toth, E. - Parajka, J. - Blöschl, G...... 52

ix

ANTICIPATORY FLOOD RISK ASSESSMENT - CASE STUDIES FROM AUSTRIA: Neuhold, C...... 53

ELBE 3D – A SPATIO-TEMPORAL ANALYSIS OF SOIL MOISTURE PATTERNS LINKED TO EXTREME FLOODS IN THE ELBE RIVER BASIN: Nied, M, - Hundecha, Y. - Merz, B...... 54

THE GRADEX-KC AND GRADEX-ZN METHODS FOR COMPUTING MAXIMUM FLOODS WITH T-YEAR RETURN PERIOD WHERE DISCHARGE MEASUREMENT SERIES ARE INCOMPLETE: Ozga-Zieliński, B...... 55

DECADES OF FLOOD SEASONALITY ACROSS THE ALPINE-CARPATHIAN RANGE: Parajka, J. - Kohnová, S. - Szolgay, J. - Merz, R. - Hlavčová, K. - Blöschl, G...... 56

AN ANALYSIS OF CHANGE IN ALPINE ANNUAL MAXIMUM DISCHARGES: IMPLICATIONS FOR THE SELECTION OF DESIGN DISCHARGES: Pistocchi, A. - Castellarin, A...... 57

T-YEAR MAXIMUM DISCHARGES ON WATER COURSES IN SLOVAKIA: Podolinská, J. - Šipikalová, H. - Kohnová, S...... 58

FLOOD PREDICTION IN UNGAUGED CATCHMENTS: REDUCING UNCERTAINTY WITH AN OPTIMUM USE OF REGIONAL INFORMATION: Randrianasolo, A. - Ramos, M-H - Andréassian, V...... 59

STOCHASTIC MODELING OF LARGE FLOODS - SPATIAL DEPENDENCE BETWEEN LOCAL RUNOFFS: Raschke, M...... 60

A NEW QUALITATIVE FIELD METHOD TO CAPTURE SHALLOW SOIL MOISTURE PATTERNS IN WET ENVIRONMENTS: Rinderer, M. - Kollegger, A. - Fischer, B. - Seibert, J...... 61

COMPARISON AND COMBINATION OF FLOOD REGIONALIZATION METHODS IN SAXONY, GERMANY: Salinas, J.L. - Merz, R. - Laaha, G. - Walther, J. - Büttner, U. .... 62

COUPLING TOPOLOGICAL AND CANONICAL KRIGING FOR DESIGN-FLOOD PREDICTION IN UNGAUGED BASINS IN THE SOUTHEASTERN UNITED STATES: Stacey A. Archfield - Castellarin, A. - Jon O. Skøien - Julie E. Kiang ...... 63

POTENTIAL OF HIGH RESOLUTION MODELLING OF RUNOFF GENERATION PROCESSES: Steinbrich, A. - Weiler, M...... 64

A THREE STEP APPROACH TO ANALYSIS OF FLOOD HAZARDS IN COASTAL BASINS: CASE SAR-ULLA RIVER, GALICIA-SPAIN: Vegas, E. - García, E. - Álvarez, C. - Listen, M...... 65

EXTREME RAINSTORMS: TESTING REGIONAL ENVELOPE CURVES AGAINST STOCHASTICALLY GENERATED EVENTS: Viglione, A. - Castellarin, A. - Rogger, M. - Merz, R. - Blöschl, G...... 66

MECHANISMS OF CLIMATE CHANGE EFFECTS ON FLOODS: Viglione, A. - Salinas, J.L. - Merz, R. - Blöschl, G...... 67

BIVARIATE FLOOD ESTIMATION BY DIFFERENT THRESHOLD MODELS: Zempléni, A. - Bálint, G...... 68

x

ON THE RELATIVE ROLES OF HILLSLOPE PROCESSES AND RIVER NETWORK IN THE HYDROLOGIC RESPONSE TO SPACE-TIME VARIABLE RAINFALL FIELDS: Zoccatelli, D. - Borga, M...... 69

Session: PREDICTIONS

TOWARDS A SEAMLESS FLOOD EARLY WARNING SYSTEM: FROM FLASH FLOODS TO GLOBAL FORECASTING: Alfieri, L. - Salamon, P. - Pappenberger, F. - Burek, P. - Krzeminski, B. - Muraro, D. - Thielen, J. - de Roo, A...... 73

THE MULTIVARIATE STATISTICAL ANALYSIS OF THE EXTREME HYDROLOGICAL EVENTS: Bačová Mitková, V...... 74

ON THE JOINT ESTIMATION OF FLOOD FREQUENCY DISTRIBUTIONS AT SEASONAL AND ANNUAL TIME SCALE: Baratti, E. - Montanari, A. - Castellarin, A. - Bezzi, A. - Di Giulio, A. - Ruzzante, E...... 75

MODELLING OF CLIMATE CHANGE IMPACTS ON FLOODS BY CONTINUOUS SIMULATION WITH UNCERTAINTY: Blazkova, S. - Beven, K. - Martinkova, M...... 76

PERSISTENCE AS A FACTOR OF FLOOD HAZARD FOR EMBANKED RIVERS: Bogdanowicz, E. - Strupczewski, W.G. - Kochanek, K...... 77

DESIGN FLOOD PROFILES ESTIMATION: FREEBOARD VERSUS UNCERTAINTY BOUNDS: Brandimarte, L. - Di Baldassarre, G...... 78

PROBABILISTIC FORECASTING OF ANTECEDENT SOIL MOISTURE CONDITION AS FLASH FLOOD PRECURSOR VARIABLES: Brigandì, G. - Aronica, G.T. - Loukas, A. - Vasiliades, L. - Papaioannou, G...... 79

FLOOD PROTECTION OF THE CITY OF LJUBLJANA AND EXPERIENCES GAINED DURING THE SEPTEMBER 2010 FLOOD: Brilly, M. - Vidmar, A. - Rusjan, S...... 80

NEW METHODS OF FLASH FLOOD FORECASTING AT THE CZECH HYDROMETEOROLOGICAL INSTITUTE: Březková, L. - Novák, P. - Jonov, M. - Kyznarová, H. - Šálek, M. - Frolík, P. - Janál, P...... 81

EFFECTS OF UNCERTAINTY IN BOUNDARY CONDITIONS ON FLOOD HAZARD ASSESSMENT: Domeneghetti, A. - Vorogushyn, S. - Castellarin, A. - Merz, B. - Brath, A...... 82

THE VALUE OF URBAN FLOOD MAPPING: APPLICATION TO THE CASE STUDY OF BARCELONNETTE (FRANCE): Dottori, F. - Di Baldassarre, G. - Alfonso, L...... 83

OPEN SOURCE GIS TOOLS FOR FLOOD HAZARD MAPPING WITHIN THE DANUBE FLOOD RISK PROJECT: Franceschi, S. - Monacelli, G. - Rigon, R...... 84

UNCERTAINTY ESTIMATION OF SIMULATED WATER LEVELS FOR THE MITCH FLOOD EVENT IN TEGUCIGALPA : Fuentes, D.C. - Halldin, S. - Xu, C-Y. - Lundin, L-C .... 85

INTEGRATION OF TIME SERIES MANAGEMENT SYSTEM AND DATA SERVING FOR MODELLING APPLICATIONS: Gál, R...... 86

ASSESSMENT OF PROJECTED CHANGES IN PRECIPITATION EXTREMES IN THE CZECH REPUBLIC USING A NON-STATIONARY INDEX FLOOD MODEL: Hanel, M. - Vizina, A. - Martínková, M...... 87

xi

SAFEFTY CRITERIA FOR FLOOD PROTECTION STRUCTURES IN SLOVENIA: Humar, N. - Kryžanowski, A. - Brilly, M. - Schnabl, S...... 88

SPRING FLOOD FREQUENCY ANALYSIS OF LITHUANIAN RIVERS: Kriaučiūnienė, J. - Šarauskienė, D...... 89

ESTIMATION OF FLOOD FREQUENCY CURVES IN POORLY GAUGED MEDITERRANEAN WATERSHEDS USING A DERIVED DISTRIBUTION PROCEDURE: Loukas, A. - Vasiliades, L. - Papaioannou, G. - Aronica, G.T...... 90

CLIMATE CHANGE IMPACT ASSESSMENT OF FLOOD RISK. CASE STUDY: VIDAA RIVER BASIN: Madsen, H. - Larsen, J. - Madsen, M.N...... 91

FLOOD RISK MAPPING UNDER UNCERTAINTY: APPLICATION TO SUNGAI JOHOR BASIN MALAYSIA: Md. Ali, A. – Di Baldassare, G. – Solomatine, D...... 92

FLOOD IN TOWN – DECISION SUPPORT TOOL (VESNA): Mišík, M. - Kučera, M. - Fridrich Tegelhoffová, M. - Mrnčo, I...... 93

UNCERTAINTY IN 1D HYDRAULIC MODELLING CAUSED BY PARAMETERIZATION AND INFLOW INACCURACY: Mukolwe, M.M. - Di Baldassarre, G. - Werner, M. - Solomatine, D...... 94

UNCERTAINTY ASSESSMENT OF STREAMFLOW SIMULATION ON UNGAUGED CATCHMENTS USING A DISTRIBUTED MODEL AND THE KALMAN FILTER BASED MISP ALGORITHM: Múnera, J.C. - Francés, F. - Todini, E...... 95

TOWARDS SPATIALLY DISTRIBUTED FLOOD FORECASTS IN FLASH FLOOD PRONE AREAS: RESULTS OF A FIRST TEST IN THE SOUTH OF FRANCE: Naulin, J-P - Payrastre, O. - Gaume, E. - Delrieu, G. - Arnaud, P. - Lutoff, C. - Vincendon, B...... 96

PROBABILISTIC FLOOD RISK MAPPING INCLUDING SPATIAL DEPENDENCE: Neal, J. - Keef, C. - Bates, P. - Beven, K. - Leedal, D...... 97

ENSEMBLE PREDICTION OF FLOODS – PROCESS CONTROLS: Nester, T. - Blöschl, G...... 98

FLOOD HAZARD UNDER CLIMATE CHANGE IN THE MEKONG DELTA: Nguyen, V.D. - Delgado, J.M. - Merz, B. - Bárdossy, A. - Apel, H...... 99

FLOOD PRONE AREAS MAPPING THROUGH GIS AND MULTI-CRITERIA ANALYSIS: Papaioannou, G. - Loukas, A. - Vasiliades, L. - Aronica, G.T...... 100

VISUALISING PROBABILISTIC FLOOD FORECAST INFORMATION: EXPERT PREFERENCES AND PERCEPTIONS OF BEST PRACTICE IN UNCERTAINTY COMMUNICATION: Pappenberger, F. - Stephens, E. - Thielen, J. - Salamon, P. - Demeritt, D. - van Andel, S.J. - Wetterhall, F. - Alfieri, L...... 101

DEVELOPMENT OF DECISION SUPPORT SYSTEMS FOR FLASH FLOOD FORECASTING IN SMALL CATCHMENTS: Poma, S. - Brigandì, G. - Aronica, G.T. - Candela, A...... 102

AN ENTROPY APPROACH FOR THE OPTIMIZATION OF RIVER CROSS-SECTIONAL SPACING: Ridolfi, E. – Alfonso,L. - Di Baldassarre, G. – Dottori, F. – Russo, F. – Napolitano, F...... 103

xii

FLOOD-RISK MAPPING – A NEW INSIGHT INTO A MAJOR ENVIRONMENTAL PROBLEM: Romanowicz, R. - Kiczko, A. - Osuch, M...... 104

BAYESIAN UNCERTAINTY ASSESSMENT OF FLOOD PREDICTIONS IN UNGAUGED URBAN BASIS FOR CONCEPTUAL RAINFALL-RUNOFF MODELS: Sikorska A. - Scheidegger A. - Banasik K. - Rieckermann J...... 105

DEVELOPMENT OF EXPECTED ANNUAL DAMAGE CURVES AND MAPS AS A BASIC TOOL FOR THE RISK-BASED DESIGNING OF STRUCTURAL AND NON-STRUCTURAL MEASURES: Tariq, M.A.U.R. - Van de Giesen, N.C. - Hoes, O...... 106

FLOOD FORECASTING WITH CONSTRAINED DOWNSTREAM CONDITION IN YUVACIK BASIN, TURKEY: Uysal, G. - Yavuz, Ö. – Şensoy, A. - Şorman, A.A. - Akgün, T. - Gezgin, T...... 107

UNCERTAINTY ANALYSIS OF HEAVY RAINFALL EVENTS REPRESENTED BY A HYDROMETEOROLOGICAL MODEL AND SATELLITE RAINFALL ALGORITHM FOR FLOOD FORECASTING: Yucel, I. - Mert, I...... 108

UNCERTAINTY ASSESSMENT OF CLIMATE CHANGE ADAPTATION OPTIONS IN URBAN FLASH FLOODS: Zhou, Q. - Arnbjerg-Nielsen, K...... 109

xiii

xiv

LEONARDO LECTURE

LOOKING AT FLOODS IN 3D - PROCESSES, PATTERNS, PREDICTIONS

Blöschl, G.

Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Vienna, Austria E-mail: [email protected]

Obtaining flood discharge values, be it for design purposes or operational flood management, has traditionally focused on the estimation problem. In flood frequency statistics we have developed ever more refined techniques for exploiting the information of the flood sample, and in flood forecasting we have focused on sophisticated updating procedures based on control theory. In this presentation I will argue that a fresh look is needed at the flood estimation problem from three perspectives – processes, patterns, predictions - to make better use of the wealth of hydrological knowledge gained in the past century. From a process perspective, understanding the causal mechanisms of flood generation at the catchment scale is the key. Analysing spatial patterns of dynamic flood characteristics helps learn from the rich diversity of flood processes across the landscape. In prediction it is essential to combine the causal, spatial and temporal information that may be available on floods in the catchment of interest. I will present a number of examples from Austria to illustrate the rich diversity of flood processes that are often site specific and difficult to capture by formal methods. On the basis of these examples, and the various sources of information, I will illustrate that hydrological reasoning and Bayesian methods give a more complete representation of flood processes at a given site than the traditional estimation methods alone. I propose the term ‘‘flood frequency hydrology,’’ as opposed to flood frequency statistics, to reflect the focus on hydrological processes and hydrological reasoning. By accounting for processes and patterns in the predictions, the single dimensional estimation problem is transformed into a three dimensional hydrological problem.

1

LEONARDO LECTURE

2

EGU Leonardo Conference Series on the Hydrological Cycle

SESSION: PROCESSES

Conveners: G. di Baldassare, M. Borga, A. Montanari, R. Moussa

This session focused on a number of gaps in scientific understanding and modelling of flood processes and in predicting flood severity, such as: • Understanding of coupled human and natural flood systems; • Comparison (and integration) of hydraulic, hydrological and stochastic flood models; • Development of techniques for the assimilation of ground and remote sensing data.

3

4

SESSION: PROCESSES

WHY IT CAN BE WORTH USING A DIFFUSIVE MODEL INSTEAD OF A FULLY DYNAMIC ONE

Aricò, C., Sinagra, M. and Tucciarelli, T.

Department of Civil, Environmental and Aerospace Engineering, University of Palermo, Viale delle Scienze 90128 Palermo, Italy Email: [email protected], [email protected], [email protected]

The paper is aimed to revise the traditional idea that the choice of a diffusive model instead of the fully dynamic one, for the solution of the continuity and momentum equations according to the shallow water hypothesis, is only a question of computational effort. Additional, often prevailing motivations can be that: 1) in the Diffusive Shallow Water (DSW) models only one boundary condition (b. c.) is required at each boundary point, where the appropriate number of b. c. in the Fully dynamic Shallow Water (FSW) models can be zero, two or three depending on the local Froude number, 2) most important, the sensitivity of the computed water depth to the topographic error is much higher in the FSW models than in the DSW ones. Motivation (1) is often based on data limitation. For example, the availability of data regarding supercritical flows entering the upstream domain boundary is often missing and in the FSW solution the normal (i.e. uniform) flow condition is usually adopted to relate water depth to discharge. Motivation (2) can be the most important one. Guinot and Cappallaere have recently analyzed the sensitivity of a FSW 2D model with respect to the parameter errors, where parameters are the topographic elevation, the Manning coefficient and the bed slope. They have shown that, in the very simple case of frictionless, horizontal bed with uniform steady-state flow, the sensitivity of the water depth with respect to the bed elevation error can be computed as the solution of a Laplace equation, where the source term is inversely proportional to the local Froude number. Because a diffusive model can be thought as a fully dynamic one where the gravity force goes to infinity, the sensitivity of the DSW model is always smaller than the sensitivity of the FSW one. The same conclusions can be obtained for the 1D case by observing the behaviour of the water depth when a topographic change is given for a constant energy value. Several numerical tests are presented, where the real geometry of the channel bed is simplified according to different approximation levels. Comparing the results of the diffusive MAST-2D model with the results of the fully dynamic MAST-2C and BreZo models, it is shown that the results of the FSW models are superior to the results of the DSW models only for very detailed geometry reconstructions and mesh densities.

5

SESSION: PROCESSES

INFLUENCE OF MICROTOPOGRAHY FOR “UNCERTAIN” FLOOD HAZARD ESTIMATION IN URBAN AREAS

Aronica, G.T.1, Neal, J.2, Candela, A.3 and Bates, P.D.2

1 Dipartimento di Ingegneria Civile, University of Messina, Italy 2 School of Geographical Sciences, University of Bristol, UK 3 Dipartimento di Ingegneria Civile, Ambientale e Aerospaziale, University of Palermo, Italy Email: [email protected]

Floods are considered the most frequent natural disaster worldwide and may have serious socio economic impacts in a community. Properties worth several billion euros are located on floodplains in Europe, and many of these are classified as being at risk. The majority of these properties are in urbanised areas. This is a problem as not only are properties located next to river channels which may flood, but also because the rapid increase in the area covered by impervious surfaces has increased runoff and led to enhanced local flooding as a consequence of excess rainfall (pluvial flooding). Such floods have high spatial and temporal variability and are strongly controlled by micro-scale features (depressions, walls, culverts) within the catchment such that they are extremely difficult to predict. The above-described phenomena make the estimation of flood risk in urban areas complex, since it must deal with both main river flooding and flooding due to excess rainfall. Local failures in inlets or culverts are often observed as a consequence of precipitation events with even lower intensities than the design rainfall. Micro-topography effects are usually responsible for those local failures, and their potential in terms of local flooding can be investigated by taking a proper degree of topographic details into consideration. Here, we test the ability of two computationally efficient hydraulic models: a two-dimensional hyperbolic finite element model and a recently developed version of the LISFLOOD-FP model which solves a reduced form of the full shallow water equations in a highly efficient manner in analysing the effects of micro-topography for producing local flooding with significant water depth and velocity in zones of flow concentration. These models are applied to a case study in the city of Glasgow.

6

SESSION: PROCESSES

APPLICABILITY OF THE NRCS CURVE NUMBER METHOD TO SMALL LOWLAND CATCHMENT IN POLAND

Banasik, K.1 and Woodward, D.2

1 Warsaw University of Life Sciences - SGGW, Department of Water Engineering, Sedimentation Laboratory, ul. Nowoursynowska 166, PL-02-787 Warsaw, Poland 2 Natural Resources Conservation Service USDA, Washington D.C. E-mail: [email protected], mailto:[email protected]

The NRCS Curve Number (known also as SCS-CN) method is well known as a tool in predicting flood runoff from small ungauged catchment. The traditional way of determination the CNs, based on soil characteristics, land use and hydrological conditions, seemed to have tendency to overpredict the floods in some cases. A few year rainfall-runoff data, collected in a small (A=82.4 km^2 ), lowland, agricultural catchment in Central Poland, were used to determine runoff Curve Number and to check a tendency of changing. The date was collected using an electronic system with a time step of recording of 10 minutes. The analysis concluded, that using CN value according to the procedure described in USDA- SCS Handbook one receives representative value for estimating runoff from high rainfall depths. This has been confirmed by applying “asymptotic approach” for estimating the watershed curve number from the rainfall-runoff data. Furthermore, the analysis indicated that CN, estimated from mean retention parameter S of recorded events with rainfall depth higher than initial abstraction, is also approaching the theoretical CN. The observed CN, ranging from 52.3 to 95.5, declines with increasing storm size, which has been classified as a standard response of watershed (Hawkins 1992). The investigation demonstrated also changeability of the CN during the year, with much lower values during the growing season.

7

SESSION: PROCESSES

HYDROMETEOROLOGICAL, HYDROLOGICAL AND HYDRAULIC ANALYSIS FOR THE NOV 1, 2010 FLOOD EVENT IN THE VENETO REGION OF ITALY

Borga, M. 1, Zoccatelli, D.1, Marra, F.1, Penna, D.1, Marchi, L.2 and Frank, E.3

1 Department of Land and Agroforest Environment, University of Padova, Legnaro, Italy 2 CNR IRPI - National Research Council of Italy, Research Institute for Hydrogeological Protection, Padova, Italy 3 BETA Studio s.r.l., Ponte San Nicolò (PD) – Italy Email: [email protected], [email protected], [email protected]

Hydrometeorological analyses of rainfall and flood response are presented for the flood event occurred over the Veneto Region in November 2010. The event – lasting three days - produced localized storm total rainfall accumulation exceeding 500 mm. Hydrometeorological and hydrologic studies illustrate the important role of orography in controlling the flood response. Analyses of high-resolution radar rainfall fields are presented, with special emphasis on radar rainfall error structure and the spatial and temporal variations of flood- producing rainfall. Diagnostic and hydrologic model studies in a number of headwater catchments were made possible thanks to availability of data from post-event surveys executed just after the flood. The consistency of data from post-event surveys and from conventional stream gauge stations is examined. Diagnostic studies over the Posina, Leogra and Agno basins are carried out to examine the controls of orographic enhancement of precipitation, space-time rainfall distribution and land surface processes on extreme flood response. Land surface processes and the contrasting distribution of rainfall in space and time from the storm systems, mainly due to the orographic control, combine to shape the scale-dependent distribution of extreme floods in the study basins. Results are reported from a hydraulic diagnostic analysis, based on a 2D hydrodynamic model, aimed to assess the numerical description of the inundation in selected areas of the floodplains and its effect on flood attenuation.

8

SESSION: PROCESSES

(FLASH) FLOODS ON 27 AUGUST 2010 IN LOWLAND CATCHMENTS IN THE NETHERLANDS AND GERMANY

Brauer, C.C.1, Teuling, A.J.1,2, Overeem, A.1,3, Van der Velde, Y.1,4,5, Hazenberg, P.1, Warmerdam, P.M.M.1, Uijlenhoet, R.1 and Hobbelt, L.G.1

1Hydrology and Quantitative Water Management Group, Wageningen University, P.O.Box 47, 6700AA Wageningen, The Netherlands, 2formerly at: Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland 3now at: Royal Netherlands Meteorological Institute, De Bilt, The Netherlands 4Soil Physics, Ecohydrology and Ground Water Management Group, Wageningen University, Wageningen, The Netherlands 5now at: Department of Physical Geography and Quaternary Geology, Stockholm University, Stockholm, Sweden E-mail: [email protected]

On 26 August 2010 the eastern part of The Netherlands and the bordering part of Germany were struck by a series of heavy rainfall events. Over an area of 740 km2 more than 120 mm of rainfall were observed in 24 h. This extreme event resulted in local flooding of city centres, highways and agricultural fields, and considerable financial loss. We investigated the unprecedented flash flood triggered by this exceptionally heavy rainfall event in the 6.5 km2 Hupsel Brook catchment, which has been the experimental watershed employed by Wageningen University since the 1960s. This study aims to improve our understanding of the dynamics of such lowland flash floods. We present a detailed hydrometeorological analysis of this extreme event, focusing on its synoptic meteorological characteristics, its space-time rainfall dynamics as observed with rain gauges, weather radar and a microwave link, as well as the measured soil moisture, groundwater and discharge response of the catchment. We found that the response of the Hupsel Brook catchment can be divided into four phases: (1) soil moisture reservoir filling, (2) groundwater response, (3) surface depression filling and surface runoff and (4) backwater feedback. During this extreme event some thresholds became apparent that do not play a role under average conditions and are not incorporated in rainfall-runoff models. Because of the large spatial extent of the rainfall event, many brooks and rivers in the Netherlands and Germany flooded. With data from several catchments we investigated the influence of rainfall and catchment characteristics (such as slope, size and land use) on the response of discharge to rainfall.

9

SESSION: PROCESSES

ASSIMILATION OF ASCAT, AMSR-E AND ECMWF DERIVED SOIL MOISTURE PRODUCT FOR FLOOD FORECASTING

Brocca, L.1, Melone, F.1, Moramarco, T.1, Zucco, G.1, Wagner, W.2, Hasenauer, S.2, De Rosnay, P.3, Albergel, C.3, Matgen, P.4, Martin, C.5

1 Research Institute for Geo-Hydrological Protection, National Research Council, Perugia, Italy 2 Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, Vienna, Austria 3 European Centre for Medium-Range Weather Forecasts, Reading, United Kingdom 4 Public Research Centre - Gabriel Lippmann, CRP, Belvaux, Luxembourg 5 UMR -6012 ESPACE, Département de Géographie, Université de Nice-Sophia-Antipolis, Nice, France E-mail: [email protected]; [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]

In recent years, steady increases in flood damages have led to a growing interest in the development of flood forecasting systems operating in near real-time. Flood forecasting has been widely approached by using rainfall-runoff models. However, uncertainties in model structure, model parameters and input/output observations strongly impact the accuracy of the predictions. To reduce this limitation, hydrological data assimilation has become an important issue for research and applications. In particular, the assimilation of soil moisture observations has the potential to provide great benefits for runoff prediction. Nowadays, soil moisture estimates from satellite sensors and global meteorological models are becoming more readily available with a spatial and temporal resolution that is suitable for hydrological applications. On the other hand, different data assimilation techniques have been developed and proposed for the optimal use of soil moisture observations in rainfall- runoff modelling. Notwithstanding these large efforts, even after a decade of attempts to improve hydrological predictions through the process of data assimilation, genuine 'success stories' continue to be rare, arguably due to differences in the quality of data and models used in different applications. Therefore, the following aspects have still to be addressed: the improvement of the hydrological information retrieval from remotely sensed data, the choice of the optimum data assimilation method, the quantitative description of model and observation errors, the choice of assimilated data and the evaluation of data assimilation effectiveness. In this study, the soil moisture products derived from two different satellite sensors, the Advanced Scatterometer and the Advanced Microwave Scanning Radiometer, and from the ECMWF Integrated Forecasting System are employed. A multi-layer continuous and distributed rainfall-runoff model, named MISDc-ML, is used to test and compare the effectiveness of the different products for improving flood prediction. The Ensemble Kalman filter is adopted to optimally incorporate the soil moisture data into MISDc-ML. Several catchments located in different climatic regions over Europe are used as case studies. The use of different soil moisture products, model structures (i.e. variable number of soil layers) and study areas provides an in-depth understanding of the value of soil moisture data assimilation in rainfall-runoff modelling. Results reveal that the three soil moisture products can be conveniently used to improve runoff prediction. However, reliability differs according to the climatic region of the catchment under investigation. Moreover, a detailed characterization of the water movement in the soil through the structure of the rainfall-runoff model is necessary in order to have a significant impact of the assimilated data in the model itself.

10

SESSION: PROCESSES

DEFINITION OF SOIL MOISTURE INITIAL CONDITION INTO A DISTRIBUTED HYDROLOGICAL MODEL THROUGH LAND SURFACE TEMPERATURE FROM REMOTE SENSING IN THE UPPER YANGTZE RIVER BASIN

Corbari, C.1, Mancini, M.1, Li, J.2 and Bob Su3

1 Politecnico di Milano, Milano, Italy 2 China Institute of Water Resources and Hydropower Research, Beijing, China 3 University of Twente, Twente, The Netherlands E-mail: [email protected]

This work investigates the potentiality of the use of operative remote sensing land surface temperature observations (LST) as a proxy of surface soil moisture, which as well-known, constitutes the soil initial condition before any storm event in real time flood forecasting model, to improve real time flash flood forecast modeling in order to reduce and to manage flood risk. The application of distributed hydrological models, both in operative and scientific research, is limited by the difficulties to verify evaporative fluxes and soil water content at the basin scale. In fact, calibration and validation of distributed models generally depend on comparison between simulated and observed discharges at the available river cross sections, which are usually very few. The use of thermal infrared data integrated with a hydrological model seems to solve the limitation of microwave remote sensor in determining the surface soil moisture. LST from remote sensing will be used for distributed hydrological model internal validation of the mass and energy balance as a complementary method to the traditional discharge measurements. A thermodynamic model for energy and mass balance computation will be implemented and included into the distributed hydrological model FEST-EWB (Flash–flood Event–based Spatially–distributed rainfall–runoff Transformation - Energy Water Balance). The energy balance is solved looking for the representative thermodynamic equilibrium temperature defined as the land surface temperature that closes the energy balance equation. This temperature is an integral temperature, which is representative of a volume and of the heterogeneity of the thermodynamic exchanges. This scheme is applied from 2000 to 2010 in the Upper Yangtze River basin (China) with an area of about 1 000 000 km2 where few ground hydrological and meteorological data are available. Land surface temperature from MODIS and AATSR sensors will be used for the validation of the distributed hydrological model FEST-EWB.

11

SESSION: PROCESSES

CASE STUDY OF USING HYDROLOGIC FLOW ROUTING FOR ESTIMATING THE RELATIONSHIP BETWEEN THE TRAVEL TIME OF FLOOD PEAKS AND DISCHARGE

Danáčová, M., Šúrek, P. and Szolgay, J.

Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Radlinskeho 11, 813 68 Bratislava, Slovakia E-mail: [email protected], [email protected], [email protected]

The relationship between the travel time of flood peaks and discharge can be obtained directly from historical hydrographs; however this method is time-consuming and requires analysis of several dozens of such waves. To overcome this, an automatic parameter optimisation technique was developed using hydrologic flow routing model for estimating the relationship between the travel time of flood peaks and discharge. The practical applicability of the concept was tested on the Morava River between Moravský Svätý Jan and Záhorská Ves, the Laborec River between Krásny Brod, on the Torysa River between Brezovica – Sabinov and on the Hron River between Brezno – Banská Bystrica. The travel-time discharge relationship was assumed as being composed from chained successive linear segments. Both the length and number of segments was allowed to vary. This relationship was used to assess the travel time parameter in the discrete multilinear cascade model. The modelled travel-time parameter vs. discharge relationships were estimated by optimisation of the model performance with the help of a genetic algorithm and ANN. The relationships between travel-time of flood peaks and peak discharge were also studied on these river reaches based on data from several floods. The performance of the proposed method was compared against relationships obtained from empirical data, and also by genetic optimization of the flood routing model performance on a small set of floods, respectively. The optimised piecewise linear relationship fitted well the empirical data on travel-times of flood peaks and was also consistent with the findings in the literature. It was shown that the relationships detected by the genetic optimisation exhibit the same behaviour as those gained from the empirical data.

12

SESSION: PROCESSES

ESTIMATION OF FUTURE FLOOD RISK FOR FLOOD WARNING SYSTEMS IN TORYSA RIVER BASIN

Daneková, J., Výleta, R. and Hlavčová, K.

Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Radlinskeho 11, 813 68 Bratislava, Slovakia E-mail: [email protected], [email protected], [email protected]

The planning for flood-related emergencies could be supported by an estimate of the return period of a forthcoming extreme flood event. A study of the relationship between the return periods of floods and flood-causing precipitation on the Torysa River basin to the outlet at Prešov located in the eastern part of Slovakia with an area of 673.89 km2 was taken as the basis for such an estimate in this poster. A methodological framework for such an estimate, based on a combination of statistical methods with a real time estimation of the soil moisture index of the basin with the help of a lumped mathematical rainfall runoff model, was proposed. A conceptual rainfall-runoff model with a daily time step was adopted for runoff modelling. The study of the relationship between the modelled soil moisture index of river basin at the beginning of the annual maximum floods and the flood-causing precipitation totals showed no dependence between these two quantities. Using event-based flood simulations, synthetic flood waves were generated for rainfall scenarios with the rainfall- runoff model. The relationships between the return periods of the synthetic precipitation and the return periods of the simulated floods were analysed, and the effect of antecedent basin saturation index on the extremity of the floods was quantified. Critical values of the basin saturation index leading to floods with higher return periods than the return period of flood- causing precipitation were suggested.

13

SESSION: PROCESSES

EVALUATING THE PERFORMANCE OF METEOROLOGICAL FORECASTS AND RE-ANALYSIS USING GLOBAL RIVER ROUTING

Dutra, E.1, Dai Yamazaki2, Wetterhall, F.1, Pappenberger, F.1 and Balsamo, G.1

1 European Centre for Medium-Range Weather Forecasts, Reading, UK 2 Institute of Industrial Science, University of Tokyo, Tokyo, Japan E-mail: [email protected]

River discharge is a natural integrator of meteorological variables. The integration is made over a spatial domain (catchment) which is geophysically appropriate, and over time. It takes into account the correlations and covariances between several meteorological variables in a meaningful way, integrating information from a multidimensional variable space. Furthermore, river discharge observations are available and generally reliable. Therefore, river discharge is an important variable to consider in verifying forecast performance, not only from a meteorological standpoint but also for its relevance to applications. A river routing scheme describing floodplain inundation dynamics is applied in the verification of meteorological forecast and reanalysis. The scheme is forced by daily runoff fields generated by the ECMWF land surface model HTESSEL in standalone mode, or runoff fields available from the reanalysis and forecast products. The present model configuration runs globally at about 25 km resolution (although the basic unit is a catchment rather than grid-point). In an initial phase the routing scheme was tested and calibrated by carrying out simulations forced by the output of HTESSEL offline. In these simulations, HTESSEL was forced by the ERA- Interim with rescaled GPCP precipitation. The calibration was a necessary step, and was performed by perturbing two model parameters (river bank height and river width) and evaluating the river discharge on a global scale using about 1000 observations. Although the limitations, and uncertainty of the calibration processes, the new set of model parameters led to improved predictions of river discharge. Using this new parameters dataset, the routing scheme was forced by the original ERA-40 and ERA-Interim runoff fields, as well as a standalone simulation of HTESSEL forced by the original ERA-Interim dataset (LSH). The LSHGPCP simulations provide the best predictions of river discharge, as expected, while LSH, ERAI and ERA4 show some problems associated to under/over estimation of precipitation. A clear signal was also found due to changes in the land surface model that generates the runoff fields. Simulations of the routing scheme forced by the ECMWF deterministic forecasts since 1995 to present are discussed, where the different model changes can be compared in terms of predicting river discharge.

14

SESSION: PROCESSES

A NEW MODELLING TOOL FOR DETERMINING THE CATCHMENT HYDROLOGICAL RESPONSE (CASE STUDY: ISONZO RIVER)

Ferri, M., Monego, M., Norbiato, D. and Bauruffi, F.

Autorità di bacino dei Fiumi dell’Alto Adriatico, Dorsoduro 3593, 30123 Venezia, Italy E-mail: [email protected]

The River Basin Authority Alto Adriatico is responsible for the management of rivers flowing into the Northern Adriatic Sea. With the purpose of contributing to the general objective of theme “Advanced models and methods to improve the understanding and mitigation of hydrological extreme events”, the paper contributes the experiences that are becoming part of the European project KULTURisk, that aims at developing a culture of risk prevention by means of a comprehensive demonstration of the benefits of prevention measures. KULTURisk first focuses on water-related hazards as the likelihood and adverse impacts of water-related catastrophes might increase in the near future because of land use and/or climate changes. To achieve the aims of the project an innovative catchment scale hydrological model has been implemented. The basin hydrological response is estimated using a semi-distributed, continuous, hydrological model based on a Geomorphoclimatic theory, relating the shape and scale of the catchment transfer function to stream network topology, channels characteristics and climate. Snow accumulation and melt are represented by using a distributed version of the UEB model, while the soil water balance is solved towards a realistic description of the temporal dynamics of soil moisture, with a physically based parameterisation which allows use of vegetation coverage, soil texture and local slope data. The model captures the essential physics of the relevant processes, rather than merely reproducing a correspondence between inputs and outputs within a finite set of observations. Predictive capabilities and robustness of the model were tested comparing simulation results with 9 years (2000-2008) of quality-controlled hourly data.

15

SESSION: PROCESSES

ANALYSIS OF THE DEPENDENCE STRUCTURE OF SIMULATED FLOOD PEAK- VOLUME DATA USING COPULAS

Gaál, L.1, Bacigál, T.1, Viglione, A.2, Szolgay, J.1

1 Faculty of Civil Engineering, Slovak University of Technology, Bratislava, Slovakia 2 Institute for Hydraulic and Water Resources Engineering, Vienna University of Technology, Vienna, Austria E-mail: [email protected], [email protected], [email protected], [email protected]

In frequency analysis of hydro-climatological extremes, one is generally interested in the extremes corresponding to low probabilities of the fitted cumulative distribution functions (CDFs). However, usually there is not enough flood data in the right tail of the empirical CDFs to derive reliable statistical inferences on the behaviour of the extremes. The scarcity of the data is even more striking in a multivariate frequency analysis of variables (such as flood peaks, volumes and durations etc.) where beside the marginal distributions of the individual variables, one also needs to consider their dependence structure. To extend the amount of information for the statistical inference, different techniques are used. For instance, temporal extension methods allow for making use of historical or paleoflood data, or spatial extension methods such as regional approaches extend the scope of the analysis by pooling data in an adequately defined pooling group or region. In this study, a different approach was adopted to increase the amount of data in the right tail of the CDFs: our methodology took the advantage of simulated flood data from rainfall-runoff modelling. The artificial flood data were then examined in terms of annual maxima of flood peaks and flood volumes, and their dependence structure was analyzed by means of copulas. In order to generate artificial runoff data (i.e. to simulate flood records of lengths of approximately 104–105 years), a two-step procedure was used. (i) First, the stochastic rainfall generator proposed by Sivapalan et al. (2005) was modified for our purpose. This model is based on the assumption of discrete rainfall events whose arrival times, durations, mean rainfall intensity and the within-storm intensity patterns are all random, and can be described by specified distributions. The mean storm rainfall intensity is disaggregated further to 15 min intensity patterns. (ii) Secondly, the simulated rainfall data entered a semi-distributed conceptual rainfall-runoff model that consisted of a snow routine, a soil moisture routine and a flow routing routine. The applicability of the proposed method was demonstrated on a selected site in Austria (Stampfangerbach Creek, Tirol). The pairs of simulated flood volumes and flood peaks were analysed in terms of their dependence structure and different families of copulas (Gumbel, Clayton, Frank, Plackett, Normal) were fitted to the observed and simulated data. The question to what extent measured data can be used to find the right copula was discussed.

16

SESSION: PROCESSES

APPLICATION OF BAYESIAN MCMC METHODS FOR INCLUSION OF HISTORICAL FLOODS IN A STATISTICAL ESTIMATION OF DESIGN MAXIMUM FLOODS

Gaál, L., Kohnová, S. and Szolgay, J.

Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Bratislava, Slovakia E-mail: [email protected], [email protected], [email protected]

The presented work deals with temporal data extension for the purposes of a local frequency analysis. The method consists in inclusion of historical hydrological events into the procedure of flood quantile estimation. The statistical modelling takes the advantage of Monte Carlo Markov chain simulations, which, within a Bayesian framework, represent a flexible tool for an inclusion of hydrological information of different origin. The practical aspects of Bayesian MCMC simulations are demonstrated through a case study that focuses on the Myjava River at the cross-section Myjava. Two considerably high flood peaks that appeared in the local series of observations are considered as ‘historical’ events, and we examine the results of the frequency analysis depending on whether i) these two exceptional events are fully excluded from the data sample; whether ii) the whole series of observations is analyzed or iii) the two discussed flood peaks are regarded as historical events. The study suggests that the most acceptable results of the frequency analysis (i.e., the narrowest confidence intervals of the 100-year flood quantiles) are related to the last option, which adopts the concept of inclusion of historical events.

17

SESSION: PROCESSES

WHAT HAS BEEN LEARNED FROM THE POST FLASH FLOOD SURVEYS RECENTLY CONDUCTED IN EUROPE?

Gaume, E.1, Borga, M.2 and Marchi, L.3

1 IFSTTAR, Bouguenais, France, 2 University of Padova, Italy, 3 CNR – IRPI, Padova, Italy, E-mail: [email protected], [email protected], [email protected]

Post-event survey and investigation is one way to gain experience on natural hazards. The importance of the systematisation and standardisation of such investigations and re-analysis is progressively recognised in all the geophysical sciences as shown by the growing number of scientific papers and programs on the subject. But how to proceed in the case of flash floods often affecting ungauged basins, what type of data should be collected for what type of analyses and which particular issues to explore? To give a first answer to these questions, a methodology for post-flash flood field investigations has been developed under the EC FLOODsite project and tested under the EC HYDRATE project. Their usefulness having been clearly demonstrated during these previous research programs, post-flood surveys have been included in the observation strategy proposed for the recently launched HYMEX project (Hydrological cycle in Mediterranean experiment). The principles of the post-flood survey methodology will be shortly presented and its possible outcomes will be illustrated on the basis of some flash floods recently analysed within the EC HYDRATE project.

18

SESSION: PROCESSES

RETROSPECTIVE CASE STUDIES OF FLOOD SIMULATIONS TO ADDRESS THE SOCIOPOLITICAL ISSUES

Hoes, O. and Tariq, M.A.U.R.

Faculty of Civil Engineering and Geosciences, Technical University Delft, Stevinweg 1 Delft, 2628CN, the Netherlands E-mail: [email protected], [email protected]

At the end of World War II nearly 120 000 German soldiers were surrounded by allied forces on a peninsula in the West of the Netherlands. As an act of despair the German army blasted the dike surrounding a low-lying area in order to prevent landings of paratroopers. As a result this 20 000 hectare large polder flooded in 48 hours with a layer of 4 meters of water. Over 7000 people succeeded to flee at once from the three villages in the area to higher grounds. Despite the damage, this flood was well archived compared to other floods in polder areas in the Netherlands. This provides today enough information to make this flood to be one of the few full-scale experiments. The objective of this paper is to compare the flood simulation results with the actual flood data since 1945. Both the context, the breach growth process and the flood are explained in our paper. Furthermore, it will be shown that to reproduce the flood - in our case - not bottom friction, but final breach size is the only suitable but sensitive calibration parameter. For the majority of all other conducted flood simulations in the Netherlands, such comprehensive data are not available. This retrospective case simulation enhances the reliability of simulations carried by the modern software tools with the help of GIS data.

19

SESSION: PROCESSES

TOWARDS ADVANCED SOLUTIONS FOR THE CALIBRATION OF HEAVY METAL DISPERSION CONTAMINATION MODELLING IN FLOODPLAIN

Hostache, R.1, Matgen, P.1, Hissler, C. 1, Kies, T.2, Tosheva, Z.2, Stille, P.3, David, E.2 and Bates, P.3

1 Environment and Agro-biotechnologies department, CRP – Gabriel Lippmann, 41 rue du Brill 4422 Belvaux, Luxembourg, 2 Radiation Physics Laboratory, University of Luxembourg, 162A avenue de la Faïencerie 1511 Luxembourg, Luxembourg 3 LHyGeS - UMR 7517, CNRS/EOST / UdS, 1 rue Blessig 67084 Strasbourg Cedex, France 4 Water Resources and Environment Division, SOGREAH Consultants, 6 rue de Lorraine B.P. 218 38432 Echirolles Cedex, France 5 School of Geographical Sciences, University of Bristol, University Road Bristol BS8 1SS, United-Kingdom E-mail: [email protected] [email protected] [email protected], [email protected] [email protected], [email protected], [email protected], [email protected]

Recent years have seen a growing awareness of the wider environmental significance of the sediment loads transported by rivers and streams. This includes the importance of suspended sediment in transporting nutrients and contaminants, such as some Potential Harmful Elements (PHE - Cr, Co, Ni, Cu, Zn, As, Cd, Sn, Sb, Hg, Pb) - which threaten the water quality and can cause severe impacts in downstream areas like wetlands and floodplains. Contemporary data on the sediment loads of rivers provide clear evidence of significant recent changes in the sediment fluxes of several rivers in response to human activities. Sediment transport models validated by in situ experimentations are the only available tool to predict the consequences of natural as well as man-induced environmental changes and impacts on sediment dynamics. Mathematical models can help our understanding of important processes and interactions that affect the water quality of water bodies. They are approximate representations of complex natural systems and the evaluation of a model with respect to its ability to reproduce multiple criteria and behaviour of a real system is still problematic. Discussions between modellers and experimentalists improve significantly the interpretation of the modelling output. These interactions between different scientific disciplines led to formulate more realistic assumptions on the behaviour of natural systems. The geochemical information, which appeared to be non-correlated with the hydrological standard parameters, provides new information and contributes to give an “orthogonal view” on the hydrologic system behaviour. Therefore, the evaluation of models by using additional information – or complementary data – may give new information to reject the erroneous hypotheses made during the model development. This applies especially to basins that are heavily contaminated by PHE. Regarding the recent development in geochemical tracer applications in models, the multi-tracer approach (elemental concentrations and ratios, stable isotopes, radionuclides) may be a necessity to decrease significantly the uncertainties in hydraulic modelling. This study focuses on these new approaches in order to propose advanced solutions for the calibration of 2D and 3D hydrodynamic models. This multidisciplinary approach is required to assess the risk of floodplain contamination due to river sediment deposition. An important set of hydrological, sedimentological and geochemical data is used in order to reach a more efficient calibration of the TELEMAC modelling system. In addition to standard techniques of hydrochemistry, new approaches of in situ suspended sediment transport monitoring, geochemical tracing and spectroscopic analyses (Near InfraRed Spectrometry) will help getting new insights on the hydraulic system behaviour.

20

SESSION: PROCESSES

OPERATIVE EVALUATION OF PEAK OUTFLOW DURING FLASH FLOOD USING ARTIFICIAL INTELLIGENCE METHODS

Janál, P.1 and Stary, M.2

1 Czech Hydrometeorological Institute, Regional Forecasting Department, Kroftova 43, 616 67 Brno, Czech Republic 2 Brno University of Technology, Faculty of Civil Engineering, Institute of Landscape Water Management, Žižkova 17, 60200 Brno, Czech Republic E-mail: [email protected], [email protected]

The torrential rainfall prediction should be followed by suitable hydrological model able to estimate at least the resultant peak outflow. The hydrological model construction interferes with high measure of uncertainty, inherent in the rainfall prediction, rainfall-runoff process and its simulation. The way how to eliminate influence of uncertainty is using fuzzy logic and other artificial intelligence methods. The article deals with the fuzzy model capable of predicting peak outflow of water from the basin in the case of flash floods. The fuzzy model was compiled through the Fuzzy Logic Toolbox in the developmental environment of MATLAB. Inputs characterizing the causal rainfall and the affected basins morphology have been chosen for construction of a model. The constructed model was tested by means of operational simulation during the summer season 2009.

The work was supported by Central Europe Programme, INCA-CE project (co-financed by European Regional Development Fund). This article was written as a part of the GAČR project: 103/07/1620 “Prediction and simulation models in the water resources control”.

21

SESSION: PROCESSES

URBAN FOOTPRINTS ON CATCHMENT STORM RUNOFF

Kjeldsen, T. R., Miller, J. D. and Packman, J. C.

Centre for Ecology & Hydrology, Maclean Building, Crowmarsh Gifford, Wallingford, UK. E-mail: [email protected]

The effect of urban land-cover on catchment flood response is evaluated using a lumped rainfall-runoff model to analyse flood events from selected catchments in the United Kingdom. The present study proposes and evaluates three extensions to an existing lumped rainfall-runoff model to enable it to represent the urban effects, namely an increase in runoff volume, reduced response time and a decrease in baseflow (resulting from decreased infiltration). Based on observed flood events from seven catchments, cross-validation methods are used to compare the predictive ability of the model variants with that of the original unmodified model. The results show that explicit inclusion of the urban effects increases the predictive ability of the model across catchments, despite significant large between-event variability of model predictive ability. More detailed investigations into the relationship between model performance and event characteristics (antecedent soil moisture, rainfall duration, depth and intensity) did not identify obvious patterns, suggesting that the model structure is unbiased, and indicating the residual level of between-event variation of storm generating processes.

22

SESSION: PROCESSES

ESTIMATING CATCHMENT SCALE RETENTION CAPACITY FROM SCATTEROMETER SOIL MOISTURE DATA

Komma, J.1, Merz R.2, Blöschl, G.1 and Wagner, W. 3

1 Institute for Hydraulic and Water Resources, Engineering, Vienna University of Technology, Vienna, Austria. 2 Department for Catchment Hydrology, Helmholtz Centre for Environmental Research (UFZ), Halle, Germany. 3 Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, Vienna, Austria. E-mail: [email protected]

Accurate knowledge of soil moisture is critically important for many purposes in hydrology and water resource management, in particular, in hydrological forecasting. The fraction of rainfall that becomes runoff (and does not infiltrate) is closely related to the soil moisture state. As the soil moisture increases so does runoff, and the relationship is highly non-linear. In this study we investigate the potential of the Advanced Scatterometer (ASCAT) sensor onboard of the Metop satellite for the assessment of antecedent soil moisture conditions in diverse physiographic regions of Austria. In a hydrologic context, the main advantages of using spaceborne sensors for soil moisture estimation is their global availability and that they provide an integral value over a large area rather than point values. The remotely sensed soil moisture data is used to estimate the antecedent wetness conditions at a catchment scale considering the relationship between SWI and the soil potential maximum retention parameter, S, of the Soil Conservation Service-Curve Number (SCS-CN) method. As a reference or benchmark observed runoff and rainfall data is used to estimate event based retention capacities S for a large number of various catchments in Austria. First results indicate that ASCAT soil moisture estimates do have the potential to improve the results from SCS-CN method in Austria.

23

SESSION: PROCESSES

3-DIMENSIONAL MODELING OF RUNOFF GENERATION OF A MOUNTAINOUS RIVER

Kuchment, L.S. and Demidov, V.N.

Institute of Water Problems of the Russian Academy of Sciences (IWP RAS), 119333, Gubkina 3, Moscow, Russia, E-mail: [email protected]

A physically based 3-dimensional model of mixed snowmelt-rainfall runoff generation of mountainous river is proposed. The model is based on the finite-element schematization of catchment area and river channel network and includes description of snow accumulation and melt with accounting their dependences on altitude and topography, soil moisture dynamics, overland, subsurface and channel flow. A case study was carried out for the Upper Kuban River basin situated in Northern Caucasus region (the catchment area is 16900 km2). The input air temperature and precipitation are interpolated on basis of the 11 available meteorological station data. The mean values of topography, river channel and soil characteristics are assigned for every area or channel finite element using available topography and soil maps and channel measurements. Five parameters have been calibrated against runoff measurements at five runoff gauges. The carried out estimations of Nash-Sutcliffe efficiency measures for each used gauge for ten-year calibration and verification periods has showed that the developed model gives acceptable accuracy of calculation of the runoff hydrographs. As an additional attempt of verification of the developed model, simulation of the catastrophic flood of June 2002 was carried out during which the maximum water stages in the Kuban region observed at various gauges exceeded the maximum stages for the entire observation period. The model has been applied for short- term hydrological forecasting.

24

SESSION: PROCESSES

ON THE VALUE OF MULTIOBJECTIVE OPTIMISATION FOR PROFITING FROM SOFT INFORMATION IN UNGAUGED BASINS

Lombardi, L., Montosi, E., Toth, E., Castellarin, A. and Montanari, A.

Faculty of Engineering, University of Bologna, Via del Risorgimento 2, I-40136 Bologna – Italy E-mail: [email protected]

The performances of single-objective and multi-objective algorithms for the calibration of a rainfall-runoff model are compared by referring to an ungauged framework. Regional procedures are applied to derive river flow statistics to be optimised in order to calibrate a rainfall-runoff model in absence of hydrometric observations. Different combinations of several objective functions were tested in calibration and validation mode. The minimisation of the set of objective functions is carried out by using the AMALGAM algorithm to identify behavioural parameter sets. The analysis refers to a wide area located in central Italy, in a region where detailed hydrological and geomorphoclimatic information is available. These catchments are considered as gauged and ungauged in turn: a baseline for model performances is first obtained under gauged conditions and it is then compared to what would be obtained in absence of observed runoff data. Alternative hydrological models and combinations of river flow statistics were tested. The results confirm the potential of the multiobjective approach and its capability to take advantage of different information sources in a context characterized by significant uncertainty.

25

SESSION: PROCESSES

RECONSTRUCTION OF THE HYDRAULIC CONDITIONS OF THE 1951 INUNDATION OF THE POLESINE REGION, ITALY

Masoero, A.1, Claps P.1, Di Baldassarre, G.2 and Asselman, N.3

1 Dipartimento di Idraulica, Trasporti ed Infrastrutture Civili, Politecnico di Torino, Torino, Italy 2 Department of Hydroinformatics and Knowledge Management, UNESCO_IHE, Delft, The Netherlands 3 Deltares, Delft, The Netherlands E-mail: [email protected], [email protected], [email protected]

Flood inundation models have become essential tools in flood risk management, due to the capability to simulate the hydraulic behavior of rivers and floodplain areas during extreme hydrologic events. In particular, 1D and 2D models can be used for the analysis of historical flood events, which can clarify baseline configurations for a more reliable assessment of the flood hazard. This study aims to reconstruct the 1951 inundation of the Polesine Region, Italy. The 1951 flooding was a mayor natural catastrophe in the XXth century in Europe. It caused an inundated area of 1,080 km2 and produced devastating social consequences, such as the displacement of more than 180,000 people. The reconstruction of the 1951 hydraulic conditions is currently based on partial knowledge of discharges and water stages at the Pontelagoscuro’s gauging station (downstream of the 1951 levee breach) using also extrapolation of the rating curves beyond the measurement range, which is, even today, something open to uncertainty. To get a more reliable reconstruction of this relevant historical event, we carried out a numerical exercise by using state-of-the-art flood inundation models. In particular, the simulation of the 1951 flooding was based on a hybrid decoupled methodology: a 1D model (HEC-RAS) was used to simulate the flow into the river and to compute the overflow through the levee breach; this result was then adopted as the inflow condition for a 2D model (SOBEK) application on the inundated area. The cross-section survey made in 1954 was used as the geometrical input. The 1D model was also used to reconstruct the historical rating curves. A good agreement between the patterns of the observed and reconstructed inundation areas was found, and the timing of the inundation was also found to be close to the information derived from the historical chronicles. Lastly, using the results of the flood inundation modeling exercise, some technical considerations about the relation between water stages and discharges, were drawn. In particular, the maximum discharge simulated in the lower portion of the Po River was found to be significantly lower than the one estimated in the past.

26

SESSION: PROCESSES

2D HYDRODYNAMIC FLOOD MODELLING IN A GRAVEL BED BRAIDED RIVER

Molnar, P., Boehringer, D., Wanner, P. and Burlando, P.

Institute of Environmental Engineering, ETH Zurich, 8093 Zurich, Switzerland E-mail: [email protected]

While numerical 2-dimensional flood simulation models have become standard tools in engineering practice (e.g. flood risk mapping), their development and improvement continues to be an important research topic. A particularly challenging scientific and practical problem is the simulation of large gravel bed braided rivers where very detailed spatial information on bed topography is required in order to accurately capture complex flow paths when the inundation of the river and floodplain increases dramatically with a rise in discharge. Apart from obvious risk floods in these environments pose to humans and infrastructure, there are also other effects floods have on the natural riverine system (e.g. riparian vegetation, sediment transport and morphology, aquatic habitat) that are relevant and need to be considered. In this paper, we conduct a modelling exercise to investigate the effects of floods on some selected processes in the natural riverine system. The study area is a 4 km reach of the Maggia River, a gravel bed braided stream with one of the last remaining sizable natural floodplain ecosystems in Switzerland. The 2D vertically integrated hydrodynamic model BASEMENT was used to simulate a sequence of steady state flows ranging from low flows to floods with return period 300 yrs. Spatially distributed maps of flow depth, velocity, bed shear stress and Froude number were produced for all flows. Aerial photographs were used to calibrate the model under low flow conditions, and to map the riparian woodland distribution (shrubs and hardwood species) on the floodplain. The simulation results were then analyzed with three main focus points: the effect of floods on morphological adjustment, riparian vegetation and fish habitat. First we show that only floods with return periods of 3 yrs or more have the capacity to conduct major morphological work. The longitudinal distribution of the average excess bed shear stress required for motion of representative grains on the bed is spatially very variable, which indicates why sediment transport in braided rivers is intermittent and most sediment is re-deposited rather than exported out of the reach. What is also very evident is the highly nonlinear response of the system to flood magnitude driven by the complex 3-dimensional floodplain topography. Second we show that the riparian vegetation distribution on gravel bars and banks is to some degree organized by floods by processes of vegetation removal by erosion and the formation of suitable germination sites where the presence of fine grains (sand) are key. On this basis we discuss the potential impacts of streamflow regulation due to hydropower system operation in the headwaters of the Maggia basin on riparian vegetation in the valley. Third we show how river habitat can be characterized on the basis of flow depth and velocity for a common species of fish in the river (Leuciscus souffia) and on that basis we estimate the potential habitat loss due to streamflow regulation. Our modelling shows that the main advancement of 2D flood modelling in gravel bed braided rivers, i.e. the ability to capture spatial variability, allows us to better explain riverine processes and study their connections.

27

SESSION: PROCESSES

EVALUATION AND MODELLING OF THE IMPACT OF LAND USE ON DISCHARGE, NITRATE AND SUSPENDED SOLIDS CONCENTRATION BY FLOOD EVENTS IN SMALL CATCHMENTS

Moravcova, J., Pavlicek, T., Ondr, P., Koupilova, M. and Kvitek, T.

University of South Bohemia, Faculty of Agriculture, Department of landscape management, Czech Republic E-mail: [email protected]

The aim of this article is to evaluate the role of landscape structure on particular water quality parameters by flood events. The partial aims were to assess the impact of other factors, which may have the influence on the change of selected water quality parameters during flood events, and to simulate the impact of landscape structure in the catchment on the discharge, nitrate and suspended solids concentration progress by SWAT model. All the experimental works were carried out in two catchments – Jenínský and Kopaninský streams in the southern respectively central part of the Czech Republic. The important result of this research was the confirmation of the theory about rainfall water infiltration by flood events by preferential flow paths, which is proved by synchronic dilution of nitrate concentration and increase of phosphate concentration with the raising runoff on the rising limb of rainfall-runoff event hydrograph. On the other hand the measured data of suspended solids by flood events show the rise of the concentrations; this coincides with water level rising. Very important is also the confirmation of significant role of infiltration areas management by runoff and nitrate concentration formation by the flood events. Model solution in SWAT interface proved the significant reduction of runoff and nitrate concentration values by agricultural use exclusion in infiltration localities, and their grassing and afforestation. The results clearly showed that special attention should be paid to the infiltration localities with regard to runoff height and water contamination. For reduction of suspended solids concentration by flood events very satisfactory results were achieved very satisfactory results by all scenarios of land use change (grassing and afforestation of particular parts of the catchment).

This article is based on results of grant of Ministry of Agriculture of the Czech Republic QH92034 Recharge zones identification by water vegetation stress in chosen catchments.

28

SESSION: PROCESSES

ON THE USE OF THE DIFFUSIVE WAVE MODEL FOR MODELLING EXTREME FLOOD EVENTS IN THE FLOODPLAIN: NATURAL vs ANTHROPOGENIC CATCHMENTS

Moussa, R.1, Ghesquière, J.1 and Bocquillon, C.2

1 INRA, UMR LISAH, 2 Place Pierre Viala, 34060 Montpellier Cedex 1, France 2 Université Montpellier II, Place Eugène Bataillon, 34095 Montpellier, France E-mail: [email protected]

During the last two decades, extreme flood events which occurred in the Mediterranean zone in Southern France are a major threat to human life and infrastructures (death of over 100 people and around a billion Euros of damage). Catastrophic flood events are frequently chosen as reference events, for modelling the impact of scenarios of river bank remediation to prevent inundations. However, these rare events are poorly understood due to the poor hydrological knowledge available, and due to the lack of data. A model for routing flood events adapted to few available data during historical extreme events was developed. The model subdivides the floodplain into cells linked to the main channel, and the diffusive wave model with overbank flow was used to route floods in the main channel and through the inundation floodplain (Moussa and Bocquillon, 2009, Journal of Hydrology, 374: 116-135). As input data, the model requires a Digital Elevation Model, geographical map and cross sections in order to subdivide the channel network into cells. The model parameters are the Manning’s roughness coefficients in the main channel and in the floodplain. The linearized equations are resolved by an implicit numerical scheme to ensure stability and convergence. A new multi-criteria analysis method, to assess the model performance at the flood event scale, is proposed and applied (Moussa, 2010 Hydrological Science Journal, 55(6):1074- 1084). We discuss the significance of the well-known Nash-Sutcliffe efficiency criteria when calculated separately on flood events and we propose a new simple quick-to-use method to analyse model performance using multi-criteria functions, on the basis of a lag translation, or a homothetic ratio of the calculated hydrograph. Application cases were conducted on two catchments located in southern France: under natural conditions in the case of the Herault basin (2540 km²) and on an anthropogenic elementary catchment characterized by a dense channel network in the case of the Roujan basin (1 km²) a subcatchment of the Herault. First, we studied the flood event of October 1958 which caused one of the largest inundations in Southern France during the XXth century (410 mm of rainfall during 48h). The model was applied to simulate the flood event of 1958, and was validated on other flood events under similar hydro-meteorological conditions. Then, the model was applied to quantify the impact of scenarios of river bank remediation. On the Herault basin, fourteen scenarios of construction of dams, embankments, or coupling both dams and embankments, were modelled. On the Roujan basin, three scenarios of land use management, agricultural practices, and channel network spatial distribution were also modelled. The model simulates hydrographs in the main channel, the water depth, the area and the volume on both right and left banks of the floodplain cells. Results show that the model developed herein, with relatively little hydraulic modelling, is well adapted to simulate extreme flood events, and is useful for environmental managers to study the impact of river bank remediation on inundations.

29

SESSION: PROCESSES

A COMPARISON OF DIFFERENT APPROACHES TO DESIGN FLOOD ESTIMATION IN ALPINE CATCHMENTS

Rogger, M.1, Kohl, B.3, Merz, R.2, Viglione, A.1 and Blöschl, G.1

1Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Austria, 2 Department for Catchment Hydrology, Helmholtz Centre for Environmental Research UFZ, Germany, 3Department of Natural Hazards and Alpine Timberline, Federal Research and Training Centre for Forests, Natural Hazards and Landscape, Austria, E-mail: [email protected], [email protected], [email protected], [email protected], [email protected],

Design floods in small alpine catchments are estimated by a variety of different methods. In general bottom up and top down methods can be distinguished. Applied bottom up methods are for instance, event based and continuous rainfall runoff models which, if detailed information on the catchments is available, can describe local hydrological processes precisely. The standard top down approach often applied is flood frequency statistics, where the design flood is estimated by fitting a distribution to measured discharge data. Top down approaches that can be used in case when no discharge measurements are available are regionalisation methods. If the different methods are applied to the same catchment the estimated design floods are in some cases consistent but may in other cases also be very different. This study presents a comparison of design flood estimates for ten small alpine catchments in Tyrol (Western Austria), where the 100-year flood was estimated applying bottom up and top down approaches. In all ten catchments a detailed analysis of the runoff processes was performed applying two bottom up approaches, a deterministic approach with an event based rainfall runoff model and a probabilistic approach including Monte Carlo simulations. As top down approaches the design floods were estimated by flood frequency statistics based on measured discharge data and in a second approach by the Top Kriging regionalisation method where the catchments were assumed to be ungauged. The results of the different approaches were analysed in detail and show that inconsistencies in the flood estimates can be attributed to factors such as specific hydrological processes on a local scale or different assumptions on which rainfall input to use. Furthermore the analysis highlights the advantages and disadvantages of each method, which should be taken into account in design flood estimation.

30

SESSION: PROCESSES

WOODY DEBRIS TRANSPORT DURING FLOODS: 2D HYDRODYNAMIC MODELLING APPROACH

Ruiz Villanueva, V.1, Bladé Castellet, E.2, Sánchez Juny, M.2, Bodoque del Pozo, J.M.3 and Díez Herrero, A.1

1 Geological Survey of Spain, Natural Hazards Division. Ríos Rosas Street, 23, 28003, Madrid, Spain. 2 Flumen Institute, Universitat Politècnica de Catalunya. Jordi Girona Street, 31, 08034 Barcelona, Spain. 3 Mining and Geological Engineering Department, University of Castilla La Mancha, Campus Fábrica de Armas, Avda. Carlos III, 45071, Toledo, Spain. E-mail: [email protected], [email protected], [email protected]. [email protected] [email protected].

Woody material mobilization in rivers has been studied during the last decades. Most of the research carried out so far has been focused on the morphological role of woody debris, the recruitment processes, and its spatial distribution in the streams. In addition to the ecological and hydro-geomorphic points of view, one of the most important topics is the potential hazard in communities adjacent to streams, since wood dramatically increases the destructive power of floods. However, very few studies have dealt with this phenomenon as a potential hazard during floods. Models that predict when and where wood will be entrained or deposited could prevent and mitigate the risk of such destructive events, but to this purpose there are no models available for it so far. In this work a numerical model has been developed in order to simulate the hydrodynamics and other fluvial processes (turbulence, sediment transport, etc.) together with the woody debris transport. This model has been integrated as a new module into the IBER hydraulic bidimensional model. IBER is a numerical model for simulating turbulent free surface unsteady flow and environmental processes in river hydraulics. Nowadays, IBER has 3 main computational modules: a hydrodynamic module, a turbulence module and a sediment transport module. All of them work in a finite volume non-structured mesh made up of triangular or quadrilateral elements. The hydrodynamic module solves the depth averaged Shallow Water Equations (2D-SWE), also known as the two-dimensional St. Venant Equations. Wood entrainment is considered in relation to a force balance model acting on wood in streams. Furthermore, turbulence may affect the woody debris transport. To incorporate this effect an additional resistance to fluid deformation resulting from the internal chaotic fluid motions that characterize turbulent flow has been added. The interactions between logs and channel configuration and between logs each other have been also taken into account in the model. The influence of woody debris (deposits or individual logs) in hydrodynamics has been solved with an additional drag. To verify the model presented here, flume experiments were carried out. Thus, the behaviour of wood in rivers was simulated performing different flume configurations with different flow conditions and using several types of dowels as LWD.

31

SESSION: PROCESSES

DISCHARGE HYDROGRAPH ESTIMATION USING 2D DIFFUSIVE MODEL AND SYNCRONOUS WATER LEVEL MEASUREMENTS

Sinagra, M.1, Nasello, C.1, Moramarco, T.2 and Tucciarelli, T.1

1 Department of Civil, Environmental and Aerospace Engineering, University of Palermo, Viale delle Scienze 90128 Palermo, Italy 2 Institute for Hydrologic Research, National Research Council, Via Madonna Alta 126, 6100 Perugia, Italy E-mail: [email protected],[email protected], [email protected] [email protected]

Discharge monitoring in rivers, especially during flood events, is very important for the calibration of hydrological and hydraulic models. Recently, new techniques have been developed that are based on 1D hydraulic modeling and unsteady-state water level data analysis. The adopted hydraulic model has the capability to compute the routed wave in terms of discharges q(x,t) and water levels h(x,t) given as input the approximated initial condition (at t=0) along all the reach, the stage hydrograph measured in the upstream section (at x=0) and a constant roughness coefficient. The upstream discharge hydrograph is computed by the model as q(0,t). An optimization process is used to estimate the roughness coefficient that minimizes the mean square error of the computed stages at the downstream section (x=L), with respect to the measured ones, within the time interval of the rising limb. This optimum roughness coefficient is adopted to compute the discharge hydrograph in the upstream section. Most of the discharge estimation methods that are based on velocity measurements along the entire cross section are aimed to the construction of the rating curve that links the measured water stage to the unknown discharge. The reliability of the rating curve strongly decreases when the measured water stage is much above the experimental points that are the couple of discharge/stage values simultaneously measured. The proposed approach only requires the use of two water level sensors located in two different sections far few kilometers from each other, does not require any direct velocity measurement and also allows the simultaneous estimation of the mean roughness coefficient along the reach between the two sections. One limit of the method is that, when extreme floods occur, the 1D assumption becomes very weak especially in the most dangerous sections, where the flood plain inundation implies consistent lateral velocity components. To overcome this limitation, the diffusive MAST 2D model is applied, in the same framework of the previous 1D approach. Because only one water level sensor is likely to be available for each monitored section, a constant water level has to be assigned, at each computational time, to all the nodes that are located along the upstream section. Similarly, the same water level measured by the instrument located in the downstream section is assumed to hold for all the nodes located along the downstream section. The methodology has been applied to the case study of the Tiber River (Italy), where water levels were recorded during two flood events. The topography of the main channel and flood plains was reconstructed using an unstructured triangular mesh, based on a 2mx2m fine DEM. The results were compared with the discharge hydrograph computed using the 1D model, as well as with the discharge hydrograph obtained using a rating curve, previously constructed using direct velocity measurements.

32

SESSION: PROCESSES

SNOWMELT RUNOFF PREDICTION IN THE UPPER EUPHRATES BASIN, TURKEY

Sorman, A.A., Gozel, E. and Sensoy, A.

Anadolu University, Department of Civil Engineering, Eskişehir, Turkey E-mail: [email protected]

Snowmelt runoff in the mountainous eastern part of Turkey is of great importance, as it constitutes nearly 2/3 in volume of the total yearly runoff during spring and early summer months. Therefore, keeping track of snow dynamics and forecasting the amount and timing of snowmelt runoff in the headwaters of the trans-boundary Euphrates River, where large dams are located, is an important task in order to plan dam regulations, control flooding and optimize water resources. Upper Euphrates Basin in Turkey with an area of 10 275 km2 and elevation ranging from 1125-3500 m is selected as the study area since it is one of the main branches feeding the Keban Dam Reservoir (first dam in the series of large dams in Euphrates Basin). Automatic snow-meteorological stations are installed at various locations and altitudes in the Upper Euphrates Basin operating in real-time. Since ground based observations can only represent a small part of the region of interest, spatially and temporally distributed snow cover data are acquired through the use of Moderate Resolution Imaging Spectroradiometer (MODIS) optical satellite. Daily 500 m resolution MODIS satellite images on Terra and Aqua platforms are used to determine cloud free snow depletion curves using certain combination and filtering techniques. Snowmelt Runoff Model (SRM) is applied in the basin to forecast runoff with a 2-day lead time for 2011 snowmelt period. Meteorological input data of precipitation and temperature are provided from numerical weather prediction models and snow extent data is estimated through stochastic modeling using MODIS images. In conclusion, promising results indicate the possible operational use of runoff forecasting in the Upper Euphrates Basin for future flood events.

33

SESSION: PROCESSES

INFORMATION EXPANSION FOR UNDERSTANDING TRANS-BASIN FLOODS

Uhlemann, S.1, Thieken, A.2 and Merz, B.1

1 Helmholtz Centre Potsdam – GFZ German Research Centre for Geosciences, Section Hydrology, Potsdam, Germany 2 Helmholtz Centre Geesthacht – Climate Service Center (CSC), Hamburg, Germany E-mail: [email protected], [email protected], [email protected]

Flood risk assessment is at foremost a subject of high societal relevance. It is therefore inherently a subject of governmental action with a large body of authorities concerned with the management of this risk and the planning of measures for flood loss reduction on all state levels. Consequently, besides the scientific body of literature, an even larger body of studies is being produced through these authorities, often endowed with a higher degree of detail. Yet, the scientific community largely ignores these sources of information and hydrologic expertise. To analyse the reasons and to assess their validity we present results of an extensive study aiming at the expansion of information for an improved understanding of trans-basin floods in Germany. Trans-basin floods are extreme events occurring on a regional scale and across catchment boundaries. Recently, Uhlemann et al. (2010) have presented a complete and consistent set of trans-basin floods in Germany for the period between 1952 and 2002 based on an analysis of multiple series of mean daily discharge. Understanding theses floods requires an integrated assessment of the whole process chain from the meteorological cause via the runoff formation in the catchments, the flood routing in the rivers, the effect of flood protection, to the negative consequences of hydrological extremes. Our objectives are three-fold: First, to identify the existing body of literature available on floods in Germany and to assess the main obstacles in their application for flood analysis. By applying a systematic literature review that includes an intensive study on sources and ways of information dissemination, we identify a total of 250 relevant reports and studies. Secondly, we exploit the information contained in the material for the task: What are the governing hydrometeorological processes that have caused trans-basin flood events in Germany over the period 1952 to 2002 and what were the consequences of these floods? Therefore, we introduce the conceptual hierarchical framework developed for data quality by Wang and Strong (1996) and adopt it to the assessment of flood report quality. The framework explicitly accounts for the context in which the report is being used and intrinsic, representational and accessibility quality criteria. We extend the concept by a pedigree scoring scheme to quantify the level of information within each of the individual quality criteria and to quantify the overall report quality. Thirdly, for the largest 30 trans-basin flood events we combine the extracted and quality- labelled information with a data based analysis of the hydrometeorological causes of trans- basin floods. We can demonstrate that the interdisciplinary combination of methods and systematic analysis of information can contribute to an improved process understanding.

REFERENCES: Uhlemann, S., et al. (2010), A consistent set of trans-basin floods in Germany between 1952-2002, Hydrology and Earth System Sciences, 14(7), 1277-1295. Wang, R. Y., and D. M. Strong (1996), Beyond accuracy: what data quality means to data consumers, Journal of Management Information Systems, 12(4), 5-33.

34

SESSION: PROCESSES

FLOOD INUNDATION MODELLING IN LARGE RIVERS UNDER UNCERTAINTY USING GLOBALLY AND FREELY AVAILABLE REMOTE SENSING DATA

Yan, K.1, Di Baldassarre, G.1, Solomatine, D.1 and Dottori, F.2

1 Department of Hydroinformatics and Knowledge Management, UNESCO-IHE Institute for Water Education, Delft, the Netherlands. 2 Università di Bologna, Dipartimento di Scienze Della Terra e Geologico-Ambientali, Via Zamboni 67, 40126 Bologna, Italy E-mail: [email protected], [email protected], [email protected], [email protected]

The extreme consequences of recent catastrophic events have highlighted that flood risk prevention still needs to be improved to reduce human losses and economic damages, which have considerably increased worldwide. Hydraulic modeling is proved to be a powerful tool for supporting flood risk management and lowland development. However, the lack of hydrological and topographical data often has often been the bottleneck of implementing hydraulic modelling and producing flood hazard and/or risk maps. Thus, the current proliferation of new space-borne data has a great potential. However, the practical value of the globally and freely available remote sensing data for hydraulic modeling still needs to be tested. This research aims to contribute to the understanding of whether the current growth in availability of globally and freely available remote sensing data may allow the production of probabilistic flood maps for the entire globe, i.e. global floodplain mapping. In this study, the SRTM DTM is used for hydraulic model building, while ENVISAT-ASAR satellite image is used for model validation. To test the usefulness of these globally and freely available data, a model based on the high resolution LiDAR DTM and ground data is used as benchmark. The work is facilitated by the use of a data-rich test site: a 98km reach of the River Po in Northern Italy. The results show that the hydraulic model based on globally and freely available data provides a satisfactory performance for a high magnitude flood event. Lastly, this technology is generalized and implemented in two case studies to produce probabilistic flood inundation maps of two large rivers: Yellow River (China) and Blue Nile (Sudan).

35

SESSION: PROCESSES

36

EGU Leonardo Conference Series on the Hydrological Cycle

SESSION: PATTERNS

Conveners: A. Castellarin, P. Claps, T. Kjeldsen, R. Merz

The analysis of flood patterns may provide key elements for addressing a number of topical issues in hydrology. In particular, the session focused on: • How analysis of space-time patterns of floods may effectively support flood hazard estimation through an improved process understanding; • An enhanced assessment and interpretation of variability and change in space-time patterns of flood data; • Integration of remotely and locally sensed data for improving understanding of flood patterns.

37

38

SESSION: PATTERNS

DESIGN FLOOD GENERATION FOR LOWLAND RIVER SECTIONS ACCOUNTING THE UNCERTAINTY GENERATED BY SUPERPOSITION AND COINCIDENCE OF FLOODS

Bálint, G.1, Lipták, G.1, Szilágyi, J.2 and Zempléni, A.3

1 VITUKI Enviromental Protection and Water Management Research Institute 2 Department of Hydraulic and Water Resources Engineering, Budapest University of Technology and Economics 3 Department of Probability Theory and Statistics, Eötvös Loránd University, Budapest E-mail: [email protected], [email protected]

Major floods in the Tisza River system are usually the results of the superposition of several flood waves from upstream sections and also their coincidence with floods of the tributaries. This phenomenon was the basis for simulation exercises which were carried out for a limited number of scenarios generated from the combination of a few historical events. A more complex approach using a hybrid, seasonal, Markov chain based model for daily streamflow generation were also used combined with a flood routing tool for the upper Tisza (Szilagyi et al 2006). Preliminary success brought the motivation for this study where probabilistic and hydrological methods were further combined in order to achieve an accurate model for possible flood scenarios in the future. The first step in model construction was to fit a non- homogenous Poisson process to the flood occurrences in the given area. The reversible jump Markov chain Monte Carlo technique, originating from Green (1995) was also applied. Results received produce possible future scenarios of flood events. For practical execution three types of hydrological stations compose the simulation scheme, with different data needs, namely: (a) Upstream stations for Markov Chain Monte Carlo (MCMC) type of simulations of daily discharge series: Long observed daily discharge series are required with minimum length of 30-50 years. (b) Flood routing stations: Observed 2-10 years of discharge and water level series, preferably covering the entire range of observed water levels, discharges, i.e. high and low water years. Additionally, rating curve representing present day or designed future conditions. In case of sufficient station density observed water level time series can also be utilized to create “interpolated” rating curves and/or discharge series. (c) Target stations, selected flood routing stations, which simulated discharge and/or water level “design” hydrographs are used as input and boundary conditions for hydrodynamic models. Data requirements are as at flood routing stations – minimum requirement. To check and validate flood frequency estimates received for simulated series additional long observed daily discharge series are beneficial (at least 30-50 years). In most cases monthly (or at least annual) maxima are sufficient. This may help water managers to prepare for events that have not yet been observed in the past but none the less can be expected in the future. Detailed analysis of target station flood distributions also supply the measure of uncertainty resulting from the coincidence and superposition of flood waves.

39

SESSION: PATTERNS

INDIRECT ESTIMATION OF DESIGN VALUES OF HEAVY RAINFALL IN THE ALPINE REGION OF SLOVAKIA: A CASE STUDY

Bara, M.1, Látečková, J.2, Kohnová, S.2, Gaál, L.2, Szolgay, J.2 and Hlavčová, K.2

1 Institute of Hydrology, Slovak Academy of Sciences, Bratislava, Slovak Republic 2 Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Bratislava, Slovakia E-mail:[email protected]. [email protected], [email protected], [email protected], [email protected], [email protected]

Design values of heavy rainfall are of very great importance in engineering hydrology, such as input data for hydrological modeling, for the prediction of flood events, or for planning and design in water resources management. In this study, the possibility of using the simple scaling method for an indirect estimation of design values of heavy rainfall in a mountainous area of Slovakia was tested. For the analysis, the Alpine region was selected, which is located in the northern part of the country. In general, the simple scaling method allows for an estimation of the design values of rainfall of selected recurrence intervals and durations (i.e. T-year quantiles) shorter than a day by using only the daily data from the regular network of non-recording rain gauges. Thus, the design values of 1-day rainfall amounts were assessed (both by local and regional estimation methods), and the T-year quantiles of higher temporal resolution were derived by scaling of the T-year quantiles of the 1-day rainfall amounts. For the case study, a number of verification meteorological stations were selected, in which the scaling exponents were estimated for particular months of the warm season and also for the whole warm season. According to this, averaged monthly and seasonal regional scaling exponents were derived for the whole selected area. Using the regionally averaged scaling exponents, the IDF curves were estimated by downscaling the quantiles of daily precipitation totals. Finally, the IDF curves were compared one to another and to those assessed locally in previous studies.

40

SESSION: PATTERNS

FLOOD POTENTIAL IN SMALL BASINS OF SLOVAKIA

Barka, I.1, Minár, J.2, Trizna, M.2 and Bonk, R.

1 National Forest Centre, Forest Research Institute, T. G. Masaryka 22, 960 92 Zvolen, Slovakia, 2 Faculty of Natural Sciences, Comenius University, Mlynská dolina, 845 15 Bratislava, Slovakia, E-mail: [email protected], [email protected], [email protected]

Landscape potential for floods (flood potential) in Slovakia has been assessed at a scale of 1:500 000 using geographic approach in 2005. Two partial flood potential were modeled: morphometric potential for floods and geoecological potential for floods. The basic spatial operational units were based on morphological-morphometrical terrain type. Current research is aimed at evaluation of flood potential in the more detailed scale. It uses similar methodological approach, but it is based on more precise data (e.g. digital elevation model - DEM with 10 m resolution, Corine Land Cover data, forestry inventory data). The small basins with an average area of approximately 25 km2 were chosen as a basic operational spatial unit. Morphometric potential express spatial differentiation of runoff velocity and was computed using slope angle and drainage density. Geoecological potential involves also soil properties and land cover as well. Overall potential for floods was estimated as a combination of geoecological potential and meteorological data (probability of maximum daily rainfall within 100 years time period). It was calibrated by hydrological data (floods frequency data and rate of mean and maximal discharge). Resulting flood potential is presented in ordinal scale. The results were interpreted separately for flash floods and large floodplain related seasonal floods. The new map of flood potential showed similar pattern comparing to assessment from 2005 but with more detailed spatial segmentation.

41

SESSION: PATTERNS

THE SURFACE WATER OCEAN TOPOGRAPHY SATELLITE MISSION: PROSPECTS FOR FLOOD SCIENCE

Bates, P.

School of Geographical Sciences, University of Bristol, Bristol, BS8 1SS, UK E-mail: [email protected]

In 2019 NASA and CNES will launch the Surface Water Ocean Topography satellite mission (see swot.jpl.nasa.gov) which will measure surface water height to centimetric accuracy every 10 days with complete global coverage at 1km resolution over the oceans and for all rivers above 100m wide. Over land the SWOT mission will provide measurements of water storage changes in terrestrial surface water bodies and will provide estimates of discharge in large (50 m-100 m width) rivers, globally. These data sets will provide a quantum improvement on our current understanding of surface water and flooding dynamics. Beyond improved characterization of the water cycle, meeting these measurements will enable numerous applications of great scientific, social, and political importance.

42

SESSION: PATTERNS

EUROPEAN FLOOD DATA AVAILABILITY AND PARENT DISTRIBUTION OF ANNUAL FLOODS

Castellarin, A.1, Kohnová, S.2 and ES0901 – WG2 Team3

1 Department DICAM, School of Civil Engineering, University of Bologna, Bologna, Italy 2 Department of Land and Water Resources Management, Faculty of Civil Engineering, STU Bratislava, Slovakia 3 ES0901 – WG2 Team and affiliations are appended below the Abstract body E-mail: [email protected], [email protected]

Flood frequency analysis is broadly used for estimating the design-flood at a given location (i.e., flood magnitude associated with the recurrence interval T, T-year flood). The approach can be implemented locally (At-Site Flood Frequency Analysis, SFFA), or regionally (Regional Flood Frequency Analysis, RFFA). SFFA and RFFA are mature disciplines and consolidated methodologies are available for many European regions. As a result, different European countries (and sometimes even different regions within a country) adopt different methodologies. Also, methodologies are often selected on the basis of traditional choices or what is known to work well for a limited number of local catchments, rather than an objective assessment of available methods. The main objective of COST Action ES0901 European procedure for flood frequency estimation (FloodFreq, http://www.cost-floodfreq.eu/) is to address this disconnected picture, promoting a pan-European and synergic approach to flood frequency analysis as requested by European Flood Directive (2007/60/EC). In particular, during the first year of implementation of FloodFreq, the Working Group focusing on assessment of statistical methods for flood frequency estimation (WG2) compiled a catalogue of flood data availability/unavailability across Europe, together with relevant information (e.g., catchment descriptors) recommended for use in flood frequency studies. This paper presents the results of this recompilation effort, together with the outcomes of some preliminary analyses that aim at identifying in an L moment-based framework the most suitable parent distributions for representing the frequency regime of annual maximum peak discharges across Europe. Results refer to some 2000 sequences of annual floods observed in 8 countries.

3 ES0901 – WG2 Team Kjeldsen, T., Centre for Ecology & Hydrology, Wallingford, Oxfordshire, UK Szolgay, J., Gaál, L., Department of Land and Water Resources Management, Faculty of Civil Engineering, SUT Bratislava, Slovak Republic Diez, J. J., PuertosyCostas Escuela de Ingenieros de Caminos Politecnica de Madrid, Ciudad Universitaria, Madrid, Spain Fleig, A., Lawrence, D., Norwegian Water Resources and Energy Directorate, Oslo, Norway Lang, M., Cemagref, Lyon, France Loukas, A., Department of Civil Engineering, University of Thessaly, Pedion Areos, Volos, Greece Macdonald, N., The University of Liverpool Department of Geography, Roxby Building, Liverpool, United Kingdom Madsen, H., DHI, Horsholm, Denmark. Mediereo, L., Département de Génie Civil: Hydraulique et Energie, Université Polytechnique de Madrid Merz, B., Helmholtz Centre Potsdam – GFZ German Research Centre for Geosciences, Section Hydrology, Potsdam, Germany Neykov, N., National Institute of Meteorology and Hydrology, Bulgarian Academy of Sciences, Sofia, Bulgaria Nicholson, O., IHP/ICID, Hydrology and Flood Response Section, Office of Public Works, Dublin, Ireland Onuşluel Gül, G., Dokuz Eylul University DEU Insaat Muh. Bolumu, Izmir, Turkey Ozga-Zielinski, B., Institute of Meteorology and Water Management, National Research Institute, Warszawa, Poland Pistocchi, A., EURAC, Institute for applied remote sensing, European Academy of Bolzano, Bolzano, Italy Salinas, J., Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Austria Kriauciuniene, J., Sarauskiene, D., Laboratory of Hydrology, Lithuanian Energy Institute, Kaunas, Lithuania Strupczewski, W., Department of Hydrology and Hydrodynamics, Institute of Geophysics, Polish Academy of Sciences Toumazis, A., Dion Toumazis & Associates, Nicosia, Cyprus Veijalainen, N., Finnish Environment Institute, Freshwater Centre, Helsinki, Finland Willems, P., K.U.Leuven, Hydraulics Division, Leuven, Belgium

43

SESSION: PATTERNS

ACCOUNTING FOR EXTREME FLOODS OCCURRED AT UNGAUGED SITES IN REGIONAL FLOOD FREQUENCY STUDIES

Cong, N.C., Payrastre, O. and Gaume, E.

IFSTTAR, Bouguenais, France E-mail: [email protected]

Flood frequency analyses are often based on continuous series at gauging stations. However, the length of the available data sets is usually too short to provide reliable estimates of extreme design floods. Hence, hydrologists have tried to make use of alternative sources of information to enrich the datasets used for their statistical inferences. Three main approaches were therefore proposed. The first one consists of a "temporal extension" of information based on the valuation of historical and paleoflood data. The second, "spatial extension”, consists in merging statistically homogeneous data to build a large regional data samples. Recently, a combination of the two techniques aiming at including regional estimated extreme discharge values at ungauged sites in regional flood frequency analyses has been proposed (Gaume et al., 2010). This new approach necessitates the calibration of a function relating the characteristics of the watersheds to a local index flood value and the calibration of a regional one. In the initially proposed approach, the index flood relation was adjusted a priori and the uncertainties associated with this calibrated relation were not considered. An improvement of this method will be presented where both (i) the index flood relation and (ii) the regional growth curve are calibrated in the same time using a Bayesian MCMC framework and enabling an accurate estimation of the uncertainties associated to the estimated peak discharge distributions (computation of credibility intervals). Two regional samples from France (167 records at 5 gauges and 35 ungauged extremes) and central Vietnam (212 records at 7 gauges and 3 ungauged extremes) are used to test and validate the method. Three types of results will be presented. 1. Comparison of this new proposed approach with a standard regional flood frequency method when applied to measured discharge series available at gauged sites only. 2. Evaluation of the added value of extremes estimated at ungauged sites, which can only be incorporated in the statistical inference with the new proposed method 3. Evaluation of the accuracy of the credibility intervals for the adjusted distributions computed by the MCMC Bayesian inference framework for the new and standard method through Monte Carlo runs. All these tests reveal the accuracy and robustness of the new proposed regional flood frequency method.

REFERENCES: Gaume, E., Gaál, L., Viglione, A., Szolgay, J., Kohnová, S., Blöschl., G., 2010. Bayesian MCMC approach to regional flood frequency analyses involving extraordinary flood events on ungauged sites. Journal of hydrology. doi:10.1016/j.jhydrol.2010.01.008

44

SESSION: PATTERNS

INFLUENCE OF OCEANIC PHENOMENA ON FLOODS

Diez, J.J., Esteban, M.D., López-Gutiérrez, J.S. and Negro, V.

Universidad Politécnica de Madrid, Spain. FP7-SMARTEST Project Researcher E-mail: [email protected]

Flood events are relatively frequent phenomena in coastal areas from the Iberian. This type of flood affects not only to the Mediterranean area of the Iberian Peninsula, but also to other ones, like Biscay Gulf. And not only to coastal areas in a strict way, but also to other interior areas, out of the Iberian Peninsula borders. Although this can occur at any zone of the Peninsula, even with extension to the whole of it, the main intensity usually runs on the eastern and southeastern areas of the Peninsula, particularly on its Mediterranean catchments and basins. Some flooding examples of this type are the following ones: Valencia, 1957; Cataluña, 1962; Almeria, 1973; Tous or the Santa Irene event, 1982; Valencia and Murcia, 1987 and 1997; Cordoba, 2010; etc. The Santa Irene flood, happened at the end of October 1982, is one of the most disastrous flood events in Spain. Its renown is mainly due to the rupture of the Tous dam, but its main meaning is to be the paradigm of the incidence of the littoral weather and the sea level on the coastal plains inland floods. The type of floods before mentioned is related to a meteorological phenomenon known as the cold drop one, which will be explained here. This phenomenon leads not only to inland floods caused by rains and fluvial currents, but also to floods in coastal areas as consequence of a combination of rains, fluvial currents, oceanic waters entrance and/or, at least, temporal increase of the sea level. There are some circumstances that can easily unleash the cold drop there: cold and dry polar air masses coming onto the Iberian Peninsula and Africa, high water temperatures and low atmospheric pressure (cyclone) areas in the western Mediterranean basin; these circumstances are characteristics of the autumn season in that region with a great space- temporal yearly variability (similar to hurricanes on Caribbean and western North-Atlantic areas). Therefore, these phenomena may have a different magnitude each time. The Santa Irene flood is analyzed, being selected because of the existence of enough information and data about it. It is tried to expose here the result of a detailed analysis and reflection about this phenomenon as a whole, on the generation of its rains and on the different natures and consequences of its flood, exposing overall the ways in which the maritime weather and consequent sea level govern the flood on the lowest zone of the hydrographical basin. Above all because this type of floods can be identified at a first view as only inland, some of them happened in areas far the coast for example the Caribbean Hurricanes and the Monsoons.

REFERENCES: AUDIENCIA TERRITORIAL DE VALENCIA. 1991. Sentencia en la causa penal 56/1982, por el desmoronamiento de la presa de Tous. 23 de octubre de 1990, Valencia. DIEZ, J.J., 1992. El fenómeno meteorológico de Santa Irene. Revista de Obras Públicas, n. 3315: 77-81.

45

SESSION: PATTERNS

INVESTIGATING RUNOFF GENERATION MECHANISMS IN A SWISS PRE ALPINE CATCHMENT BY SPATIAL INTERCOMPARISON

Fischer Benjamin, M. C.1 and Seibert J.1, 2

1 University of Zurich, Department of Geography - Hydrology and Climate Winterthurerstrasse 190, 8057 Zürich, Switzerland 2 Uppsala University, Department of Earth Sciences, Uppsala, Sweden E-mail: [email protected], [email protected]

Long term detailed data in mountainous catchments are generally sparse and even when they exist, internal data such as groundwater or isotopic data are usually lacking. This makes difficult to identify runoff generation processes. Here we present first results from a new data set collected in the Zwäckentobel, a 4.3 km2 Swiss pre-alpine catchment. The high annual precipitation input, flashy character of streams and dominant wet conditions, make this catchment an exciting study area for hydrologists. However the contribution and response of streams to storm flow in space and time is largely unknown. This study tried to address this issue by comparing several sub-catchments. The aim was to learn from differences, similarities and to put each catchment into perspective at different scales. The basic question was whether there is a difference in response to rainfall and if so are the runoff contribution processes different? For the Zwäckentobel and six ungauged sub-catchment rainfall, runoff and isotope concentrations were measured and compared with the one sub-catchment with long-term observations as reference. Here we present first results of this study, which show the difference in rainfall runoff response and stable isotopes within this relative small catchment scale. Especially the comparative isotope hydrograph separations were found to be informative.

46

SESSION: PATTERNS

SCENARIOS OF HEAVY, FLOOD-TRIGGERING PRECIPITATION IN THE CARPATHIAN AREA, ON THE BASIS OF HIGH RESOLUTION REGIONAL CLIMATE MODELS

Gaál, L.1,2, Beranová, R.2, Kyselý, J.2 and Hlavčová, K.1

1 Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Bratislava, Slovakia 2 Institute of Atmospheric Physics, Czech Academy of Sciences, Prague, Czech Republic E-mail: [email protected], [email protected], [email protected], [email protected]

Heavy precipitation events are of great importance since they may cause soil erosion, landslides and floods. The present study examines possible changes in the regime of heavy precipitation, aggregated on two different time scales: 1 hour and 5 days. Both time scales have significant role in hydrology: while the heavy 1-hour precipitation, predominantly from localized convective systems, may be one of the triggering factors of flash floods, heavy 5- day precipitation events, usually of frontal origin, may lead to devastating large scale floods. We analyze simulated precipitation outputs from 12 regional climate models (RCMs) with high spatial resolution 0.5° (~25 km) over Europe made available within the EU-FP6 ENSEMBLES project. The RCMs are driven by several global climate models; nevertheless, all the simulations are based on the SRES A1B emission scenario. The study examines the 50-year return values of the seasonal 1-hour and 5-day precipitation maxima that are expected to occur at the end of the 21th century (time slice 2070–2099) in comparison with the same statistics estimated for the control period (time slice 1961–1990), both for the summer (June–August) and the winter (December–February) seasons. The high quantiles of heavy precipitation are assessed by means of (i) local (at-site) frequency analysis in individual grid boxes, and (ii) regional frequency analysis based on the region-of-influence methodology as modified by Kyselý et al. (2011). In the latter method, a set of similar grid boxes (a ‘region-of-influence’) is formed for each grid box, from which information on extreme precipitation is taken when estimating extreme value distribution at the target grid box. Although the grids of the RCMs cover the entire Europe and all the frequency modelling is carried out for the whole set of grid boxes for each RCM, the study focuses on a narrower geographical area encompassing the Carpathian mountain range (17°–28°E and 44.5°– 50.5°N). The selected area has a complex orography and includes the whole territory of Slovakia and Hungary as well as major parts of Romania and Ukraine. Slovakia is of a particular interest of the study since as far as we are concerned, (i) no climate-change scenarios of precipitation regime were based on such a dense network of grid boxes, and (ii) no studies concerning the expected changes in precipitation at sub-daily scales appeared in the regional scientific literature.

REFERENCES: Kyselý, J., Gaál, L., Beranová, R., Plavcová, E., 2011. Climate change scenarios of precipitation extremes in Central Europe from ENSEMBLES regional climate models. Theoretical and Applied Climatology 104: 529–542. doi:10.1007/s00704-010-0362-z.

47

SESSION: PATTERNS

CLIMATOLOGICAL AND GEOLOGICAL CONTROLS OF REGIONAL PATTERNS OF FLOOD TIME SCALES IN AUSTRIA

Gaál, L.1, Szolgay, J.1, Kohnová, S.1, Parajka, J.2, Merz, R.3 and Blöschl, G.2

1 Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Bratislava, Slovakia 2 Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Vienna, Austria 3 Department of Catchment Hydrology, Helmholtz Zentrum für Umweltforschung UFZ, Halle, Germany E-mail:[email protected], [email protected], [email protected], [email protected], [email protected], [email protected]

In this study, the dependence structure of two characteristics of extreme rainfall-runoff events, i.e., flood peaks and flood volumes are examined. Unlike the common practice where the dependence between two variables is usually described by means of linear or rank correlation coefficients, here we focus on process based analysis of residual similarity of dependence between these two variables that is expressed by the flood time scale defined as the ratio of the flood volume to the flood peak. The case study is based on approximately 400 catchments in Austria. In order to compare the variability of the flood time scales between and within the regions, several smaller groups of pilot basins are used, within each one having homogeneous geological background. The flood time scales of these are compared in terms of their cumulative distributions functions. The hydrological and climatological driving factors of flood generation are taken into consideration through a stratification of the time scales according to the seasons of flood occurrence and flood types defined by Merz and Blöschl (2003). Pronounced regional differences in the similarity of hydrological response were found, which can be explained by different hydrological processes. For example, in Weinviertel (NE of Austria), short duration flash floods play an important role with median time scale of ~10 hours. On the other hand, in the Alpine region, the rain-on-snow process dominates, leading to large time scales exceeding 80–100 hours. In some areas, the time scales are consistent between years as the majority of the flood events belong to a given flood type, such as long rain floods, while in other ones (e.g., Innviertel and Mühlviertel), there is a mix of long duration rain-on-snow and short duration frontal floods leading to a large variance of the time scale.

REFERENCES: Merz, R., Blöschl, G., 2003. A process typology of regional floods. Water Resources Research, 39(12), 1340. doi:10.1029/2002WR001952.

48

SESSION: PATTERNS

USING PROXIMITY ALONG THE RIVER NETWORK TO IMPROVE DESIGN FLOOD ESTIMATION AT UNGAUGED SITES

Ganora, D., Laio, F. and Claps P.

Dipartimento di Idraulica, Trasporti ed Infrastrutture Civili, Politecnico di Torino, Torino, Italy E-mail: [email protected]

Standard regional approaches for flood estimation at ungauged sites normally do not preserve the information related to the hierarchy among gauged stations, which derives from their location along the river network. However, this information is particularly important when estimating runoff at a site located immediately upstream or downstream a gauging station. We propose here a novel approach, referred to as Along-Stream Estimation (ASE) method, to estimate a flood statistics at an ungauged site (target site), as a function of regional estimates of the same variable (at the donor and target site) and of ancillary information about the physical characteristics of both basins. A criterion to define the domain of application of the ASE approach is proposed, and the uncertainty inherent with the obtained estimates is evaluated. This allows one to compare the variance of the along-stream estimates to that of other models eventually available for application (e.g. regional models), and thus to choose the most accurate method (or to combine different estimates). This approach is particularly useful in areas where many gauged stations with short record are available. The ASE model has been applied in the northwestern part of Italy in connection to an existing regional model for flood frequency analysis. The analysed variables are the first L- moments of the annual discharge maxima. The application demonstrates that the ASE approach can be effectively used to improve the regional estimates for the L-moment of order one (the index flood), particularly when the area-ratio of a pair of donor-target basins is less or equal to ten. On the other hand, the method does not provide significant improvements to the estimation of higher-order L-moments.

49

SESSION: PATTERNS

IMPROVING SIMULATIONS OF INUNDATION PATTERNS BY AIRBORNE LASER SCANNING ANALYSES – A CASE STUDY IN A DANUBE WETLAND

Gschöpf, C.1, 2, Vetter, M.1, 3 and Blöschl, G. 1, 2

1 The Centre for Water Resource Systems, Vienna University of Technology, Karlsplatz 13, A-1040 Wien 2 Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Karlsplatz 13/222, A-1040 Wien 3 Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, Gusshausstraße 27-29, A- 1040 Wien E-mail: [email protected] , [email protected], [email protected]

This study presents an analysis of inundation patterns covering both the extent of the inundated area and the dynamics of a flood event. The frequency and duration of flooding at partial areas of the flood plain is included. The flow regime is simulated by hydrodynamic modelling. The aim of this study is to describe hydrological habitat characteristics that limit the performance of ecosystems. This particularly applies to the flood regimes which are part of the morphological equilibrium of riverine systems. Open-channel hydraulics, therefore, are important to microbial and ecological research. This integrated interdisciplinary approach raise river-basin management to a new scientific level. Two-dimensional modelling, which is used for non-parallel overbank flow, requires a great amount of input data from areas that are often poorly monitored relative to their size. To satisfactorily represent the heterogeneous topographic surface in hydraulic models, the commonly used parameterisation of the topographic elevation is derived from highly accurate airborne laser scanning (ALS) data. To this end, digital terrain models (DTMs) are used; however, in many cases the available accuracy of terrain information cannot be implemented due to computer performance. Flow resistance, generally known as surface roughness, is represented by Manning’s n. The surface roughness is traditionally estimated from land cover maps which are a result of time consuming field campaigns and aerial images. A further aim of this study is to present the estimation of surface roughness based on the same data set as that used for the creation of the DTM. This new method provides an automatic classification of the vertical structure of the ALS point cloud. This could be very important when terrestrial field campaigns for land cover mapping cannot be realised at all, or not to a sufficient degree. Currently used land cover maps could also have become obsolete due to morphological changes that caused an alteration in the vegetation cover. These morphological changes are due to the floods that took place in the interim since the date of mapping. This study is focused on an urban flood plain, the Lobau, which is part of the Danube riverine landscape within the city limits of Vienna. The complex transient flow regime arises from the back flooding of the river Danube through an opening at the downstream end of the flood protection levee. This occurs after a certain threshold water level has been exceeded and the flood spreads out over the flood plain.

50

SESSION: PATTERNS

CAN WE ATTRIBUTE FLOOD TRENDS?

Merz, B., Delgado, J., Hundecha, Y., Nguyen, D., Uhlemann, S. and Vorogushyn, S.

Helmholtz Centre Potsdam – GFZ German Research Centre for Geosciences, Section Hydrology, Potsdam, Germany E-mail: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]

Over the last decade, the number of studies on trend detection in series of flood flows is ever increasing. Frequently, when significant trends are detected, explanations on the causes of the detected changes are given. This attribution step, i.e. establishing the most likely causes for the detected changes, is very important in order to put the identified changes into context and to evaluate the implications for flood risk management. We review the state-of-the-art of flood trend attribution and assess to which extent attribution conclusions of recent studies are credible. A classification for attribution studies is proposed, and we argue that approaches are needed that integrate both a critical and broad usage of available statistical approaches for trend detection as well as an attribution that extends beyond the confirmation of a single hypothesis towards an in-depth analysis of a variety of physical and human induced explanatory causes.

51

SESSION: PATTERNS

CROSS-CALIBRATION AT REGIONAL SCALE FOR RAINFALL-RUNOFF MODELING IN UNGAUGED CATCHMENTS

Montosi, E.1, Montanari, A.1, Toth, E.1, Parajka, J.2 and Blöschl, G.2

1.University of Bologna, DICAM, Bologna, Italy 2 Institute for Hydraulic and Water Resources Engineering, Vienna University of Technology, Vienna, Austria E-mail: [email protected]

The objective of this study is to explore one possible solution to the question whether it is possible to satisfactorily parameterize hydrological models in case of lack of runoff measures. In ungauged or scarcely gauged basins it is impossible to turn to the classical calibration of rainfall-runoff (R-R) models. Therefore it is common to transfer the parameters from gauged catchments that are hydrologically similar, to the ungauged catchment with a regionalization technique. We propose a cross-calibration procedure based on the adoption, for all the gauged donor catchments, of a unique, space- invariant parameter set. Therefore the proposed scheme implies that the same parameters can be directly applied to the ungauged basin. A basin in turn, for whom we pretend the local runoff data are missing, is selected and identified as target catchment. We will refer to all the remaining catchments of the study area as the donors. In detail, if there are n watersheds in the region of the ungauged catchment, the process will need (n-1) steps. Firstly, the R-R model is calibrated on the most similar donor, chosen according to a catchment distance measure. With a second step the process develops calibrating the R-R model on the two most similar donors, and so on adding a new donor at each step. For each set of donors we consider, as stated above, always a unique parameter set, and the calibration is based on the simultaneous optimization of the simulated streamflow over all the gauged catchments for the same calibration period. The parameter set identified at each step is then applied, as it is, to the ungauged catchment, without any further transfer technique. With the identified cross-calibration we want to analyze the trade-off between the disadvantage of assuming the parameters homogeneous in space and the benefit of adding precious new information as the cross-calibration evolves, investigating if involving an increasing number of catchments in the simultaneous cross-calibration leads to more robust parameters estimation.

52

SESSION: PATTERNS

ANTICIPATORY FLOOD RISK ASSESSMENT - CASE STUDIES FROM AUSTRIA

Neuhold, C.

Universitaet fuer Bodenkultur Department fuer Wasser - Atmosphaere - Umwelt Institut fuer Wasserwirtschaft, Hydrologie und konstruktiven Wasserbau E-mail: [email protected]

The assessment of flood risk comprises hazard assessment and vulnerability assessment. Flood hazard varies due to hydrological characteristics, land use change and climate change. Flood vulnerability strongly depends on human behaviour, flood mitigation and flood protection as well as susceptibility and exposure. The exposure to floods increased substantially during the past decades due to change in land use from agricultural to industrial and residential areas in former flood plains, leading to a significant increase in physical and human exposure and consequently higher damage potential. This paper aims at developing new approaches and methodologies to assess anticipatory flood risk to enable strategic regional development within flood prone areas. Therefore, state of the art approaches – assessing the current state of development – are extended by future development scenarios based on several data sources referring to demography, land use, building types, settlement characteristics, historic data, etc. Based on this information the scenarios (1) “status quo”, (2) “extrapolation of previous development” and (3) “realistic maximum development” are assessed based on various hydrologic scenarios (HQ30, HQ100, HQ300 and extreme events). New approaches are developed, tested and validated based on two case studies in Austria. Both case study sites are characterised by rural settlements exposed to medium sized rivers (Mattig and Große Rodl) whereas the municipality of Pfaffstätt/Mattig (case study 1) is not protected against floods at all, and the municipalities Scharzgrub, Rodl and Höflein/Großen Rodl (case study 2) are protected against a 100-year flood. Therefore, aspects of risk and residual risk are discussed by this paper and are incorporated in the presented flood risk assessment methodologies. Both case study sites show remarkable increase of flood risk (discounted on 2010) – for case study 1 an increase of the expected annual losses is estimated at roughly 40% until 2100 (2010: 180000€/a; 2100: 250000€/a); for case study 2 tremendous increases are expected as areas without protection are highly likely to be adapted (filling up to HQ100 level) and settled. Nevertheless, an increase of the expected annual losses by a minimum of five times the current values has to be expected (2010: 5000€/a; 2100: 28000€/a). The results clearly indicate that the consideration of prospective development scenarios is needed to enable comprehensive and anticipatory flood risk assessment providing a reliable basis for adequate flood risk management strategies.

53

SESSION: PATTERNS

ELBE 3D – A SPATIO-TEMPORAL ANALYSIS OF SOIL MOISTURE PATTERNS LINKED TO EXTREME FLOODS IN THE ELBE RIVER BASIN

Nied, M., Hundecha, Y. and Merz, B.

Helmholtz Centre Potsdam, GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germany E-mail: [email protected], [email protected], [email protected]

Large scale flood disasters, such as the Elbe / Danube flood in August 2002, are the result of a complex interaction between meteorological event characteristics and catchment conditions. This work is, therefore, aimed at developing a framework for flood risk assessment that considers the interaction between meteorological and hydrological initial conditions explicitly. The methodology is based on identifying extreme flood flows and classifying important hydrological state variables (e.g. soil moisture) corresponding to the identified events. The state variables are simulated using a semi-distributed conceptual rainfall-runoff model that is calibrated using a Monte Carlo scheme with an objective function that emphasizes high flows. Daily soil moisture index (soil water content standardized by field capacity) at the subbasin scale is used as an indicator of the current catchment state. A principal component analysis is applied to identify patterns of soil moisture index that explain most of the variability of the soil moisture dynamics within the catchment. A cluster analysis is applied to the leading principal components to identify days of similar soil moisture patterns. The study was applied to the Elbe catchment, which was subdivided into 2268 subbasins with a mean size of 65 km², using daily hydrological and meteorological data over the period 1951-2003. Our analysis shows that the first three principal components explain the soil moisture spatio- temporal variability related to flooding within the Elbe catchment. Cluster analysis applied to these principal components identifies flood prone clusters. The relative frequency of flood days within a cluster ranges between 0.0 and 0.32. The probability of discharge within each cluster is computed to provide a quantitative probabilistic statement on high flows within a cluster and thus on a soil moisture pattern. In a subsequent step, precipitation patterns will be identified and conditional probability distributions of discharge within each cluster of soil moisture and precipitation pattern will be computed. This will enable us to quantify the effect of interaction of catchment state and event precipitation characteristics on flood risk.

54

SESSION: PATTERNS

THE GRADEX-KC AND GRADEX-ZN METHODS FOR COMPUTING MAXIMUM FLOODS WITH T-YEAR RETURN PERIOD WHERE DISCHARGE MEASUREMENT SERIES ARE INCOMPLETE

Ozga-Zieliński B.

Institute of Meteorology and Water Management (IMGW-PIB), National Research Institute, Podleśna 61, 01-673 Warsaw, Poland E-mail: [email protected]

The Gradex-KC method enables estimation of annual maximum floods with T-year return period on the basis of short, say 10 to 15 years, series of annual maximum river discharge and long i.e. 40 or more years annual maximum precipitation series. The method can be used for relatively small basin areas i.e. up to 500 km2. In case of small catchments, particular rainfall incident can cover whole basin area, what leads to assumption of its uniform distribution on entire basin. Also small catchments can be characterized relatively small variability of physiographical conditions, which allows to assume that direct runoff is similar on entire basin area. Two main assumptions are underlying the Gradex-KC method. Firstly, the rainfall is assumed to yield an exponential, i.e. EV I distribution and secondly, the rainfall over certain threshold Po is totally transformed into direct runoff. Accepting these both assumptions, the maximum floods distribution function, above a value of discharge corresponding to a value of precipitation Po, can be approximated by the maximum rainfalls distribution function. This approximation is possible thanks to two parameters and Po, which are estimated in the Gradex-KC method. Parameter a called GRADEX (GRADient EXtremes), as a parameter of rainfalls distribution function, is evaluated on the basis of maximum precipitation series. But parameter Po is calculated on the basis of the relationship between short series of river discharge observations and corresponding to them rainfall observations. The second problem, which is discussed, concerns estimation of maximum floods with T- year return period in cases of ungauged watersheds due to proposed regional relationships i.e. extended Gradex-ZN method. The Gradex-ZN method is based on theoretical assumptions adopted in the Gradex-KC method introducing changes in parameters estimation taking into account requirements of ungauged watershed. The parameter is estimated for all rainfall stations situated on the area of ungauged watershed. Then on the basis of particular gradex parameter of each rainfall station, isolines of gradex are drawn and then they are used to evaluate a parameter for ungauged watershed. Whereas Po parameter for ungauged watershed is proposed to be calculated on the basis of relationship between river discharges and rainfalls for controlled basin on which ungauged watershed area is situated.

55

SESSION: PATTERNS

DECADES OF FLOOD SEASONALITY ACROSS THE ALPINE–CARPATHIAN RANGE

Parajka, J.1, Kohnová, S.2, Szolgay, J.2, Merz, R.3, Hlavčová, K.2 and Blöschl, G.1

1 Vienna University of Technology, Vienna, Austria, www.hydro.tuwien.ac.at, 2 Slovak University of Technology, Bratislava, Slovakia 3 Helmholtz Zentrum für Umweltforschung UFZ, Halle, Germany E-mail: [email protected]

The aim of this contribution is to analyze the differences in the long-term regimes of extreme precipitation and floods across the Alpine–Carpathian range, which includes the southeastern part of France, Switzerland, the northern part of Italy, Austria, the southern part of Germany, Slovakia, Romania and a small region along the Ukraine–Hungarian border. This region represents diverse climatologic conditions between the Atlantic, the Mediterranean and the continental part of Europe. The assessment is based on data collected in the HYDRATE project (Gaume et al., 2009) and includes the time series of annual maximum runoff at 577 stream gauges and annual maximum daily precipitation at 1945 climate stations with at least 20 years of measurements in the period 1961– 2000. The seasonality assessment is based on directional statistics, which represents an effective method for defining similarity measures on the basis of the timing of hydrological extreme events with a year. We used the Burn index vector to analyze the mean date and variability of occurrence of the extreme events. In order to investigate the change of the seasonality with time period and magnitude of extreme events, the seasonality index is separately estimated for different decades and for different number of the largest annual maximum values. The results will present similarities and differences in flood regime seasonality across study region and discuss the shift in flood generating processes within different decades.

REFERENCES: Gaume, E., Bain, V., Bernardara, P., Newinger, O., Barbuc, M., Bateman, A., Blaškovicová, L., Blöschl, G., Borga, M., Dumitrescu, A., Daliakopoulos, I.,Garcia, J., Irimescu, A., Kohnova, S., Koutroulis, A., Marchi, L., Matreata, S.,Medina, V., Preciso, E., Sempere-Torres, D., Stancalie, G., Szolgay, J., Tsanis, J.,Velasco, D., Viglione, A., 2009. A collation of data on European flash floods. Journal of Hydrology 367, 70–78. doi:10.1016/j.jhydrol.2008.12.028.

56

SESSION: PATTERNS

AN ANALYSIS OF CHANGE IN ALPINE ANNUAL MAXIMUM DISCHARGES: IMPLICATIONS FOR THE SELECTION OF DESIGN DISCHARGES

Pistocchi, A.1, 2 and Castellarin, A.3

1 GECOsistema srl Rimini/Cesena, Cesena, Italy 2 Romagna Rivers Regional Basin Authority, Forlì, Italy 3 Department DICAM, School of Civil Engineering, University of Bologna, Bologna, Italy E-mail: [email protected]

The contribution presents an analysis of 17 long annual maximum series (AMS) of flood flows for Swiss Alpine basins, aimed at checking the presence of changes in the frequency regime of annual maxima. We apply Pettit´s change-point test, the nonparametric sign test and Sen´s test on trends. We also apply a parametric goodness-of-fit test for assessing the suitability of distributions estimated on the basis of annual maxima collected up to a certain year for describing the frequency regime of later observations. For a number of series the tests yield consistent indications for significant changes in the frequency regime of annual maxima and increasing trends in the intensity of annual maximum discharges. In most cases, these changes cannot be explained by anthropogenic causes only (e.g., streamflow regulation, construction of dams). Instead, we observe a statistically significant relationship between the year of change and the elevation of the catchment outlet. This evidence is consistent with the findings of recent studies that explain increasing discharges in alpine catchments with an increase in the temperature controlling the portion of mountain catchments above the freezing point. Finally, we analyze the differences in return periods (RP’s) estimated for a given flood-flow on the basis of recent and past observations. For a large number of the study AMS we observe that, on average, the 100-year flood for past observations corresponds to a RP of approximately 10 to 30 years on the basis of more recent observation. From a complementary perspective, we also notice that estimated RP- year flood (i.e., flood quantile associated with RP) increases on average by approximately 20% for the study area, irrespectively of the return period RP. Practical implications of the observed changes are illustrated and discussed.

57

SESSION: PATTERNS

T-YEAR MAXIMUM DISCHARGES ON WATER COURSES IN SLOVAKIA

Podolinská, J.1, Šipikalová, H.1 and Kohnová, S.2

1 Slovak Hydrometeorological Institute, Regional Centre Banská Bystrica, Zelená 5, 974 04 Banská Bystrica, Slovak Republic 2 Faculty of Civil Engineering, Slovak University of Technology in Bratislava, Radlinskeho 11, 813 68 Bratislava, Slovak Republic E-mail: [email protected], [email protected], [email protected]

The T-year maximum discharges estimation is one of the most difficult tasks in hydrology. To choose the method for design discharge calculation is also complicated and depends on the catchment size and data availability. The Slovak Hydrometeorological Institute (SHMI) processes and provides these data according technical standards. It is a tradition that the whole territory of Slovakia is processed and the system of river network is considered. This way was also applied for data updating with new methods and technologies (GIS). Background materials were time series of hydrological data obtained from surface water stream-gauging stations during the whole observation period (minimum 20 years). T-year maximum discharges in 340 stream-gauging stations were calculated using mathematical- statistic methods. Pooling scheme of 100-year maximum specific discharge (q100.max) was used in the profiles without observation. The cluster analysis was applied for regional types and relative estimate deviation q100.max determined using non-linear regression formula expressing dependence of this value on specific physical-geographical characteristics. Input data were time series from 197 selected stations with a catchment area from 20 to 300 km2 and minimum impact of anthropogenic activities.

58

SESSION: PATTERNS

FLOOD PREDICTION IN UNGAUGED CATCHMENTS: REDUCING UNCERTAINTY WITH AN OPTIMUM USE OF REGIONAL INFORMATION

Randrianasolo, A., Ramos, M-H. and Andréassian, V.

Cemagref, Hydrology Research Group, UR HBAN, Antony, France E-mail: [email protected]

The most commonly used tools in applied hydrology are essentially data-driven, which poses a problem for several drainage basins in the world that are ungauged or poorly gauged. These basins are characterized by the absence of discharge measurements or by incomplete data series, with several gaps and a lack of information content for hydrological applications. In flood forecasting, one of the main challenges when dealing with ungauged basins concerns the need of data for, on one hand, the calibration of model parameters and, on the other hand, the updating of initial conditions at the onset of the forecasts. Information has to be retrieved from other hydrologically similar gauged catchments, which usually correspond to neighbouring catchments within a homogeneous region, and transposed to the ungauged site. The uncertainty associated with such procedure is considerable and strategies to reduce the predictive uncertainty need to be implemented to produce reliable forecasts that can be useful to decision-makers. The objective of this study is to investigate different strategies to reduce the predictive uncertainty in ungauged catchments, with the help of the geographical neighbourhood. Different scenarios are considered, combining both the transfer of model parameters from neighbour catchments ("donors") for rainfall-runoff simulation, and the transfer of observed discharges for updating the initial conditions at the time of forecast. In search of an optimal use of the information coming from neighbours, the assumption that using different donors to infer model parameters and to derive initial conditions for updating will improve forecast performance is also tested. A leave-one-out cross validation approach is applied for 211 catchments in France, considering each catchment as ungauged at a time. Daily forecasts are issued by a lumped soil-moisture accounting type rainfall-runoff model, using observed rainfalls as input, as well as ensemble precipitation forecasts from the French meteorological centre (Météo-France) available for a 4-year period. Flow forecasts are evaluated using skill scores and the performances of the different tests are compared. The results show that the main contribution of neighbouring gauged catchments is in the transfer of model parameters. The transfer of specific discharges for updating requires a more meticulous search for donors. The added value of having at least local data available at the time of the forecast to perform the real-time updating is illustrated. Key issues on alternative strategies to apply when no local updating is possible are discussed.

59

SESSION: PATTERNS

STOCHASTIC MODELING OF LARGE FLOODS – SPATIAL DEPENDENCE BETWEEN LOCAL RUNOFFS

Raschke, M.

Fa.M.Raschke, Gustav-Frextag-Str.24, 04277 Leipzig, Germany E-mail: [email protected]

A stochastic model for floods in a large region has to include a model for the spatial dependence between different runoffs beside the frequency functions for the local runoffs. These dependencies are considered already in some approaches, for example by the application of copulas. A disadvantage of such individual models is the large number of parameters. Beside this, a possible relation between dependency and geographic parameters of the catchment areas is not considered. Last but not least, the definition of a locale, discrete flood event by a continuous local runoff variable is difficult and can influence the entire model. The concepts of random fields and max-stable random fields are the basic for a general formulation for the spatial dependency between local runoffs. The concept of the parameter estimation for a max-stable random field, the statistical analysis of block maxima, can overcome the problem of event definition. This is shown by the data analysis of Thuringia. Limits of the existing stochastic models and the issues of parameter estimation are also discussed.

60

SESSION: PATTERNS

A NEW QUALITATIVE FIELD METHOD TO CAPTURE SHALLOW SOIL MOISTURE PATTERNS IN WET ENVIRONMENTS

Rinderer, M., Kollegger, A., Fischer, B. and Seibert, J.

Department of Geography, Hydrology and Climate, University of Zurich, Winterthurerstr. 190, CH-8057 Zurich, Switzerland E-mail: [email protected]

Soil moisture patterns and their spatio-temporal dynamics are of central interest in catchment hydrology. It has been demonstrated, that information about these can give insightful information on dominant runoff processes. Common measuring techniques are time consuming and/or destructive (e.g., gravimetric method) or need sometimes expensive instruments (Time Domain Reflectometry, TDR). Furthermore in very wet soils with a high organic content TDR often does not provide reliable results. Here we present a qualitative field method, which has been developed to offer a supplement to common quantitative measuring techniques as it allows a quick and inexpensive mapping of soil moisture conditions. This soil wetness classification scheme is based on qualitative but well-defined topsoil indicators which can easily be mapped in the field. The scheme consists of seven wetness classes from ‘very wet’, when water is visible at the soil surface to ‘very dry’, when a person could sit at that spot for a longer time period without getting wet trousers. A test of the new method with in total 18 people showed that soil wetness classes could be reasonably well assigned to all 52 sampling points. The test persons, who did not have any prior expert knowledge of soil moisture mapping, could classify all 52 points within 20 to 30 minutes. Analysis of the test data set showed a small number of classification errors compared to a reference classification mapped by an ‘expert’ when assigning the driest or wettest class. Slightly higher variability could be seen for the classification of intermediate sampling points. Overall, the test showed the potential of the new qualitative method. Further tests were performed to link the qualitative classes to quantitative values of soil water content. This promising qualitative method does not intent to replace common quantitative methods but offers a potential to supplement them with the capability of covering a large number of spatially distributed sampling points. These spatial soil moisture datasets offer a potential to constrain model parameters and to study hydrological processes and their spatial organization.

61

SESSION: PATTERNS

COMPARISON AND COMBINATION OF FLOOD REGIONALIZATION METHODS IN SAXONY, GERMANY

Salinas, J.L.1, Merz, R.2, Laaha, G.3, Walther, J.4 and Büttner, U.5

1 Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Vienna, Austria 2 Helmholtz Centre for Environmental Research, Halle, Germany 3 University of Natural Resources and Life Sciences, Vienna, Austria 4 DHI-WASY, Dresden, Germany 5 Saxonian Ministry of Environment and Agriculture, Dresden, Germany E-mail: [email protected]

The spatial extension of flood frequency estimations is of vital importance in practical engineering purposes, particularly in the case of prediction in ungauged catchments. From a theoretical point of view there have been many different approaches, which could be classified in two main groups: pooling and interpolation/regression methods. In this case study in Saxony (Germany), the performance of different approaches was first separately compared: Index-Flood method belonging to the first group; georegression and Top-kriging to the second. On a second step, different combinations of the methods were analyzed with different goodness-of-fit metrics, obtaining in some cases better results than with the original methods.

62

SESSION: PATTERNS

COUPLING TOPOLOGICAL AND CANONICAL KRIGING FOR DESIGN-FLOOD PREDICTION IN UNGAUGED BASINS IN THE SOUTHEASTERN UNITED STATES

Stacey A. Archfield1, Castellarin, A.2, Jon O. Skøien3 and Julie E. Kiang4

1 Massachusetts-Rhode Island Water Science Center, U.S. Geological Survey, Northborough, MA,US 2 Department DICAM, School of Civil Engineering, University of Bologna, Bologna, Italy 3 Institute for Environment and Sustainability, Joint Research Centre, European Commission, Italy 4 Office of Surface Water, U.S. Geological Survey, Reston VA E-mail: [email protected]

Spatial interpolation techniques of point data have been shown to be effectively applicable for predicting streamflow statistics (i.e. flood flows and low-flow indices) in ungauged basins. Literature reports successful applications of two techniques, Canonical kriging (or physiographical-space based interpolation, PSBI) and Topological kriging (or Top-kriging). PSBI performs the spatial interpolation of the streamflow statistics of interest in the two- dimensional space of catchment descriptors. Top-kriging predicts the index along river networks taking both the catchment area and nested nature of catchments into account. A recent comparative analysis shows that PSBI and Top-kriging have complementary performance, that is Top-kriging outperforms PSBI at larger river branches, and vice-versa for headwater catchments. We focus on the possibility to couple PSBI and Topological kriging for enhancing the prediction of 10-, 50-, 100- and 500-year floods in ungauged conditions using a set of gages located across the southeastern US, which we simulate through a leave-one-out cross-validation procedure. In particular, the analysis is twofold and symmetric: (a) Top-kriging is initially used to estimate the flood-frequency statistics and PSBI is then applied to the residuals and used to adjust the Top-kriging estimates; (b) PSBI is initially used to estimate the flood-frequency statistics and Top-kriging is then applied to the residuals and used to adjust the PSBI estimates. Our preliminary analysis aims at assessing whether or not coupling PSBI and Top-kriging improves the performance of each methodology applied as a standalone technique and, if so, providing users with some indications on how most effectively to couple them.

63

SESSION: PATTERNS

POTENTIAL OF HIGH RESOLUTION MODELLING OF RUNOFF GENERATION PROCESSES

Steinbrich, A. and Weiler, M.

Institute of Hydrology, University Freiburg, Germany E-mail: [email protected]

The Distributed RunOff Generation (DROGen) model was developed with the aim to quantify runoff formation processes with respect to floods at a height temporal and spatial resolution (up to 1m²). The model is GIS based and not calibrated at all. All required parameters are derived from geo data sets and experience, obtained from many experimental studies carried out worldwide with respect to runoff generation processes. One benefit of the model is that important runoff generation processes, as infiltration and the subsurface flow via macropores and preferential flow pathways that often are neglected in models, are implemented based on land-use and soil properties. A further advantage of the model is its GIS based architecture that enables the modeller to visualize and analyse all model states at every time step in very height spatial resolution. Since the model is adapted to the unique conditions of its spatial environment by the information incorporated with the geo data, it is expected to reproduce runoff generation and accumulation correct without a priori information about precipitation - runoff relationship as commonly used when models are calibrated. The potential of the new model was tested with several applications. Its capability to reproduce observed flood hydrographs was tested in eight meso-scale catchments situated in different regions (geology, soils, land-use) of the state on Baden-Württemberg. Short with height intensity and long-lasting rain events with moderate intensities and different antecedent moisture conditions were used to test the models ability to reproduce different runoff generation mechanisms. It was found that especially the dynamics of the observed flood hydrographs was well reproduced highlighting the eligibility of the model to correctly account for the flood formation processes. An important potential of the uncalibrated model when applying to catchments in different regions was the ability to test the correctness of the modelling concepts and representation of processes when the model was not reproducing the observed hydrographs correctly under certain antecedent and event conditions. Example given the occurrence of tile drainage or impermeable soil horizons in podsolic soils, what was not represented by the used geo data. The model was also applied to the whole area of the state of Baden-Württemberg (~ 36,000 km²) at a resolution of 5*5m² using several scenarios representing different types of rainfall events and antecedent soil moisture conditions. The results were visualised for the typical pattern of runoff generation processes and also the dominant processes responsible in a catchment for floods under various events and antecedent conditions.

64

SESSION: PATTERNS

A THREE STEP APPROACH TO ANALYSIS OF FLOOD HAZARDS IN COASTAL BASINS: CASE SAR-ULLA RIVER, GALICIA-SPAIN

Vegas, E.1, García, E.2, Álvarez, C.1,2 and Listen, M.1

1 University of Cantabria, 2 Environmental Hydraulics Institute IH Cantabria, Avenida de los Castros s/n, 39005 Santander, Cantabria, Spain. E-mail: [email protected]

Flood events in Galicia (northern Spain) are associated with long duration rainfall episodes. The confluence of Sar and Ulla rivers has a long record of catastrophic floods. The study area is characterized by the proximity of the sea, scarce stream gauge data, complex terrain and a variety of land uses. Due to these characteristics, a three step approach is proposed to analyze flood hazards. The first step consists in a frequency analysis of rainfall to derive the Intensity-Duration-Frequency curves. These curves are used to obtain a particular rainfall event (project rainfall) and simulated discharge data using a distributed hydrological model previously calibrated and validated. In the second step, the primary outputs become the input for a dynamic 2D hydraulic model which simulates the effects of the flood in the confluence between the rivers Sar and Ulla, i.e. flow levels velocities and large-scale patterns. The third step consists in a flood hazard assessment based on several criteria, some proposed by the authors, and some others suggested by different water management organizations, as the Australian Commonwealth Scientific and Research Organisation (CSIRO). Hazard maps have been evaluated for both steady and unsteady hydraulic simulations, and the results vary notably due to dynamic effects. The main conclusions are that the flood hazard results are very sensitive to the formula applied, and also depend largely on the variability of velocities and water levels during the flood event. In summary, this three-level integrated approach, with an especial focus on hazard evaluation, facilitates the assessment of flood risks in complex and scarce gauged basins.

65

SESSION: PATTERNS

EXTREME RAINSTORMS: TESTING REGIONAL ENVELOPE CURVES AGAINST STOCHASTICALLY GENERATED EVENTS

Viglione, A.1, Castellarin, A.2, Rogger, M.1, Merz, R.3 and Blöschl, G.1

1 Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Vienna, Austria 2 DISTART, School of Civil Engineering, University of Bologna, Bologna, Italy 3 Helmholtz Centre for Environmental Research, Halle, Germany E-mail: [email protected]

The Depth-Duration Envelope Curves (DDEC) are regional upper bounds on observed rainfall maxima for several durations. Recently, a probabilistic interpretation has been proposed in the literature in order to associate a recurrence interval T to the DDEC's and, consequently, to retrieve point rainfall quantiles for ungauged sites. Alternatively, extreme rainfall quantiles can be retrieved from long synthetic rainfall series obtained with stochastic rainfall generators calibrated to local time series of rainfall events. While DDEC's are sensitive to outliers and data errors, the stochastic rainfall generator performance is affected by the limited record lengths used for calibration. The objective of this study is to assess the reliability of the two alternative methods by verifying if they give consistent results for a wide study region in Austria. Relative to previous studies, we propose some generalisations of the DDEC procedure in order to better represent the Austrian data. The comparison of rainfall quantiles estimated with the two methods for large T shows an excellent agreement for intermediate durations (from 1 h to 6 h), while the agreement worsens for very short (15 min) and long (24 h) durations. The results are scrupulously analyzed and discussed in light of the exceptionality of rainfall events that set the regional envelopes and the characteristics of the stochastic generator used. Our study points out that the combined use of these regional and local methods can be very useful for estimating reliable point rainfall quantiles associated with large T within regions where many rain gauges are available, but with limited record lengths.

66

SESSION: PATTERNS

MECHANISMS OF CLIMATE CHANGE EFFECTS ON FLOODS

Viglione, A.1, Salinas, J.L.1, Merz, R.2 and Blöschl, G.1

1 Institute of Hydraulic Engineering and Water Resources Management, Vienna University of Technology, Vienna, Austria 2 Helmholtz Centre for Environmental Research, Halle, Germany E-mail: [email protected]

Understanding climate impacts on floods is essential for developing climate adaptation strategies. The traditional scenario method of climate impact studies is problematic due to the large uncertainties which are difficult to estimate. Moreover, it is rarely clear how the uncertainties in the assumptions propagate to the results. The focus of this work is on the mechanism to allow a more nuanced and transparent assessment of cause-effect than is possible by scenarios alone. This allows us to separate changes that are likely to occur (hard facts) from changes that are possible but not supported by data evidence (soft facts). For instance, we found that some mechanisms allow us to suggest likely changes of floods with some confidence, i.e., the increase of winter floods due to higher temperatures (rising snow fall lines) and the decreasing summer floods due to earlier snow melt. Other mechanisms, like future changes in convective precipitation, remain elusive.

67

SESSION: PATTERNS

BIVARIATE FLOOD ESTIMATION BY DIFFERENT THRESHOLD MODELS

Zempléni, A.1 and Bálint, G.2

1 Department of Probability Theory and Statistics, Eötvös Loránd University, Budapest 2 VITUKI Enviromental Protection and Water Management Research Institute E-mail: [email protected], [email protected]

Univariate modelling of floods is a well-understood question, with sound theoretical background. One may choose either the method of annual maxima and the extreme value distributions or the peaks-over-threshold models and the corresponding Generalised Pareto Distribution (GPD). However, in practice the joint investigation of two different gauges is also important (because of the more difficult defence or the dangerous downstream floods caused by such simultaneous events). There are different approaches for solving the above problem of bivariate modelling. The annual maxima approach is not advisable here, most of all because of the completely different time points of the two maxima. The peaks-over threshold (MGPD) models are more promising. One of these approaches defines an observation extreme only, if both components exceed the threshold (MGPD type I), while the other considers all observations that are extreme in at least one component (MGPD type II). In the talk I shall introduce parametric models for both cases and illustrate their potential by applying them to Hungarian flood data from the Danube.

68

SESSION: PATTERNS

ON THE RELATIVE ROLES OF HILLSLOPE PROCESSES AND RIVER NETWORK IN THE HYDROLOGIC RESPONSE TO SPACE-TIME VARIABLE RAINFALL FIELDS

Zoccatelli, D. and Borga, M.

Department of Land and Agroforest Environment, University of Padova, Legnaro, Italy. Email: [email protected]

This paper develops a new framework to analyze the relevance of hillslope with respect to channel processes in the generation of the hydrologic response to space-time heterogeneous rainfall fields. The dependence between rainfall spatial organisation, the geomorphic features of channels and hillslopes, and the basin response, is examined by suitably extending the concept of ‘Spatial moments of catchment rainfall’. The moments describe the spatial rainfall organisation in terms of concentration and dispersion statistics as a function of the distance measured along the flow routing coordinate. The introduction of these statistics permits derivation of a simple relationship for the quantification of storm velocity at the catchment scale. The first order moment controls the timing of the flood hydrograph, whereas the second order moment and the storm velocity statistics both control the temporal dispersion of the runoff hydrograph (i.e., the amplitude). The different processes taking place in hillslopes and channel network are represented by using two different values of celerity in hillslopes and channels. The effects of hillslope processes on the hydrologic response to space-time heterogeneous rainfall fields is examined by varying the ratio between the two runoff celerities. Applications are reported to five extreme flash floods occurred in various European regions in the period 2002-2007. High resolution radar rainfall fields and a distributed hydrologic model are employed to examine how effective are the derived statistics in describing the degree of spatial rainfall organisation which is important for runoff modeling with varying the role of hillslope contribution. Results are obtained concerning the influence of hillslope contribution on the first two moments and on the storm velocity statistics. It is shown that, with increasing the influence of the hillslope contribution to the catchment response, the influence of the space-time rainfall variability on flood response is progressively smoothed. The main controls of this influence are identified, quantified and discussed.

69

SESSION: PATTERNS

70

EGU Leonardo Conference Series on the Hydrological Cycle

SESSION: PREDICTIONS

Conveners: G.T. Aronica, H. Madsen, B. Merz, R.J. Moore

Flood risk predictions are characterized by considerable uncertainty that needs to be evaluated and clearly communicated to decision-makers. For that reason the session dealt with: • Estimation of uncertainty in extremes of rainfall, discharge, overbank flow and inundation extent for flood risk predictions and the characterisation of their uncertainty; • Predictions in catchments for which little or no measured data are available, with a particular focus on reduction and characterisation of uncertainty in those predictions; • Communication of flood risk and uncertainty to community and national decision- makers.

71

72

SESSION: PREDICTIONS

TOWARDS A SEAMLESS FLOOD EARLY WARNING SYSTEM: FROM FLASH FLOODS TO GLOBAL FORECASTING

Alfieri, L.1, Salamon, P.1, Pappenberger, F.2, Burek, P.1, Krzeminski, B.2, Muraro, D.1, Thielen, J.1 and de Roo, A.1

1 European Commission - Joint Research Centre, Institute for Environment and Sustainability, via E. Fermi, 2749, 21027 Ispra (VA), Italy 2 European Centre for Medium Range Weather Forecasts, Shinfield Park, Reading RG2 9AX, UK E-mail: [email protected]

Ongoing changing climate has raised the attention towards weather driven natural hazards. Flash floods and debris flows following exceptional downpours often come without any adequate warning and cause heavy tolls to the human society. Also, recent large-scale meteorological events have shown the catastrophic impact of floods in large river basins, often affecting developing countries with scarce infrastructures for flood prevention and mitigation. Current Numerical Weather Predictions (NWP) can skillfully forecast extreme meteorological events, several days in advance. Yet, operational flood early warning systems based on hydro-meteorological simulation often do not make use of the full capabilities of state-of-the-art NWP. Besides, their coverage is mostly limited to country-wide extents, even within trans-national river basins, leading to very heterogeneous level of preparedness to extreme flood events. Focus of this work is to give an overview of operational and experimental flood warning systems currently being developed at the Joint Research Centre of the European Commission, which tackle different flood processes through similar approaches, though at different space-time scales and geographic extents. An experimental system for extreme precipitation events and flash flood early warning is run on a daily basis using 5-day limited-area ensemble forecasts covering most of Central and Southern Europe. It is based on the detection of extreme precipitation events of short duration, through comparison with a 20-year reforecast climatology, consistent with the input forecasts. The European Flood Alert System is running pre-operationally since 2003 and focuses on riverine floods in mid-size to large river basins in Europe. Streamflow simulations are produced through distributed hydrological modeling of 120 ensemble weather forecasts every day. In addition, a prototype of a Global Flood Alert System was set up in 2011 aimed at detecting large river floods with lead time as long as one month. This system has been run continuously on global coverage with ensemble hindcasts from January 2010 provided by the European Centre for Medium-Range Weather Forecasts (ECMWF). A number of case studies are shown to demonstrate strengths and weaknesses of the different systems. Although the three systems have limited requirements in terms of local measurements, a major challenge is faced in the scarce availability of observed discharge data for comparison, particularly in remote unsettled regions and in ungauged catchments. Finally, we make use of the acquired experience to comment on ongoing and future challenges of operational warning systems.

KultuRISK and IMPRINTS FP7 projects are gratefully acknowledged for the financial support to the research work.

73

SESSION: PREDICTIONS

THE MULTIVARIATE STATISTICAL ANALYSIS OF THE EXTREME HYDROLOGICAL EVENTS

Bačová Mitková, V.

Institute of Hydrology SAS, Račianska 75, 818 11 Bratislava, Slovak Republic E-mail: [email protected]

One-dimensional analyses of flood characteristics are now mainly used in studies of hydrological extreme events. It means that the peak flow of flood wave or volume of flood wave is statistically analyzed separately. A drawback of this approach is that the same family of marginal distributions is assumed for all flood characteristics, while analysed variables could show different margins. Hydrological phenomena are often multidimensional and hence they require the joint modeling of several random variables. Therefore, for more comprehensive analysis of the flood wave it is necessary to know and understand relationship and dependence between the characteristics of the flood wave, what requires a multidimensional statistical approach. For the multidimensional distribution copula functions are usually used. The concept of copulas relaxes the restriction of traditional flood frequency analysis and connects one-dimensional marginal distribution of random variables with their joint distribution. The aim of this paper is to analyze and evaluate statistically the occurrence of hydrological variables such as flow and volume of the phenomenon. In our work we focus on the use of copula function in multidimensional statistical analysis of maximum annual discharges and their assigned volumes of the flood waves on the Bodrog River in Streda nad Bodrogom gauging station during the period 1950-2009. Some of three Archimedean copulas were used for the analysis: Clayton copula, Gumbel copula and Frank copula. The parameter theta of the each Archimedean copulas can be obtained by considering the mathematical relationship between Kendall`s coefficient of correlation and generating function. The simulations of the 1000 pairs of points from copulas were analyzed. Results obtained by the multidimensional analysis of the variables, which characterize the hydrological waves (flow, volume, time) can contribute to more reliable assessment of flood risks. They give an overview of the flood event as a whole and practical use of these results can be in water management and in the design of flood protective systems.

74

SESSION: PREDICTIONS

ON THE JOINT ESTIMATION OF FLOOD FREQUENCY DISTRIBUTIONS AT SEASONAL AND ANNUAL TIME SCALE

Baratti, E.1, Montanari, A.1, Castellarin, A.1, Bezzi, A.2, Di Giulio, A.2 and Ruzzante, E.2

1 Department DICAM, University of Bologna, Via del Risorgimento 2, I-40136 Bologna, Italy 2 Studio ing. G. Pietrangeli Srl, Via Cicerone 28, I-00193 Rome, Italy E-mail: [email protected], [email protected], [email protected]

Flood frequency distribution is usually estimated at annual time scale, by fitting observations of annual maximum peak lows or peak flows exceeding a given threshold. However, in many cases one is also interested in inferring the seasonal behaviour of the flood frequency distribution, for instance when one needs to estimate the risk of flood in different periods of the year. Seasonal flood frequency analysis has been rarely investigated. A key issue is to ensure the compatibility between seasonal and annual flood frequencies. We propose an approach to jointly estimate the parameters of intra-annual and annual probability distribution of floods. The approach is based on the preliminary identification of an optimal number of seasons within the year, which depends on the climatic and hydrological behaviours of the investigated watershed. Then, parameters of intra-annual and annual flood distributions are jointly estimated by using (a) an approximate optimization techniques and (b) a formal maximum likelihood approach. The proposed methodology is applied to the case study of the Blue Nile River, for which extended hydrological information is available at annual and seasonal scale.

75

SESSION: PREDICTIONS

MODELLING OF CLIMATE CHANGE IMPACTS ON FLOODS BY CONTINUOUS SIMULATION WITH UNCERTAINTY

Blazkova, S.1, Beven, K.2 and Martinkova, M.1

1 T.G. Masaryk Water Research Institute, Hydrology, Prague 6, Czech Republic 2 LancasterEnvironment Centre, Lancaster, UK E-mail: [email protected]

The climate change is supposed to affect the occurrence of climate extremes. The presented study is focused on estimation of climate change impacts on floods while using the outputs from regional climate models of ENSEMBLES data set (http://ensemblesrt3.dmi.dk/). The methodology used in this study is continuous simulation (i.e. the input is a series from a precipitation simulator) within the uncertainty framework (many behavioural simulations are taken into account). With the behavioural parametersets computed in the study, 100 thousand year series are modelled and hydrographs of large floods are extracted from them. A special attention is paid to the initial conditions before extremely heavy rains. The problem to whatdegree the volume and shape of the flood hydrograph is affected by a drought of various depths on a catchment inessentially wet Central European climate is analyzed. This research is supported by the Czech Grant Agency (P209/11/2045) and by the Czech Ministry of Education (OC10024 and 7A08036).

The present study is a follows up of the paper of Blazkova and Beven (2009).

Blazkova, S., and K. Beven (2009), A limits of acceptability approach to model evaluation and uncertainty estimation in flood frequency estimation by continuous simulation: Skalka catchment, Czech Republic, Water Resour. Res., 45, W00B16, doi:10.1029/2007WR006726.

76

SESSION: PREDICTIONS

PERSISTENCE AS A FACTOR OF FLOOD HAZARD FOR EMBANKED RIVERS

Bogdanowicz, E.1 , Strupczewski, W.G.2 and Kochanek, K.2

1 Institute of Meteorology and Water Management, Podlesna 61, 01-673 Warsaw, Poland. 2 Department of Hydrology and Hydrodynamics, Institute of Geophysics, Polish Academy of Sciences, Ksiecia Janusza 64, 01-452 Warsaw, Poland. E-mail: [email protected], [email protected], [email protected]

Sense of security in floodplains of embanked rivers results from the belief that levees protect against flood magnitude on which they were designed. So, if the actual forecasted flood peak does not exceed the safety levels related to Levee designed value one can assume that the risk that flood water will overtop the dike crest is negligible and so is the risk of flooding in the protected area. The history of floods shows this is not true. An important factor to be taken into consideration is the duration of high water levels. Long-lasting high stages may weaken the levee structure (soaking) and cause dangerous leaks, blurs and breaks that threaten their destruction. The break usually happens during the falling phase of flood hydrograph. That’s why the classical FFA concerning only the frequency of the maximum flows is not suitable in this case and ought be extended to the analysis of the duration of flows and corresponding states over the given threshold. In Poland, as in many countries in the world, for each hydrological station two benchmark water levels, called the warning stage and the alarm stage, have been specified. The frequency of annual maximum uninterrupted duration, D (in days), of flows over the flood alarm stage can be regarded as the measure of the risk of flooding due to waning of the levees' strength. This study introduces formal aspects of the Duration–flow–Frequency (DqF) modeling in stationary and non-stationary conditions. To cater for the conventional FFA, the flow discharge corresponding to the alarm stage is used here, i.e. the upper limb of the rating curve is regarded as time invariant. In the presented statistical model, the duration (D) is considered as a random variable while the alarm flow discharge is the fixed value.

77

SESSION: PREDICTIONS

DESIGN FLOOD PROFILES ESTIMATION: FREEBOARD VERSUS UNCERTAINTY BOUNDS

Brandimarte, L. and Di Baldassarre, G.

UNESCO-IHE, Institute for Water Education, Delft, The Netherlands E-mail: [email protected], [email protected]

The prediction of flood defence design profiles has always been associated with some degree of uncertainty. The traditional deterministic approach takes uncertainty into account by adding a freeboard to the simulated flood profile; the more recent probabilistic approaches have provided sophisticated techniques for taking into account different sources of uncertainty in hydraulic modelling and flood frequency analysis. Nevertheless, in engineering practice, the deterministic approach, based on the outcomes of a well-calibrated model, is still preferred to the probabilistic one. This might be due to the unfriendliness of the implementations of methods to quantify uncertainties from different sources. However, although easier to apply, the freeboard is often arbitrarily defined and may lead to a false perception of additional safety level. This paper proposes a pragmatic methodology to estimate design flood profiles and the associated uncertainty, which is based on the use of uncertain flood profiles: the most significant sources of uncertainty are explicitly analyzed by using a simple to apply approach. An application to the Po river reach between Cremona and Borgoforte (Italy) is used to illustrate the proposed framework and compare it to the traditional freeboard approach.

78

SESSION: PREDICTIONS

PROBABILISTIC FORECASTING OF ANTECEDENT SOIL MOISTURE CONDITION AS FLASH FLOOD PRECURSOR VARIABLES

Brigandì, G.1, Aronica, G.T.1, Loukas, A.2, Vasiliades, L.2 and Papaioannou, G. 2

1 Department of Civil Engineering, University of Messina, Messina, Italy 2 Department of Civil Engineering, University of Thessaly, Volos, Greece E-mail: [email protected]

Flood forecasting is a rather complicated task, particularly in those catchments which are prone to flash flood formation or which the response time is of the order of few hours and, even brief anticipation are important and welcomed. In this context, some kind of hydrological precursors can be considered to improve the effectiveness of the emergency actions (i.e. early flood warning). Now, in literature has been widely recognized how soil moisture is an important factor in flood formation, because the runoff generation is strongly influenced by the antecedent soil moisture conditions of the catchment. The basic idea of the work here presented is to use soil moisture conditions in a probabilistic framework to define a first alert phase in a flash flood forecasting chain. For the soil moisture conditions modelling, the IHACRES model which is a spatially lumped rainfall-runoff model that can be employed to reproduce the continuous daily response of the catchment was used to derive a series of wetness indexes. Soil moisture conditions are defined using an Antecedent Moisture Conditions index (AMC) similar to this widely used for the implementation of the Soil Conservation Service – Curve Number methodology. Instead, for the soil moisture conditions forecasting a Markov chain model has been implemented and tested. Application of the proposed methodology has been carried out with reference to a river basin in Cyprus.

79

SESSION: PREDICTIONS

FLOOD PROTECTION OF THE CITY OF LJUBLJANA AND EXPERIENCES GAINED DURING THE SEPTEMBER 2010 FLOOD

Brilly, M., Vidmar, A. and Rusjan, S.

University of Ljubljana, Faculty of Civil and Geodetic Engineering, Chair of Hydrology and Hydraulic Engineering E-mail: [email protected]

In the contribution the flood protection problematic of the capital of the Republic of Slovenia, the Ljubljana city, will be presented. Ljubljana lies in the southern part of the Ljubljana basin, crossing the moor on the south and the north of Ljubljana field. The tectonic subsidence of the area in the geological past has made it an important confluence of the rivers. The area of the City of Ljubljana has a long history of various flood protection measures (e.g. first waterworks in the Ljubljanica River channel by the Romans, Grubar flood canal excavation in 1780 for diversion of Ljubljanica moor floodwaters away from the city center, weir construction on the Ljubljanica River in 1950s for floodwater manipulation and extended widening of the Mali graben channel in the 1970s). However, despite the abovementioned flood protection efforts, many parts of the urban area of the City of Ljubljana is presently heavily threatened by the floods as the one experienced in September 2010. The southern part of the city, particularly in the Ljubljana moor, is exposed to a risk of catastrophic, medium and even small flood events. In the northern part of Ljubljana, at the Sava River area, there is a risk of catastrophic medium flood events. Most heavily endangered is the southern part of the city in the vicinity of the Ljubljanica River and its tributaries. The western part between Podutik and Rožna dolina is endangered by Glinščica stream high waters and its tributaries, south western part of the city (the whole Vič area) by Gradaščica with Horjulka, southern part of the Rudnik suburbs with moor floodwaters and the central and northern part of the Rudnik by tributaries from Golovec and inland waters. The main reasons for the present insufficient flood protection of the City of Ljubljana lies especially in the discontinuities and mutual exclusion of flood protection measures planning and overall spatial development of the urbanized areas. As a consequence, some of the past flood protection measures could no longer function properly due to intensive urbanization of some areas (e.g. the Mali graben channel). On the other hand, some of the urbanized Ljubljanica moor areas are becoming increasingly flood vulnerable due to past long-term intensive moor area drainage and consequent slow decreasing of the overall ground levels. In the past twenty years, as environmentally friendly maintenance, riverbank transformed in bush, that also tremendously decrease flood discharge and increase flood in urban area.

80

SESSION: PREDICTIONS

NEW METHODS OF FLASH FLOOD FORECASTING AT THE CZECH HYDROMETEOROLOGICAL INSTITUTE

Březková, L., Novák, P., Jonov, M., Kyznarová, H., Šálek, M., Frolík, P. and Janál, P.

Czech Hydrometeorological Institute, Regional Forecasting Department, Kroftova 43, 616 67 Brno, Czech Republic E-mail: [email protected]

The Czech Hydrometeorological Institute is the primary agency responsible for the monitoring and forecasting of river stages at national level. For prediction of large scale floods the hydrological models using the data from NWP models are used in daily routine. However, the flash floods caused by heavy precipitation, which hit small catchmets in summer season, still remain almost unpredictable phenomena due to their rapid rise. In recent years, the precipitation estimates and nowcasting tools derived from the radar data were established. Together with the hydrological models these tools were tested for the flash flood forecasting. The high uncertainty of the predicting of such a type of phenomena leads to using the various nowcasting methods for estimation of predicted rainfall sum. This “variant-approach“ was successfully tested on case studies and was set up in testing operation for pilot catchments. In the paper the first experience with the new flash flood forecasting system is presented.

The work was supported by Central Europe Programme, INCA-CE project (co-financed by European Regional Developement Fund).

81

SESSION: PREDICTIONS

EFFECTS OF UNCERTAINTY IN BOUNDARY CONDITIONS ON FLOOD HAZARD ASSESSMENT

Domeneghetti, A.1, Vorogushyn, S.2, Castellarin, A.1, Merz, B. 2 and Brath, A.1

1 Department DICAM, School of Civil Engineering, University of Bologna, Bologna, Italy 2 Section 5.4: Hydrology, Deutsches GeoForschungsZentrum (GFZ), Potsdam, Germany E-mail: [email protected]

Comprehensive flood-risk assessment studies should quantify the global uncertainty in flood hazard estimation, for instance by mapping inundation extents together with their confidence intervals. This appears of utmost importance, especially in the case of flood hazard assessments along dike-protected reaches, where dike failures have to be considered. This paper focuses on a 50km reach of River Po (Italy) and three major sources of uncertainty in inundation mapping: uncertainties in the (i) upstream and (ii) downstream boundary conditions, and (iii) uncertainties in the dike-failure location and breach morphology. We derive confidence bounds for flood hazard maps by means of the Inundation Hazard Assessment Model (IHAM) – a hybrid probabilistic-deterministic model. IHAM couples in a dynamic way a 1D hydrodynamic model and a 2D raster-based hydraulic model through a probabilistic dike-breaching analysis that considers three different failure mechanisms: overtopping, piping and micro-instability due to seepage. To address the randomness resulting from the variability in boundary conditions and dike-failures the system is run in a Monte Carlo framework. Uncertainties in the definition of upstream boundary conditions (i.e. design-hydrographs) are assessed by applying different bivariate copula families to model the frequency of flood peaks and volumes. Uncertainties in the definition of downstream boundary conditions are characterized by associating the rating-curve used as boundary condition with confidence intervals which reflect discharge measurement errors and interpolation errors. The results of the study are presented in terms of the Monte Carlo-based flood hazard mapping for different flood-intensity indicators (e.g., inundation depth, flow velocity, inundation duration, etc.) together with the corresponding uncertainty bounds. We conclude on the influence of uncertainty in boundary conditions and provide decision makers with an important piece of information for devising and implementing flood-risk mitigation strategies.

82

SESSION: PREDICTIONS

THE VALUE OF URBAN FLOOD MAPPING: APPLICATION TO THE CASE STUDY OF BARCELONNETTE (FRANCE)

Dottori, F.1, Di Baldassarre, G.2 and Alfonso, L.2

1 University of Bologna 2 Unesco-IHE, Delft E-mail: [email protected]

Let us consider an urban settlement near a river, for which a flood alert system is installed. In case of a flood alarm, it is evident that people living close to the river should be evacuated anyhow, while people leaving very far away will not be affected by the inundation. In these two opposite situations, any additional information to make a decision (evacuate / not evacuate) has no value. However, in the flood prone areas between these two extremes, urban flood mapping has a value, providing information on the dynamics of the flood event and its consequences and therefore reducing the uncertainty associated to the event. In this context, the goal of this study is to analyze the value of urban flood mapping, using the Value of Information (VOI) concept. In particular, VOI is evaluated in terms of prevented damage with probabilistic flood inundation maps. Such maps are ensemble-based, for which a recently developed inundation model, based on cellular automata approach, is used. The methodology is applied to a specific case study: The Ubaye River at Barcelonnette (France Alps).

83

SESSION: PREDICTIONS

OPEN SOURCE GIS TOOLS FOR FLOOD HAZARD MAPPING WITHIN THE DANUBE FLOOD RISK PROJECT

Franceschi, S.1, Monacelli, G.1 and Rigon, R.2

1 ISPRA Italy (Higher Institute for Environmental Protection and Research) 2 University of Trento, Department of Civil and Environmental Engineering, Italy E-mail: [email protected]

The Danube River is one of the most important natural axes in South-East-Europe. It links most of the countries in the SEE area. The DANUBE FLOODRISK project focuses on the most cost-effective measures for flood risk reduction: risk assessment, risk mapping, involvement of stakeholders and risk reduction by adequate spatial planning. 19 institutions all along the Danube countries, central public bodies, universities, research institutes and operational agencies, NGOs are involved in the project. ISPRA (Higher Institute for Environmental Protection and Research in Italy) will give its contribution to the Danube Flood Risk Project focusing on a set of studies and analyses on small Alpine catchments of the study area of the Drau basin closed in Lienz. The attention will be focused on the influence that planning and water management in small basins can have on the floods in the main Danube River. For this purpose some algorithms and methodologies have been integrated into the new GIS library JGrassTools to be used in the project for hazard mapping, with particular attention to the evaluation of the maximum discharge for floods with high content of transported sediments and debris flow, as recommended by the European Flood Directive 2007/60CE. The main reason for choosing the Drau basin as study area is to compare different approaches and tools for hazard mapping. Therefore during the project specific tools will be developed and the achieved results will be compared with those obtained with other proprietary software. The development of GIS tools is a good starting point for data harmonization and integration between different countries and also for stakeholders’ involvement because of the use of maps and colours for the presentation of results to the public and because technicians and administrations can use the same tools for their analysis. These tools are in fact used also in some academic and post academic courses on hydro-geological risk evaluation. The final purpose of the work is to have a set of well working Open Source tools that can be used for planning, data management and integration and stakeholders’ involvement. These tools will make the exchange of data and knowledge between neighbouring countries easier and, on the other hand, they will be a guarantee of the quality of the work done by professionals for the public administrations.

84

SESSION: PREDICTIONS

UNCERTAINTY ESTIMATION OF SIMULATED WATER LEVELS FOR THE MITCH FLOOD EVENT IN TEGUCIGALPA

Fuentes, D. C.1,2, Halldin, S.1, Xu, C-Y.1,3 and Lundin, L-C. 1

1 Department of Earth Sciences, Uppsala University, Villavägen 16, SE-752 36 Uppsala, Sweden 2 Earth Science Institute, Universidad Nacional Autónoma de Honduras, Blv. Suyapa, Ciudad Universitaria, Tegucigalpa, Honduras 3 Department of Geoscience, University of Oslo, P O Box 1047, Blindern, NO-0316, Oslo, Norway E-mail: [email protected]

Hurricane Mitch in 1998 left a devastating flood in Tegucigalpa, the capital city of Honduras. Simulation of elevated water surface provides a better way to understand the hydraulic mechanism of large flood events. In this study the one-dimensional HEC-RAS model for steady flow conditions was used to estimate the water level for the Mitch event in the river reaches at Tegucigalpa. Parameters uncertainty of the model was investigated using the generalized likelihood uncertainty estimation (GLUE) framework. Because of the extremely large magnitude of the Mitch flood, no hydrometric measurements could be taken during the event. However, post-event indirect measurements of the discharge and observed water level were obtained in some previous studies. To overcome the problem of lacking direct hydrometric measurement data, uncertainty in the discharge was considered by sampling the value from a beta distribution function. Furthermore, the fuzzy set theory was used to provide the model with flexibility in selecting behavioral parameter sets by considering an uncertainty range in the post-event observed water level. If behavioral parameter sets can reproduce water surface levels well for past events such as Mitch, more reliable predictions for future events can be expected. The results acquired in this research will provide guidelines to deal with the problem of modeling past floods when no direct data was measured during the event, and to predict future large events taking uncertainty into account. The obtained range of the uncertain flood extension will be an outcome useful for decision makers.

85

SESSION: PREDICTIONS

INTEGRARTION OF TIME SERIES MANAGEMENT SYSTEM AND DATA SERVING FOR MODELLING APPLICATIONS

Gál, R.

KISTERS AG, Charlottenburger Allee 5, 52068 Aachen, Germany E-mail: [email protected]

The core technology, which provides the backbone of key services for water management data processing is called KISTERS Time Series Management (KiTSM). KiTSM is developed in JAVA and is designed to organize, compute and share time series mass data in local networks but as well in the Internet. KiTSM is combining interdisciplinary demands on time series processing with regard to mass data capabilities, scalability, modular design and flexibility to work in diverse specialist areas, high level of automation, reliability, security, integration potential, broad platform independence, redundancy and resilience. Standard time series data structure contains time stamp, value, quality and interpolation type. Time interval of time series can be regular or flexible, with possibility to add time offset. Between two points is clearly defined the interpolation type, which could be changed over the time. All calculations and auto validation routines comprise the organic part of execution system. Data operations as algorithms, called Agents, are operating on time series data and their main purpose is to either auto validate time series or derive other time series. Agents are generally triggered on data change events and executed automatically each time the data is accessed (such as by reports, graphs, exports, calculations); with KiTSM time series are also automatically updated in the background. WISKI as a client application desktop tool is connected to all metadata and time series data. Key lists and basic data are organised in different hierarchies. Environment of graphical view allows providing various types of data corrections, arranging graphical layout and elements as well as print, copy or save the graph. Extreme events and exceptional situations are controlled by AlarmManager. It handles incoming event messages, classifies them, and reacts accordingly. Alarm messages are created based on templates, then filled with additional information, and are distributed to recipients via several media types. Relevant hydrological data are available online in a web browser using WISKI Web Public which disseminates data for general public covering peak accesses during flood situation. Requirements of specific expert groups are fulfilled by WISKI Web Pro which is highly interactive and works with dynamic filtering, mapping, lists and graphs. The KISTERS Web Interoperability Solution KIWIS has been developed as a contribution to interoperability experiments at Open Geospatial Consortium (OGC). KIWIS offers combined services such as SOS, WOF, WMS and WFS for different data sources in one instance. As a standard requirement the KISTERS archives have to serve data to a range of modelling applications. In the majority of cases this data transfer is scheduled in time and is based upon a variation of model specific ASCII file formats. To ease this process and to provide the advantage of linking models to our customers the KISTERS time series server became OpenMI compliant by the end of 2010. With the KISTERS OpenMI wrapper, local and/or remote modelling users can establish the communication to a KISTERS time series server, search and identify the appropriate input time series and retrieve data through the internet directly into a chain of integrated modelling applications.

86

SESSION: PREDICTIONS

ASSESSMENT OF PROJECTED CHANGES IN PRECIPITATION EXTREMES IN THE CZECH REPUBLIC USING A NON-STATIONARY INDEX FLOOD MODEL

Hanel, M., Vizina, A. and Martínková, M.

T. G. Masaryk Water Research Institute, p.r.i., Prague, Czech Republic E-mail: [email protected]

Fourteen transient simulations of regional climate models (RCMs) are used to assess the possible future changes in seasonal precipitation extremes of durations 1 to 30 days for the Czech Republic. The statistical model used for analysis assumes that these precipitation maxima follow a generalized extreme value (GEV) distribution with time-varying parameters. In addition, spatial pooling of the maxima at various grid boxes is applied to reduce the uncertainty of the parameters of the GEV model. The ensemble mean changes in the quantiles are positive for all durations, return periods and seasons considered. Large quantiles of the distribution of summer precipitation extremes increase more than moderate quantiles and this increase is relatively large at short durations. The dependence of the changes on duration and return period is much weaker in the other seasons, in particular in winter and spring. Although the spread in the RCM ensemble is considerable, most of the RCM simulations agree on the sign of the changes. The value of RCM projections of hourly precipitation extremes is also briefly addressed.

87

SESSION: PREDICTIONS

SAFEFTY CRITERIA FOR FLOOD PROTECTION STRUCTURES IN SLOVENIA

Humar, N.1, Kryžanowski2, A., Brilly, M.2, Schnabl, S.2

1 Hidrotehnik d.d. Ljubljana, Slovenčeva 97, SI-1000, Ljubljana, Slovenia 2 Chair of Hydrology and Hydraulic Engineering, Faculty of Civil and Geodetic Engineering, University of Ljubljana, Jamova 2, Si-1000 Ljubljana, Slovenia E-mail: [email protected], [email protected], [email protected], [email protected]

In Slovenia, several structures (e.g., dams, levees, floodwalls, etc.) which intend to protect the affected flood prone areas have been built recently. Most of these structures have been built in areas where failure would potentially results in massive damage or great loss of lives. As a result, all these structures have to be designed, constructed, maintained, and inspected appropriately. Thus, in this paper, design/safety criteria and practices used in relation to these structures are reviewed and critically compared to other national standards applicable to similar structures. Finally, based on this comparison, some improvements of the safety criteria/standards adopted in Slovenia will be suggested.

88

SESSION: PREDICTIONS

SPRING FLOOD FREQUENCY ANALYSIS OF LITHUANIAN RIVERS

Kriaučiūnienė, J. and Šarauskienė, D.

Laboratory of Hydrology, Lithuanian Energy Institute, Breslaujos 3, LT-44403 Kaunas, Lithuania E-mail: [email protected]

Flood frequency analysis is useful in forecasting flood magnitude and behavior. Statistical prediction of design flood is very important for management of extreme flood control, prevention of present and planned hydro technical structures, hydropower stations and dams. Statistical methods of flood frequency analysis could be applied for dam safety analysis (flood-hazard mapping). Since small rivers often have short series of observation data or no data at all, there is a possibility to use available flood characteristics of other rivers of the same homogenous region, applying flood frequency distribution as well. The probabilistic behavior of floods has been studied for the Lithuanian rivers. Analysis of different flood frequency distributions (Gumbel, Pearson Type III, General Normal, Log Normal, Generalised Extreme Value) using long-term data of 32 hydrological gauging stations was performed in order to estimate the best fit distributions for Lithuanian conditions. The water measurement stations (WMS) were distributed in three hydrological regions of Lithuania: 1 – Western Lithuania, 2 – Central Lithuania, 3 – Southeastern Lithuania. Additionally the hydrological data from 5 WMS which are on the biggest rivers of Lithuania (the Nemunas and the Neris) were investigated. The results revealed that in 1961-1990 Log-Pearson type 3 probability distribution best fitted actual data of spring flood in the Western region (where precipitation is the main river feeding source) and GEV – for the Central region (that has mixed river feeding type) and Southeastern region (where groundwater feeding prevails); in the period of 1991-2008 GEV was the probability distribution that received the highest ranks in all regions. Recently, in the period of significant climatic changes (1991-2008), GEV probability distribution is the most suitable distribution for flood frequency analysis and could be used to predict of flood maximum discharges in Lithuanian rivers. This probability distribution could also be used for simulating flood frequency distribution in engaged rivers. LN and EV1 probability distributions seemed to be the least suitable (out of the five studied) for the modeling of the analyzed spring flood data.

89

SESSION: PREDICTIONS

ESTIMATION OF FLOOD FREQUENCY CURVES IN POORLY GAUGED MEDITERRANEAN WATERSHEDS USING A DERIVED DISTRIBUTION PROCEDURE

Loukas, A.1, Vasiliades, L.1, Papaioannou, G.1 and Aronica, G.T.2

1 Department of Civil Engineering, University of Thessaly, Pedion Areos, 38334 Volos, Greece 2 Department of Civil Engineering, University of Messina , Via Nuova Panoramica dello Stretto, 98166 Villaggio S. Agata - Messina, Italy E-mail: [email protected], [email protected]

For engineering applications of hydrology, the estimation of the peak discharge for a given return period at basin outlet is important because the planning and design of water resources projects and floodplain design depend on the frequency and magnitude of peak discharges. In ungauged and poorly gauged watersheds, it is not possible to apply statistical analysis of flood data or using a calibrated rainfall-runoff model of varying degrees of complexity. A way to circumvent the lack of data is to model the flood formation processes using simplified descriptions of the hydrological processes characterized by a reduced number of robust parameters in order to ensure a reduced uncertainty in model predictions. Such a modelling procedure is the derived distribution approach, which uses a rainfall generator and a simple rainfall-runoff model. In this work, an event rainfall-runoff simulation procedure based on the derived distributions method is proposed for the estimation of flood frequency in ungauged or poorly gauged watersheds in the Mediterranean. The procedure uses a stochastic generation rainfall generation model and a rainfall-runoff model. The results of previous research on rainfall characteristics and watershed response have been incorporated into the two models. The rainfall characteristics are storm depth, storm duration and the storm spatial and temporal distribution. The rainfall-runoff model is a simple model which describes the hydrological processes with a reduced number of parameters. Some of the parameters of the rainfall generation and hydrological model parameters are stochastic in nature and they are generated using the Monte Carlo method simulating the flood hydrographs. The proposed procedure has been applied in a watershed in Cyprus, and the results have been validated against the observed flood frequency.

90

SESSION: PREDICTIONS

CLIMATE CHANGE IMPACT ASSESSMENT OF FLOOD RISK. CASE STUDY: VIDAA RIVER BASIN

Madsen, H., Larsen, J. and Madsen, M.N.

DHI, Agern Allé 5, DK-2970 Horsholm, Denmark E-mail: [email protected]

A methodology is presented for flood risk assessment under climate change. The methodology is demonstrated for analysing changes in flood risk and dike safety in the Vidaa River basin, a cross-border catchment located in the southern part of Jutland, Denmark and northern Germany. The river discharges to the Wadden Sea through a tidal sluice, and extreme water level conditions in the river system occur in periods of high sea water levels where the sluice is closed and increased catchment runoff take place. To assess the impacts of future climate change both changes in the meteorological forcing (precipitation, temperature and potential evapotranspiration) and changes in sea water level are considered. Changes in meteorological forcing data are based on an ensemble estimate of downscaled regional climate model projections from the ENSEMBLES data archive. Future sea water levels are estimated from current projections of the mean sea level rise in the area, estimated isostatic changes, and changes in storm surge statistics predicted from a hydrodynamic model simulation forced with a regional climate model. To estimate the changes in flood risk in the Vidaa River basin an integrated hydrological and hydraulic model has been set up and calibrated. This model forms the basis for simulation of water levels in the river system using meteorological forcing and sea water level data for current (using observed records) and future climate (using projected records), considering two projection horizons, 2050 and 2100. Extreme value analysis is applied to estimate the risk of dike overtopping. More extreme precipitation events are expected in future, which result in an increase of the extreme runoff in the river basin of about 8% in 2050 and 14% in 2100 (ensemble averages). Hydrodynamic model results show increases in extreme storm surge levels at the Vidaa sluice of up to 0.8 m in 2100. However, durations of extreme water levels are not seen to increase. Currently, there is a relatively low risk of dike overtopping with annual exceedance probabilities of 0.1% or less in most parts of the river system. For the worst-case scenario in 2100, pronounced changes in flood risk are seen with flood risks of 5% or more in the downstream part of the river.

91

SESSION: PREDICTIONS

FLOOD RISK MAPPING UNDER UNCERTAINTY: APPLICATION TO SUNGAI JOHOR BASIN, MALAYSIA

Md. Ali, A., Di Baldassarre, G. and Solomatine, D.

UNESCO-IHE Institute of Water Education, Department of Hydroinformatics and Knowledge Management, Westvest 7, 2601 DA, Delft, The Netherlands E-mail: [email protected]

Flood risk can be defined as a combination of flood hazard, vulnerability and exposure. In particular, while flood hazard is usually derived by elaborating the outcomes of hydraulic models, flood vulnerability and exposure are commonly evaluated via socio-economic analyses, land use studies, and assessment of the capability of the population to deal with flood disasters (e.g. evacuation plans). Currently, flood risk mapping is often implemented without a serious consideration of the large number of uncertainty sources affecting the risk assessment. A slight uncertainty in estimating hydrologic flow can cause uncertainty in estimating the water surface elevation in a channel, which in turn can cause uncertainty in mapping the flood inundation. Choices of performance measures and observation sets could significantly influence the estimation of flood hazard. Given the current state of uncertainty and recent advances in various individual components, an integrated framework for flood inundation mapping is needed. Advances in technology, data collection techniques, and modelling approaches have led to improvements in overall simulation of hydrologic and hydraulic processes. The goal of this research is to identify and quantify the main sources of uncertainty associated with flood risk mapping and develop an efficient methodology to deal with them.

92

SESSION: PREDICTIONS

FLOOD IN TOWN – DECISION SUPPORT TOOL (VESNA)

Mišík, M., Kučera, M., Fridrich Tegelhoffová, M. and Mrnčo, I.

DHI SLOVAKIA, s.r.o, Hattalova 12, 831 03 Bratislava, Slovakia E-mail: [email protected], [email protected], [email protected], [email protected]

Main idea and ambition of the VESNA project was to create a powerful tool, determined to simplify decision-making in crisis flood situations for responsible personnel, on a local town level – town Levice in central Slovakia (35 thousand inhabitants). Levice town is endangered by small river Podlužianka, typical with fast flood response after heavy rainfall, flowing from hilly catchment to a flat populated area. The River Podlužianka in Levice has mean discharge 0,5 m3.s-1, flood discharge with 100 year return period is 65 m3.s-1. The main scope of the project was creating a tool for decision-making support, which will use standard meteorological and hydrological forecasts and operational data of SHMI (Slovak Hydrometeorological Institute), published for public use on Internet. On the basis of predicted (or even actual) meteorological and hydrological situation, the user chooses one category, which matches the current prediction or situation best. Several categories are defined in the system, covering a whole range of possible meteorological and consequently also hydrological situations in the area. Each meteorological and hydrological category is represented by one representative flood scenario, which was simulated in detail. Representative scenarios also include a possibility of breaking the flood protection lines. The system is oriented on single event heavy rainfall and consequent floods with typical delay of flood 8 – 12 hours after rainfall. Simulations of flood scenarios are not calculated in real time during actual flood, but were calculated and mapped in advance, for the particular defined model area during project. To its user the system offers a description of flood situation, its consequences, extend and time advancement of flood in the text part. In the graphical part of the tool, the user has a chance to view maps of water surface elevation, water depth, flow velocity and direction, as well as dangerous zones in a detailed scale. The results are complied with animations and 3D visualisation of the flooded area together with a photo documentation of past floods. Decision-making part of the system includes also other subsequent steps, which were defined and formulated during the project in cooperation with its future users (town mayor, flood commissions, river basin authority, rescue...). It is expected that after a testing and practical use of the tool, it will be further supplemented by other useful recommendations for action in urgent situations (announcing of flood level activity, evacuation of people, rescue works, necessary material and technique...). Information on result uncertainty is also at one’s disposal. The tool contains also contacts to involved organs and a complete flood plan of the town. It is possible to launch the VESNA system from flash memory stick (or HDD) on any PC equipped with standard Internet browser. System does not require any additional installation or licensed specialised software. Controlling of the system by the user is intuitive and requires only a minimal training. Created system is open, which means that it will be possible to input any necessary information according to current needs, to modify it, or add any other steps required for decision-making.

93

SESSION: PREDICTIONS

UNCERTAINTY IN 1D HYDRAULIC MODELLING CAUSED BY PARAMETERIZATION AND INFLOW INACCURACY

Mukolwe, M.M., Di Baldassarre, G., Werner, M. and Solomatine, D.

UNESCO-IHE, Institute for Water Education, PO Box 3015, 2601 DA Delft, the Netherlands Email: [email protected]

It is widely acknowledged that the use of accurate techniques to predict the effects of water related disasters, such as floods, forms a basis for the mitigation of the effects of such occurrences. Consequently, emphasis is placed on the better understanding of the methods and tools, including their associated uncertainty, that are applied in flood risk management. This paper presents the analysis of uncertainty in hydraulic modelling, with focus on the inaccuracy caused by inflow errors and parameter uncertainty. In particular, the study develops a method to propagate the uncertainty induced by: (1) the application of a stage- discharge rating curve (i.e. parameterisation of the power function), and (2) the calibration of the channel roughness coefficients (i.e. Manning's roughness coefficient) through a 1D hydraulic model (i.e. HEC-RAS). The methodology is applied and tested for a 98Km reach of the River Po (Italy), between Cremona and Borgoforte. In particular, the hydraulic model was evaluated by using two independent sets of hydrometric data (along the river reach), i.e. observed water levels, within a GLUE (Generalised Likelihood Uncertainty Estimation) framework. Independent calibration and validation, as well as the uncertainty analysis, showed a good performance of the model. In fact, when conditioned on a flood event, the model provided satisfactory results on predicting flood levels for another (independent) event as the observations were found within 95% confidence interval of the ensemble simulation. However, it should be noted that poor results were obtained when the model was conditioned on a data type different from the one used to evaluate model predictions.

94

SESSION: PREDICTIONS

UNCERTAINTY ASSESSMENT OF STREAMFLOW SIMULATION ON UNGAUGED CATCHMENTS USING A DISTRIBUTED MODEL AND THE KALMAN FILTER BASED MISP ALGORITHM

Múnera, J.C.1, Francés, F.1 and Todini, E.2

1 Instituto de Ingeniería del Agua y el Medio Ambiente, Universidad Politécnica de Valencia. Camino de Vera, s/n C.P. 46022, Valencia - España. Email: 2 DIPARTIMENTO di SCIENZE della TERRA e GEOLOGICO-AMBIENTALI, Università di Bologna. Piazza di Porta San Donato, 1 – 40126, Bologna - Italia. Email: [email protected], [email protected], [email protected]

In this work, an approach based on the MISP technique (Todini, 1978) is presented, which make use of two interacting Kalman filters coupled in parallel seeking to improve the historical hourly stream flow predictions made with a hydrologic distributed model in some ungauged catchments. On the other hand, the Kalman filter approach allows assessing the uncertainty associated to these predictions. The two coupled filters allow representing the dynamic behavior of a discrete lineal system; the first filter is responsible for the minimum variance estimation of the system state given a set of parameters, and the second one deals with updating the parameters of the state transition matrix of the first filter, also with minimum variance. Each moment is taken into account, the interaction between the two filters to get the best estimate of the system state and parameters, which essentially is equivalent to solve a non-linear problem (Todini, 1978). In the configuration of the state vector measurable quantities must be related at discrete time intervals. The state variables included in the state vector are the observed and simulated stream flows in a gauging station at the basin outlet, which is used as a pivot point to transfer information to an ungauged sub-catchment within it, as well as the stream flows simulated with the model in the latter. Additionally, we have also included these variables at previous time step, to account for autocorrelation in the process. Prior to execution of the filter, we made a logarithmic transformation of data, seeking a better approximation of the hypothesis of white noise, both in system and measurement errors. Finally, we have incorporated in the covariance matrix of the measurement errors a cross covariance term, to take into account the spatial correlation of errors between both stations. This approximation of the MISP approach was applied to some gauging stations located on the Illinois River basin, that were included in the second phase of the Distributed Model Intercomparison Project (DMIP 2) of NOAA/NWS. The hydrological model used in this case study is TETIS model (Vélez et al, 2002; Francés et al, 2009). In order to test the effectiveness of the algorithm, every ungauged point of simulation has been matched with an existing gauging station, in which observed data are used only for comparison purposes. The obtained results suggest that the proposed methodology is a robust and very useful tool for the uncertainty estimation related to streamflow predictions made with a hydrological model on ungauged catchments, and in some cases allow to improve such estimates.

95

SESSION: PREDICTIONS

TOWARDS SPATIALLY DISTRIBUTED FLOOD FORECASTS IN FLASH FLOOD PRONE AREAS: RESULTS OF A FIRST TEST IN THE SOUTH OF FRANCE

Naulin, J-P. 1, Payrastre, O.1, Gaume, E.1, Delrieu, G.2, Arnaud, P.3, Lutoff, C.4 and Vincendon, B.5

1 IFSTTAR, Bouguenais, France, 2 CNRS - LTHE, Grenoble, France 3 Cemagref, Aix en Provence, France 4 CNRS – UMR Pactes, Grenoble, France 5 GAME/CNRM (Meteo France, CNRS), Toulouse France E-mail: [email protected]

Accurate flood forecasts are crucial for an efficient flood event management. Until now, hydro-meteorological forecasts have been mainly used for early-warnings in France (Meteorological and flood vigilance maps) or over the world (Flash-flood guidances). Forecasts are generally limited to the main streams covered by the flood forecasting services or to specific watersheds with particular assets like hydropower dams which are in most cases well gauge river sections, leaving aside large parts of the territory Distributed hydro-meteorological forecasting models, able to take advantage of the now available high spatial and temporal resolution rainfall measurements, are promising tools for anticipating and quantifying the consequences at the ground of storm events all over a region. They would be very useful, especially in regions frequently affected by severe storms with complex spatio-temporal patterns. They would provide the necessary information for flood event management services to identify the areas at risk and to take the appropriate safety and rescue measures: prepositioning of rescue means, stopping traffic on exposed roads, determination of safe accesses or evacuation routes. Some preliminary tests conducted by the LCPC within the European project FLOODsite have shown encouraging results of a distributed hydro-meteorological forecasting model. It seems possible, despite the limits of the available rainfall measurements and the shortcomings of the rainfall-runoff models, to deliver distributed forecasts of possible local flood consequences – road submersion risk rating at 2500 different locations of the Gard department in the tested case - with an acceptable level of accuracy. The PreDiFlood project (http://heberge.lcpc.fr/prediflood/) aims at consolidating and extending these first results with the objective to conduct pre-operational tests with possible end-users at the end of the project. Such a tool will not replace, but complement existing flood forecasting approaches in time and space domains, short term forecasting at a regional scale, that have not been covered until now. It will produce a completely new type of forecasts and the usefulness of such data for the emergency services for their real-time decision making will be assessed within the project. Beyond the direct operational objectives, this project aims at demonstrating, on a specific application – the now-casting of road submersions, the possibilities and also the limits and hence the needed improvements of tools that are still underused: high spatial and temporal resolution rainfall measurements including radar quantitative precipitation estimates but also precipitation now-castings, rainfall-runoff models and the recent knowledge acquired on flash-floods consequence evaluation as well as event management.

96

SESSION: PREDICTIONS

PROBABILISTIC FLOOD RISK MAPPING INCLUDING SPATIAL DEPENDENCE

Neal, J.1, Keef, C.2, Bates, P.1, Beven, K.3 and Leedal, D.3

1 School of Geographical Sciences, University of Bristol, Bristol, BS8 1SS, UK. 2 JBA Consulting, South Barn, Broughton Hall, Skipton, N Yorkshire, BD23 3AE, UK. 3 Lancaster Environment Centre, Lancaster University, Lancaster, LA1 4YQ, UK. E-mail: [email protected]

Effective flood risk management depends on methods for estimating flood hazard and an appraisal of the dominant uncertainties in the analysis. Typically hydraulic models are used to simulate the extent of flooding for an estimate of the flow in a particular reach for a chosen return probability. However, this definition causes problems at river confluences where flows derive from multiple sources. Here a model-based approach was adopted to describe the multisite joint distribution of river flows for three rivers that converge on the city of Carlisle (UK). Monte-Carlo methods were used to generate a best estimate of the set of flood events with realistic spatial dependence between tributaries which would occur over a 100 year period. Each event was then simulated with a new computationally efficient two-dimensional hydraulic model of the whole city. Simulations from the hydraulic model were used to map inundation frequency, with the largest event in the event set assumed to represent the 1 in 100 year flood. To estimate uncertainty in the statistical model used to create the event set block bootstrapping was used to produce a further 100 runs of the event generator. Each of the ~47000 dynamic flood events created by this process was then simulated using the hydraulic model. Flood risk was calculated for each of the 100 event generator runs based on a depth damage curve. The 90th percentile of the resulting damage distribution was 55 times greater than the 10th percentile, indicating significant uncertainty in the flood damage and the limitations of typical discharge records.

97

SESSION: PREDICTIONS

ENSEMBLE PREDICTION OF FLOODS – PROCESS CONTROLS

Nester, T. and Blöschl, G.

Vienna University of Technology, Austria, Institute for Hydraulic and Water Resources Engineering E-mail: [email protected]

The runoff forecasting model for the Danube tributaries in Upper Austria and Lower Austria has been in operational use since 2006. The model is based on a conceptual semi- distributed water balance model with catchment sizes ranging from 25 to 25.000 km². The aim of this study is to examine how the spread in ensemble runoff forecasts and the differences between observed runoff and deterministic runoff forecasts are related. We are also interested in the contributions of the model and precipitation errors to the total forecast errors. Assuming that all ensemble members are equally likely, the statistical analyses indicate that the ensemble spread of the flood forecasts is always narrower than the distribution of the forecast errors. For shorter lead times, the contribution of the model error is larger than the contribution of the precipitation error while for longer lead times the contribution of the precipitation is the dominant factor.

98

SESSION: PREDICTIONS

FLOOD HAZARD UNDER CLIMATE CHANGE IN THE MEKONG DELTA

Nguyen, V.D.1,3, Delgado, J.M.1, Merz, B.1, Bárdossy, A.2 and Apel, H.1

1 GFZ German Research Centre for Geosciences, Section 5.4 Hydrology, Potsdam, Germany 2 Institute of Hydraulic Engineering, University of Stuttgart, Stuttgart 70569, Germany 3 Southern Institute of Water Resources Research SIWRR, Ho Chi Minh City, Vietnam E-mail: [email protected]

Flood hazard analysis is an indispensable input for flood risk assessment. An essential part is the determination of probabilities of occurrence of floods of different magnitudes. The methods applied for the determination are mostly based on the stationary assumption. However, this assumption does not hold for most of the observed discharge time series in general, and in particular not for future climate conditions. It is now commonly agreed that climate change leads to changing flood hazards. Therefore, this study aims at developing a novel approach for flood hazard mapping considering the change in climate variability with the Mekong Delta as an example. We explicitly take non-stationarity and dependence in the discharge time series into consideration, for both annual maximum flood peak discharge and flood volume. The dependency in the marginal distribution of flood peak discharge and volume is modeled by copulas, while characteristic flood hydrograph shapes are identified through a cluster analysis. We further analyzed the dependency of the statistical distributions of flood peak and volume on the variance in monsoon indexes, which yielded a reliable proxy for flood intensity in the Mekong Delta, avoiding time consuming and error prone downscaling of GCM results and hydrological modeling. Future development of the flood intensity and frequency was consequently estimated by the identified non-stationarity in the extreme value statistics, the dependency of the frequency distributions to monsoon intensity, and the monsoon variability simulated by 7 different GCM’s. The changes in the frequency of flood peak and duration were mapped into probabilistic inundation maps by driving a large scale hydrodynamic model for the Mekong Delta in a Monte Carlo framework. A comparison of these maps for the projection year 2050 with the reference year 2000 illustrates the changes in flood hazard spatially explicit for the whole delta.

99

SESSION: PREDICTIONS

FLOOD PRONE AREAS MAPPING THROUGH GIS AND MULTI-CRITERIA ANALYSIS

Papaioannou, G.1, Loukas, A.1, Vasiliades, L.1 and Aronica, G.T.2

1Department of Civil Engineering, University of Thessaly, Pedion Areos, 38334 Volos, Greece, 2Department of Civil Engineering, University of Messina , Via Nuova Panoramica dello Stretto, 98166 Villaggio S. Agata - Messina, Italy, E-mail: [email protected], [email protected]

Flood risk mapping is a basic component of the European natural disaster management policy, directly connected to water resources management with the European Directive (2007/60). This study presents a methodology for flood prone areas mapping using geographical information systems (GIS) and multi-criteria evaluation techniques. Factors are divided in different groups which do not have the same level of trade off. These groups are related to geophysical, morphological, climatological/meteorological and hydrological characteristics of the basin as well as to human activities. GIS are used for geographic data acquisition and processing. The analytical hierarchy process (AHP) is the multicriteria evaluation technique used, enhanced with different standardization methods. The method has been applied to the watershed and the floodplain of Xerias River in the region of Thessaly, Greece. In this work, a methodology which combines a risk based analysis and geographical information systems is proposed for the mapping of flood prone areas. Xerias river was repeatedly flooded in the last few years. These floods had major impacts on agricultural areas, road networks and other technical infrastructures. Historical flood inundation data in the study area have been used for the validation of the methodology.

100

SESSION: PREDICTIONS

VISUALISING PROBABILISTIC FLOOD FORECAST INFORMATION: EXPERT PREFERENCES AND PERCEPTIONS OF BEST PRACTICE IN UNCERTAINTY COMMUNICATION

Pappenberger, F.1, Stephens, E.2, Thielen, J.5, Salamon, P.5, Demeritt, D.3, van Andel, S.J.4, Wetterhall, F.1 and Alfieri, L.5

1 European Centre For Medium Range Weather Forecasts, Reading, UK 2 University of Bristol, Bristol, UK 3 King’s College London, London, UK 4 Unesco IHE, Delft, The Netherlands 5 Joint Research Centre of the European Commission, JRC, Ispra E-mail: [email protected]

The aim of this paper is to understand and to contribute to improved communication of the probabilistic flood forecasts generated by Hydrological Ensemble Prediction Systems (HEPS) with particular focus on the inter expert communication. Different users are likely to require different kinds of information from HEPS and thus different visualizations. The perceptions of this expert group are important both because they are the designers and primary users of existing HEPS. Nevertheless, they have sometimes resisted the release of uncertainty information to the general public because of doubts about whether it can be successfully communicated in ways that would be readily understood to non-experts. In this paper we explore the strengths and weaknesses of existing HEPS visualization methods and thereby formulate some wider recommendations about best practice for HEPS visualization and communication. We suggest that specific training on probabilistic forecasting would foster use of probabilistic forecasts with a wider range of applications. The result of a case study exercise showed that there is no overarching agreement between experts on how to display probabilistic forecasts and what they consider essential information that should accompany plots and diagrams. In this paper we propose a list of minimum properties that, if consistently displayed with probabilistic forecasts, would make the products more easily understandable.

101

SESSION: PREDICTIONS

DEVELOPMENT OF DECISION SUPPORT SYSTEMS FOR FLASH FLOOD FORECASTING IN SMALL CATCHMENTS

Poma, S.1, Brigandì, G.2, Aronica, G.T.2 and Candela, A.1

1 Dipartimento di Ingegneria Civile, Ambientale e Aerospaziale, University of Palermo, Italy 2 Dipartimento di Ingegneria Civile, University of Messina, Italy E-mail: [email protected]

In the recent past throughout the Mediterranean area, many extreme events such as floods, debris flows and landslides occurred. The use of a reliable flood forecasting model can play an important role in managing land and water resources. The purpose of this work is the development of a Decision Support System (DSS) for flash flood forecasting. A system of alerts, based on the use of rainfall thresholds and structured in three phases, has been implemented in order to achieve an integrated system of support for the prediction of hydraulic and hydro-geological instability. Starting from precipitation and temperature air data, the first phase of the DSS deals with the implementation of a procedure to model the soil moisture conditions (saturation degree) using API index in order to generate a first warning based on the antecedent moisture condition of the soil. In the second phase the comparison between rainfall events and rainfall thresholds calculated for a given river section of the catchment are displayed. Finally, the third phase concerns the imminence of a possible hydrogeological risk by comparing the time when cumulative rainfall and rainfall thresholds meet each other. The output of the first phase gives the probability of producing a flood event if the catchment has a certain degree of saturation. To activate the other two alarm phases it is necessary to derive, for a given river section where the critical discharge has been fixed, rainfall thresholds as a function of the antecedent soil moisture conditions and the temporal evolution of the rain storm. The implementation of the DSS allows to build a user interface for the operator of the flood risk easy to understand. The system has been successfully applied to the Giampilieri catchment, located in the north-east of Sicily. Actually, the system could help to program a precise system of civil protection in order to put in place all necessary emergency actions to reducing the loss of lives and damage to property.

102

SESSION: PREDICTIONS

AN ENTROPY APPROACH FOR THE OPTIMIZATION OF RIVER CROSS-SECTIONAL SPACING

Ridolfi, E.1, Alfonso, L.2, Di Baldassarre, G.2, Dottori, F.3, Russo, F.1 and Napolitano, F.1

1 Dipartimento di Ingegneria Civile, Edile e Ambientale, Sapienza Università di Roma, Rome, Italy. 2 Department of Hydroinformatics and Knowledge Management, UNESCO-IHE Institute for Water Education, Delft, the Netherlands. 3 Università di Bologna, Dipartimento di Scienze Della Terra e Geologico-Ambientali, Via Zamboni 67, 40126 Bologna, Italy. E-mail: [email protected], [email protected], g.dibaldassarre@unesco- ihe.org, [email protected], [email protected], [email protected]

Accurate definition of rivers geometry is a relevant task in the preparation of one-dimensional hydraulic models. In particular, the selection of river cross section spacing is the key for accurately describing the river hydraulics. In this work an entropy approach that allows evaluating the optimum number of river cross sections and their location on the river geometry is presented. Given a redundant collection of pre-established cross section locations, the aim is to select a subset of locations such that together provide the least redundant information and, simultaneously, the highest information content. The problem can be posed as a MultiObjective Optimization Problem (MOOP), in which two objectives are compromised: minimising total correlation (as a measure of redundancy) and maximising joint entropy (as a measure of information content). The MOOP is solved using the non-sorted genetic algorithm NSGA-II and a Pareto front of optimal sets of cross section locations is obtained. The Pareto front is then analysed and the most appropriate solution is chosen. The proposed method is applied to an existing hydraulic model for the Po river between Cremona and Borgoforte, built with an important number of redundant cross sections. In order to accurately describe the profile of the watermarks of a flood event, the MOOP is applied to find the optimal number and locations of cross sections. It is demonstrated that the proposed approach matches very well with a previous methodology already presented in literature.

103

SESSION: PREDICTIONS

FLOOD-RISK MAPPING – A NEW INSIGHT INTO A MAJOR ENVIRONMENTAL PROBLEM

Romanowicz, R., Kiczko, A. and Osuch, M.

Institute of Geophysics Polish Academy of Science, Księcia Janusza 64, 01-452 Warsaw, Poland E-mail: [email protected]

Derivation of flood risk maps requires estimation of maximum flood inundation extent with assumed probability of exceedance, e.g. 100 or 500 year flood. The results of numerical simulation of flood wave propagation are used to overcome the lack of relevant observations. In practice, deterministic 1-D models are used for flow routing, giving a simplified image of a flood wave propagation process. The solution of a 1-D model depends on the initial and boundary conditions and estimates of model parameters which are usually identified using the inverse problem based on the available noisy observations. Therefore, there is a large uncertainty involved in the derivation of flood risk maps. This work is an attempt to quantify this uncertainty. We compare the flood risk maps obtained using standard deterministic methods with the maps that take into account some aspects of the uncertainty related to a design flood input, observation errors and simplifications of the model structure. We apply the Bayesian conditioning to assess the uncertainties of observations used for the calibration of model parameters. The methods are compared using the Upper Narew catchment as a case study. The results indicate that the uncertainties have a substantial influence on the flood risk assessment. In this paper, we present a simplified methodology allowing the influence of that uncertainty to be assessed.

104

SESSION: PREDICTIONS

BAYESIAN UNCERTAINTY ASSESSMENT OF FLOOD PREDICTIONS IN UNGAUGED URBAN BASIS FOR CONCEPTUAL RAINFALL-RUNOFF MODELS

Sikorska, A.1,2, Scheidegger, A.1, Banasik, K.2 and Rieckermann, J.1

1 Eawag: Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland 2 Warsaw University of Life Sciences - SGGW, Faculty of Civil and Environmental Engineering, Dept. of Hydraulic Engineering, Warsaw, Poland E-mail: [email protected], [email protected], [email protected], [email protected], [email protected]

Urbanization and the resulting land-use change strongly affect a watershed’s water cycle. Especially with regard to precipitation, the slow natural response in discharge drastically changes into a rapid and fast response of the urbanized catchment due to the additional impervious area. Particularly in Eastern European cities, many urbanized areas experienced rapid growth in recent years, resulting in more frequent urban flooding events. However, often it is not trivial to predict the magnitude of a flood event, because catchment data and hydrological measurements are not generally available. This usually prohibits the use of detailed process-based rainfall-runoff models. On the one hand, it is difficult to inform the parameters, e.g. soil moisture or hydraulic conductivity. On the other hand, most small urban catchments are ungauged and therefore suffer from a lack of monitoring data. Because of that, to predict the consequences of future urbanization, we are bound to using simpler models. Our goal of the present study is two-fold: First, we investigate how far we can use conceptual hydrological models to characterize runoff from ungauged urbanized catchments. Second, we wanted to provide urban planners with a reliable measure of uncertainty of the model predictions, which explicitly considers uncertainty from the simple model structure as well as input uncertainty, e.g. from inaccurate rainfall measurements or outdated GIS catchment data. To this aim, we applied a methodology to assess the reliability of predicted discharges which takes into account uncertainty in the model input, the model parameters, measurement errors and uncertainty stemming from the simple model structure. A key feature of the method is that it avoids common unrealistic assumptions on independently and identically (normally) distributed residuals. Instead, the likelihood function is constructed by combining a Box–Cox data transformation with a continuous-time autoregressive error model, which is then used for Bayesian parameter estimation. To the best of our knowledge, this has not been previously applied to assess conceptual models of ungauged urbanized catchments. We demonstrate the potential of our approach on a case study from Sluzew creek, which drains an urbanized catchment within Warsaw, PL. Specifically, we identified three potentially suitable conceptual models from literature and investigated their predicted performance based on several years of rainfall and streamflow measurements. Our first results show a promising agreement between modelled and observed hydrographs as well as a satisfactory prediction performance. However, prediction uncertainty is dominated by the error due to the simple model structure and less by uncertainty in model parameters. Therefore, the “manual” derivation of the parameters does not introduce significantly more uncertainty. So far, we can conclude that conceptual models can deliver reasonable predictions of the response from small ungauged urban catchments and – in combination with Bayesian techniques – a sound description of the uncertainties is possible.

105

SESSION: PREDICTIONS

DEVELOPMENT OF EXPECTED ANNUAL DAMAGE CURVES AND MAPS AS A BASIC TOOL FOR THE RISK-BASED DESIGNING OF STRUCTURAL AND NON- STRUCTURAL MEASURES

Tariq, M.A.U.R., Van de Giesen, N.C. and Hoes, O.

Faculty of Civil Engineering and Geosciences, Technical University Delft, Stevinweg 1 Delft, 2628CN, the Netherlands E-mail: [email protected]

At present, probabilistic design standards for the flood protection measures are being practiced in most parts of the word. These probability-based methods cannot be used to design a non-structural measures and are unable provide an optimal design of structural measures. Risk-based designs of flood management measures are getting more attention due to their ability to optimize both the structural and non-structural measures. A flood measure that is designed against a specific flood frequency may produce an under protection design or a cost-inefficient solutions at its both extremes. Flood damages are probabilistic events and flood intensity also greatly varies over the floodplain. Floods impacts are also distributed variantly over spatial and temporal frames. It is important to consider the probable floods and their possible damages while designing a measure. In early 1990’s, US Army Corps of Engineers used the relationships of damage, frequency, discharge, and stage for a floodplain to estimate expected annual damages (EAD) using a hydroeconomic model. The problem with hydroeconomic model is that it estimates losses using pre-established relationship between flood frequency and damages. Whereas, flow regime may be changed significantly once a structural measure is introduced. Thence, this method cannot provide accurate estimates of expected losses in such cases. Calculation of losses on the bases of GIS data and hydraulic modeling is proposed in our research. The proposed methodology recommends the calculation of EAD based on (modified) actual flow regime. These indicators can also be used for any development project or non-structural measure in a floodplain. Each of these indicators has its specific use. Categorically speaking, EAD value can be used as a deciding tool while choosing between the flood management options or strategies. Any flood management plan will correspond to a single EAD value; which can be used to conclude a decision. It is obvious that the strategy having lowest EAD is more acceptable in general. EAD curves and EAD maps provide a detailed illustration of risk distributions. EAD Maps are also known as risk maps. At present, these maps are mostly used to develop flood zoning. These maps combine the hazard and vulnerability. Flood risk maps can be developed to estimate and alleviate the potential flood damages. EAD curve depicts the trend of damages against probable floods. Whereas, EAD map helps in determining flood impacts over both time and space frames. EAD map is also helpful for land-use planners. EAD curves, maps, and values were calculated in our case study areas of the Chenab and Swan rivers in Pakistan. These basic indicators were used in further development of structural and non-structural measures as a part of our project. On the basis of their design philosophy, their capabilities and convenient usability, the use of EAD is highly recommended for the design almost all kind of flood management measures.

106

SESSION: PREDICTIONS

FLOOD FORECASTING WITH CONSTRAINED DOWNSTREAM CONDITION IN YUVACIK BASIN, TURKEY

Uysal, G.1, Yavuz, Ö.1, Şensoy, A.1, Şorman, A.A.1, Akgün, T.2 and Gezgin, T.2

1 Anadolu University, Iki Eylul Campus, Department of Civil Eng., 26555, Eskisehir, Turkey, 2 Akifer Su Hizmetleri Ltd. Sti., Izmit Icme Suyu Aritma Tesisleri, Basiskele, Kocaeli, Turkey E-mail: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]

Yuvacık Dam Basin, located in the Marmara region of Turkey with a 248 km2 drainage area, has high flood potential due to its steep topography, mild and rainy climate. The basin shows fast flow response, especially to early spring and fall precipitation events. Moreover, the basin’s elevation ranges between 80 – 1548 m, allowing considerable snowmelt contribution to the streamflow during melt season. The dam reservoir, having a total volume of 56 million m3, is designed to provide 142 million m3 of domestic and industrial water annually for the city of Kocaeli with a population of 1.5 million people, despite the relatively small capacity of its reservoir. The 12 km length downstream reach of the stream passes initially from a narrow valley near a rural district and thereafter flows into the Marmara Sea after sharp curvature by a manmade channel next to industrial and urban areas. Hence, the maximum amount of water to be released is set as 100 m3/s by the regional water authority taking the drainage discharge condition into consideration although the spillway capacity is 1560 m3/s. The reservoir spillway is operated by radial gates behind which excess water is stored especially during the flood season (late February to June) and operations require release regulation through downstream. A hydrological model with an embedded snow module is used to establish relationship between the rainfall and runoff to calculate the inflow into the dam reservoir. The basin is divided into four sub-basins, along with the three elevation zones for each basin. Hydro- meteorological data are collected via 11 automated stations in and around the basin and a semi-distributed rainfall runoff model, the HEC-HMS, is calibrated for sub-basins that drain into the reservoir lake. Daily total precipitation and daily average temperature forecast data that constitute the hydrological model input are provided from the Mesoscale Model 5 (MM5) numerical weather prediction data. Predicted precipitation and temperature data are compared with ground observations to examine consistency. Thereafter, the representative MM5 data pixels are assigned to rain gages. A kind of lapse rate is applied to remove temperature difference due to the topography. The study is carried out to test reservoir operation discharges especially during flood events in conditions of using different reservoir level scenarios. Predicted inflows are used as main inputs in HEC-ResSim for the hourly and daily reservoir simulation of flood events. In this case study, Yuvacık Dam reservoir operation for the 6 – 11 April 2010 flood event, during which spillway has been operated, is selected and simulated with different reservoir level scenarios. The operations and release decisions are analyzed to propose an improvement in its decision support system. Model parameters calibrated for several flood events are used in real time forecasting of reservoir inflows and several scenarios are tested to improve management of the reservoir.

107

SESSION: PREDICTIONS

UNCERTAINTY ANALYSIS OF HEAVY RAINFALL EVENTS REPRESENTED BY A HYDROMETEOROLOGICAL MODEL AND SATELLITE RAINFALL ALGORITHM FOR FLOOD FORECASTING

Yucel, I.1 and Mert, I.2

1Middle East Technical University, Civil Engineering Department, Water, Resources Lab, 06800 Ankara Turkey 2Turkish State Meteorological Service, Kalaba 06120 Ankara Turkey E-mail: [email protected], [email protected]

Evidence is showing that global warming or climate change has a direct influence on changes in precipitation and the hydrological cycle. Extreme weather events such as heavy rainfall and flooding are projected to become much more frequent as climate warms. On the other hand, hydrologic predictions are critical for decision-making related to flood mitigation and water resources. In data sparse regions, alternative products such as atmospheric model forecasts and remotely sensed data should be used for flood forecasting and warning issues as they provide more continuous monitoring of precipitation data both spatially and temporally over large extent of coverage. Hydrologic model simulations depend heavily on the availability of reliable precipitation estimates. Difficulties in estimating precipitation in data sparse regions impose an important limitation on the possibility and reliability of hydrologic forecasting and early warning systems. Therefore, as supplement to rain gauge and radar rainfall data the use of remotely sensed and numerical model-based precipitation products becomes critical in hydrological simulations. This study investigates the performance of the rainfall products obtained from the Multi Precipitation Estimates (MPE) and Weather Research and Forecasting (WRF) model and provide an uncertainty analysis among them by comparing with gauge rainfall measurements. The precipitation estimates obtained by using the three-dimensional variational (3D-VAR) data assimilation scheme within WRF system are also used in uncertainty analysis. Uncertainty analysis is performed for the selected number of heavy rainfall events and associated flooding conditions over the mountainous West Black Sea Basin in Turkey. The MPE and WRF-estimated rainfall showed capabilities in capturing the timing of the flood events and in some extent spatial distribution and magnitude of the heavy rainfall events. WRF with assimilation improved area-averaged precipitation forecasts by 9 percent and at some points there exists quantitative match for heavy rainfall events, which are critical for hydrological forecast.

108

SESSION: PREDICTIONS

UNCERTAINTY ASSESSMENT OF CLIMATE CHANGE ADAPTATION OPTIONS IN URBAN FLASH FLOODS

Zhou, Q. and Arnbjerg-Nielsen, K.

Technical University of Denmark E-mail: [email protected]

Adaptation is necessary to cope with the increasing flood risk in cities due to anthropogenic climate change in many regions of the world. The choice of adaptation strategies can and should be based on a comprehensive risk-based economic analysis to indicate the net benefits of proposed options. However, the analysis is complicated by irreducible uncertainties about present and future hydrologic conditions as well as the present and future vulnerability of the area in question. Further, modeling of the actual hazards given the hydrologic conditions also entails substantial uncertainty. The work presented is based on a flood risk framework that is in accordance with the EU flood directive, but adapted and extended to incorporate anticipated future changes due to city development and hydrologic extremes. The framework is used to study the importance of inherent uncertainties in order to find robust adaptation options. The case study is a small urban catchment where no significant city development is anticipated. Therefore the main focus is on estimation of impact of uncertainties related to present and future hydrological conditions, impacts on assets, and costing of the damages. The uncertainties are calculated using Monte Carlo simulations and thus the resulting uncertainties are described by probability density functions. Two different adaptation options are studied to reduce the increase in risk of flooding, namely increasing the pipe capacity and the use of local infiltration measures to hold water back from flood prone areas. The two options represent classical engineering solutions and water sensitive urban design, respectively. These options are compared to a business-as-usual scenario, where no adaptation is foreseen in the area. The results indicate that infiltration is less cost-effective regardless of the uncertainties from climate change impacts and /or damage estimation procedure when considering the ability to reduce the risk of flooding. The description of the correlation structure between the key inputs proved to be important in order to obtain a correct description of the resulting uncertainties. The study shows that uncertainties associated with climate adaptation benefits are large, but still it is possible to choose no-regret adaptation options. The introduced procedure provides an important tool for achieving an explicit and thorough uncertainty analysis of the risk-based economic evaluation.

109

SESSION: PREDICTIONS

110