Dissertations / Theses on the topic 'Flood and Flash Flood forecasting'

A major area targeted for hydrometeorological forecast service improvements is in flash flood forecasting. Verification data show that general public service products of flash flood forecasts do not provide enough lead time in order for the public to make effective response. Sophisticated users of flash flood forecasts could use forecast probabilities of flash flooding in order to make decisions in preparation for the predicted event. To this end, a systematic probabilistic approach to flash flood forecasting is presented. The work first describes a deterministic system which serves as a conceptual basis for the probability system. The approach uses accumulated rainfall plus potential rainfall over a specified area and time period, and assesses this amount against the water holding capacity of the affected basin. These parameters are modeled as random variables in the probabilistic approach. The effects of uncertain measurements of rainfall and forecasts of precipitation from multiple information sources within a time period and moving forward in time are resolved through the use of Bayes' Theorem. The effect of uncertain inflows and outflows of atmospheric moisture on the states of the system, the transformation of variables, is resolved by use of convolution. Requirements for probability distributions to satisfy Bayes' Theorem are discussed in terms of the types and physical basis of meteorological data needed. The feasibility of obtaining the data is evaluated. Two alternatives for calculating the soil moisture deficit are presented--one, an online automatic rainfall/runoff model, the other an approximation. Using the soil moisture approximation, a software program was developed to test the probabilistic approach. A storm event was simulated and compared against an actual flash flood event. Results of the simulation improved forecast lead time by 3-5 hours over the actual forecasts issued at the time of the event.

An empirical method is developed for constructing likelihood functions required in a Bayesian probabilistic flash flood forecasting model using data on objective quantitative precipitation forecasts and their verification. Likelihoods based on categorical and probabilistic forecast information for several forecast periods, seasons, and locations are shown and compared. Data record length, forecast information type and magnitude, grid area, and discretized interval size are shown to affect probabilistic differentiation of amounts of potential rainfall. Use of these likelihoods in Bayes' Theorem to update prior probability distributions of potential rainfall, based on preliminary data, to posterior probability distributions, reflecting the latest forecast information, demonstrates that an abbreviated version of the flash flood forecasting methodology is currently practicable. For this application, likelihoods based on the categorical forecast are indicated. Apart from flash flood forecasting, it is shown that likelihoods can provide detailed insight into the value of information contained in particular forecast products.

Flash flooding in the semi-arid United States poses a significant danger to life and property. One effective way to mitigate flood risk is by implementing a rainfall-runoff model in a real-time forecast and warning system. This study investigated the feasibility of using the mechanistic, distributed semi-arid rainfall-runoff model KINEROS2 driven by high resolution radar rainfall input estimates obtained from the NEXRAD WSR-88D DHR reflectivity measurements in such a system. The original procedural paradigm-based KINEROS2 Fortran 77 code with space-time looping was recoded into an object-oriented Fortran 90 code with time-space looping for this purpose. The recoded form is now applicable to large basins, is easily future-extensible, and individual modules can be incorporated into other models.Sources of operational uncertainty in the above system were investigated for their influence over several events within a sub-basin of the USDA-ARS Walnut Gulch Experimental Watershed. Uncertainties considered were in the rainfall estimates, the model parameters, and the initial conditions. The variance-based Sobol' method of global sensitivity analysis conditioned on the observed streamflow showed that the uncertainty in the modeled response was heavily dominated by the operational variability of biases in the radar rainfall depth estimates. Sensitivities to KINEROS2 parameters indicates the need for improved representation of semi-arid hillslope hydrology in small basins, while pointing to specific influential, but poorly identified model parameters towards which field investigations should be directed. The significant influence of initial hillslope soil moisture showed the requirement of a sophisticated inter-storm model component for a continuous forecasting model.A synthetic study data was used to further explore the phenomena seen in the above real data study, of behavioral modifier set inconsistency across all events and of irreducibility in the spatial modifier ranges. The former was found to be attributable to wide uncertainty ranges in the sources of uncertainty, and the latter to the high distributed model non-linearity with associated interactions. These contribute towards a high predictive uncertainty in operational forecasting.Overall, the GLUE-based predictive uncertainty method with behavioral classification and accommodation of wide operational source uncertainty ranges is recommended as a simple and effective setup for operational flash flood forecasting.

Au XXIe siècle, la prévision de l'aléa hydrométéorologique et des impacts associés aux crues rapides demeurent un défi pour les prévisionnistes et les services de secours. Les mesures structurelles et / ou les avancées des systèmes de prévision hydrologique ne garantissent pas, à elles seules, la réduction des décès lors de ces phénomènes d'inondation rapide. La littérature souligne la nécessité d'intégrer d'autres facteurs, liés aux processus de vulnérabilité sociaux et comportementaux, afin de mieux prendre en compte les risques encourus par les populations lors de ces épisodes extrêmes. Cette dissertation conduit une analyse théorique couplés à ceux de une analyse des accidents historiques mortels afin d'expliquer les interactions qui existent entre les processus hydrométéorologiques et sociaux responsables de l'apparition de vulnérabilités humaines lors de crues rapides aux États-Unis. Des données d'enquêtes liées aux crues rapides sont examinées afin d'élaborer un système de classification des circonstances du décès (en voiture, à l'extérieur, à proximité d'un cours d'eau, dans un camping, dans un bâtiment ou en mobile-home). L'objectif est d'établir un lien entre la conception des vulnérabilités et l'estimation des pertes humaines liées à ces catastrophes naturelles. "Random forest" est utilisé et est basé sur un arbre de décision, qui permet d'évaluer la probabilité d'occurrence de décès pour une circonstance donnée en fonction d'indicateurs spatio-temporels. Un système de prévision des décès liés à l'usage de la voiture lors des crues rapides, circonstance la plus répandue, est donc proposé en s'appuyant sur les indicateurs initialement identifiés lors de l'étude théorique. Les résultats confirment que la vulnérabilité humaine et le risque associé varient de façon dynamique et infra journalière, et en fonction de la résonance spatio-temporelle entre la dynamique sociale et la dynamique d'exposition aux dangers. Par exemple, on constate que les jeunes et les personnes d'âge moyen sont plus susceptibles de se retrouver pris au piège des crues rapides particulièrement soudaines(par exemple, une durée de près de 5 heures) pendant les horaires de travail ou de loisirs en extérieur. Les personnes âgées sont quant à elles plus susceptibles de périr à l'intérieur des bâtiments, lors d'inondations plus longues, et surtout pendant la nuit lorsque les opérations de sauvetage et / ou d'évacuation sont rendues difficiles. Ces résultats mettent en évidence l'importance d'examiner la situation d'exposition aux risques en tenant compte de la vulnérabilité dynamique, plutôt que de se concentrer sur les conceptualisations génériques et statiques. Ce concept de vulnérabilité dynamique est l'objectif de modélisation développée dans cette thèse pour des vulnérabilités liés aux véhicules. À partir de l'étude de cas sur les crues rapides survenues en mai 2015, et en analysant principalement les états du Texas et de l'Oklahoma, principaux états infectés par ces évènements,le modèle montre des résultats prometteurs en termes d'identification spatio-temporelle des circonstances dangereuses. Cependant, des seuils critiques pour la prédiction des incidents liés aux véhicules doivent être étudiés plus en profondeur en intégrant des sensibilités locales non encore résolues par le modèle. Le modèle établi peut être appliqué, à une résolution journalière ou horaire, pour chaque comté du continent américain. Nous envisageons cette approche comme une première étape afin de fournir un système de prévision des crues rapides et des risques associés sur le continent américain. Il est important que la communauté scientifique spécialisée dans l'étude des crues éclairs récoltent des données à plus haute résolution lorsque ces épisodes entrainement des risques mortels, et ce afin d'appuyer la modélisation des complexités temporelles et spatiales associées aux pertes humaines causées par les futures inondations soudaines
In the 21st century the prediction of and subsequent response to impacts due to sudden onset and localized flash flooding events remain a challenge for forecasters and emergency managers. Structural measures and/or advances in hydrological forecasting systems alone do not guarantee reduction of fatalities during short-fuse flood events. The literature highlights the need for the integration of additional factors related to social and behavioral vulnerability processes to better capture risk of people during flash floods. This dissertation conducts a theoretical analysis as well as an analysis of flash flood-specific historic fatalities to explain complex and dynamic interactions between hydrometeorological, spatial and social processes responsible for the occurrence of human life-threatening situations during the "event" phase of flash floods in the United States (U.S.). Individual-by-individual fatality records are examined in order to develop a classification system of circumstances (i.e., vehicle-related, outside/close to streams, campsite, permanent buildings, and mobile homes). The ultimate goal is to link human vulnerability conceptualizations with realistic forecasts of prominent human losses from flash flood hazards. Random forest, a well-known decision-tree based ensemble machine learning algorithm for classification is adopted to assess the likelihood of fatality occurrence for a given circumstance as a function of representative indicators at the county-level and daily or hourly time steps. Starting from the most prevalent circumstance of fatalities raised from both the literature review and the impact-based analysis, flash flood events with lethal vehicle-related accidents are the subject to predict. The findings confirm that human vulnerability and the subsequent risk to flash flooding, vary dynamically depending on the space-time resonance between that social and hazard dynamics. For example, it is found that younger and middle-aged people are more probable to get trapped from very fast flash floods (e.g., duration close to 5 hours) while participating in daytime outdoor activities (e.g., vehicle-related, recreational). In contrary, older people are more likely to perish from longer flooding inside buildings, and especially in twilight and darkness hours when rescue and/or evacuation operations are hindered. This reasoning places the importance of situational examination of dynamic vulnerability over generic and static conceptualizations, and guides the development of flash flood-specific modeling of vehicle-related human risk in this thesis. Based on the case study of May 2015 flash floods with a focus in Texas and Oklahoma, the model shows promising results in terms of identifying dangerous circumstances in space and time. Though, critical thresholds for the prediction of vehicle-related incidents need to be further investigated integrating local sensitivities, not yet captured by the model. The developed model can be applied on a daily or hourly basis for every U.S. county. We vision this approach as a first effort to provide a prediction system to support emergency preparedness and response to flash flood disasters over the conterminous U.S. It is recommended that the flash flood disaster science community and practitioners conduct data collection with more details for the life-threatening scene, and at finer resolutions to support modeling of local temporal and spatial complexities associated with human losses from flash flooding in the future

Flash flood is among the most hazardous natural disasters, and it can cause severe damages to the environment and human life. Flash floods are mainly caused by intense rainfall and due to their rapid onset (within six hours of rainfall), very limited opportunity can be left for effective response. Understanding the socio-economic characteristics involving natural hazards potential, vulnerability, and resilience is necessary to address the damages to economy and casualties from extreme natural hazards. The vulnerability to flash floods is dependent on both biophysical and socio-economic factors. This study provides a comprehensive assessment of socio-economic vulnerability to flash flood alongside a novel framework for flash flood early warning system. A socio-economic vulnerability index was developed for each state and county in the Contiguous United States (CONUS). For this purpose, extensive ensembles of social and economic variables from US Census and the Bureau of Economic Analysis were assessed. The coincidence of socio-economic vulnerability and flash flood events were investigated to diagnose the critical and non-critical regions. In addition, a data-analytic approach is developed to assess the interaction between flash flood characteristics and the hydroclimatic variables, which is then applied as the foundation of the flash flood warning system. A novel framework based on the D-vine copula quantile regression algorithm is developed to detect the most significant hydroclimatic variables that describe the flash flood magnitude and duration as response variables and estimate the conditional quantiles of the flash flood characteristics. This study can help mitigate flash flood risks and improve recovery planning, and it can be useful for reducing flash flood impacts on vulnerable regions and population.

Flash floods caused by localized thunderstorms are a natural hazard of the semi -arid Southwest, and many communities have responded by installing ALERT flood forecasting systems. This study explored a rainfall- runoff modeling approach thought to be appropriate for forecasting in such watersheds. The kinematic model KINEROS was evaluated because it is a distributed model developed specifically for desert regions, and can be applied to basins without historic data. This study examined the accuracy of KINEROS under data constraints that are typical of semi -arid ALERT watersheds. The model was validated at the 150 km2, semi -arid Walnut Gulch experimental watershed. Under the conditions examined, KINEROS provided poor simulations of runoff volume and peak flow, but good simulations of time to peak. For peak flows, the standard error of estimate was nearly 100% of the observed mean. Surprisingly, when model parameters were based only on measurable watershed properties, simulated peak flows were as accurate as when parameters were calibrated on some historic data. The accuracy of KINEROS was compared to that of the SCS model. When calibrated, a distributed SCS model with a simple channel loss component was as accurate as KINEROS. Reasons for poor simulations were investigated by examining a) rainfall sampling errors, b) model sensitivity and dynamics, and c) trends in simulation accuracy. The cause of poor simulations was divided between rainfall sampling errors and other problems. It was found that when raingage densities are on the order of 1/20 km2, rainfall sampling errors preclude the consistent and reliable simulation of runoff from localized thunderstorms. Even when rainfall errors were minimized, accuracy of simulations were still poor. Good results, however, have been obtained with KINEROS on small watersheds; the problem is not KINEROS itself but its application at larger scales. The study also examined the hydrology of thunderstorm -generated floods at Walnut Gulch. The space -time dynamics of rainfall and runoff were characterized and found to be of fundamental importance. Hillslope infiltration was found to exert a dominant control on runoff, although flow hydraulics, channel losses, and initial soil moisture are also important. Watershed response was found to be nonlinear.

Improving the capacity to make predictions in ungauged basins is one of most difficult challenge for the scientific community (see for example the current initiative Prediction Ungaged Basins (PUB) launched by the International Association of Hydrological Sciences, IAHS). Whatever hydrological models are used, in view of the tremendous spatio-temporal heterogeneity of climatic and landscape properties, extrapolation of information, or knowledge, from gauged to ungauged basins remains fraught with considerable difficulties and uncertainties, especially in the light of the generally poor understanding of where water goes when it rains, what flow path it takes to the stream, and the age of the water that emerges in the channel. The PUB problem is the key concept of this thesis and it is analysed from several point of view. Methodologies able to observe, model and predict the hydrological response at the regional scale are proposed.

Eastern Kentucky is a 35-county region that is a part of the Cumberland Plateau of the Appalachian Mountains. With mountaintop removal and associated land cover change (LCC) (primarily deforestation), it is hypothesized that there would be changes in various atmospheric boundary layer parameters and precipitation. In this research, we have conducted sensitivity experiments of atmospheric response of a significant flash flood-producing rainfall event by modifying land cover and topography. These reflect recent LCC, including mountaintop removal (MTR). We have used the Weather Research and Forecasting (WRF) model for this purpose. The study found changes in amount, location, and timing of precipitation. LCC also modified various surface fluxes, moist static energy, planetary boundary layer height, and local-scale circulation wind circulation. The key findings were the modification in fluxes and precipitation totals. With respect to sensible heat flux (H), there was an increase to bare soil (post-MTR) in comparison to pre-MTR conditions (increased elevation with no altered land cover). Allowing for growth of vegetation, the grass simulation resulted in a decrease in H. H increased when permitting the growth of forest land cover (LC) but not to the degree of bare soil. In regards to latent heat flux (LE), there was a dramatic decrease transitioning from pre-MTR to post-MTR simulations. Then with the subsequent grass and forest simulations, there was an increase in LE comparable to the pre-MTR simulation. Under pre-MTR conditions, the total precipitation was at its highest level overall. Then with the simulated loss of vegetation and elevation, there was a dramatic decrease in precipitation. With the grass LC, the precipitation increased in all areas of interest. Then forest LC was simulated allowing overall slightly higher precipitation than grass.

Two main approaches to enhance urban pluvial flood prediction were developed and tested in this research: (1) short-term rainfall forecast based on rain gauge networks, and (2) customisation of urban drainage models to improve hydraulic simulation speed. Rain gauges and level gauges were installed in the Coimbra (Portugal) and Redbridge (UK) catchment areas. The collected data was used to test and validate the approaches developed. When radar data is not available urban pluvial flooding forecasting can be based on networks of rain gauges. Improvements were made in the Support Vector Machine (SVM) technique to extrapolate rainfall time series. These improvements are: enhancing SVM prediction using Singular Spectrum Analysis (SSA) for pre-processing data; combining SSA and SVM with a statistical analysis that gives stochastic results. A method that integrates the SVM and Cascade-based downscaling techniques was also developed to carry out high-resolution (5-min) precipitation forecasting with longer lead time. Tests carried out with historical data showed that the new stochastic approach was useful for estimating the level of confidence of the rainfall forecast. The integration of the cascade method demonstrates the possibility of generating high-resolution rainfall forecasts with longer lead time. Tests carried out with the collected data showed that water level in sewers can be predicted: 30 minutes in advance (in Coimbra), and 45 minutes in advance (in Redbridge). A method for simplifying 1D1D networks is presented that increases computational speed while maintaining good accuracy. A new hybrid model concept was developed which combines 1D1D and 1D2D approaches in the same model to achieve a balance between runtime and accuracy. While the 1D2D model runs in about 45 minutes in Redbridge, the 1D1D and the hybrid models both run in less than 5 minutes, making this new model suitable for flood forecasting.

Expert computer programs have recently emerged from research on artificial intelligence as a practical problem-solving tool. An expert system is a knowledge-based program that imitates the problem-solving behaviour of a human expert to solve complex real-world problems. While conventional programs organize knowledge on two levels: data and program, most expert programs organize knowledge on three levels: data, knowledge base, and control. Thus, what distinguishes such a system from conventional programs is that in most expert systems the problem solving model is treated as a separate entity rather than appearing only implicitly as part of the coding of the program. The purpose of this thesis is twofold. First, it is intended to demonstrate how domain-specific problem-solving knowledge may be represented in computer memory by using the frame representation technique. Secondly, it is intended to simulate a typical flood estimation situation, from the point-of-view of an expert engineer. A frame network was developed to represent, in data structures, the declarative, procedural, and heuristic knowledge necessary for solving a typical flow estimation problem. The control strategy of this computer-based consultant (FLOOD ADVISOR) relies on the concept that reasoning is dominated by a recognition process which is used to compare new instances of a given phenomena to the stereotyped conceptual framework used in understanding that phenomena. The primary purpose of the FLOOD ADVISOR is to provide interactive advice about the flow estimation technique most suitable to one of five generalized real-world situations. These generalizations are based primarily on the type and quantity of the data and resources available to the engineer. They are used to demonstrate how problem solving knowledge may be used to interactively assist the engineer in making difficult decisions. The expertise represented in this prototype system is far from complete and the recommended solution procedures for each generalized case are in their infancy. However, modifications may be easily implemented as the domain-specific expert knowledge becomes available. It is concluded that over the long term, this type of approach for building problem-solving models of the real world are computationally cheaper and easier to develop and maintain than conventional computer programs.
Applied Science, Faculty of
Civil Engineering, Department of
Graduate

A fast catchment response usually leads to a shorter lag time, and under these conditions the forecast lead time obtained from a rainfall-runoff model or correlation between upstream and downstream flows may be infeasible for flood warning purposes. Additional lead time can be obtained from short-term quantitative rainfall forecasts that extend the flood warning time and increase the economic viability of a flood forecasting system. For this purpose algorithms which forecasts the quantitative rainfall amounts up to six hours ahead have been developed, based on lumped and distributed approaches. The lumped forecasting algorithm includes the essential features of storm dynamics such as rainband and raincell movements which are represented within the framework of a linear transfer function model. The dynamics of a storm are readily captured by radar data. A space-time rainfall model is used to generate synthetic radar data with known features, e.g. rainband and raincell velocities. This enables the algorithm to be assessed under ideal conditions, as errors are present in observed radar data. The transfer function algorithm can be summarised as follows. The dynamics of the rainbands and raincells are incorporated as inputs into the transfer function model. The algorithm employs simple spatial cross-correlation techniques to estimate the rainband and raincell velocities. The translated rainbands and raincells then form the auxiliary inputs to the transfer function. An optimal predictor based on minimum square error is then derived from the transfer function model, and its parameters are estimated from the auxiliary inputs and observed radar data in real-time using a recursive least squares algorithm. While the transfer-function algorithm forecasts areal rainfalls, a distributed approach which performs rainfall forecasting at a fine spatial resolution (referred to as the advection equation algorithm) is also evaluated in this thesis. The algorithm expresses the space-time rainfall on a Cartesian coordinate system via a partial differential advection equation. A simple explicit finite difference solution scheme is applied to the equation. A comparison of model parameter estimates is undertaken using a square root information filter data processing algorithm, and single-input single-output and multiple-input multiple-output least squares algorithms.

A flash flood is a flood that follows the causative storm event in a short period of time. The term “flash” reflects a rapid response, with water levels in the drainage network reaching a crest within minutes to a few hours after the onset of the rain event, leaving extremely short time for warning [Creutin and Borga, 2003; Borga et al., 2008]. Flash floods are localized phenomena that occur in watersheds of few hundred kilometres or less, with response times of a few hours or less [Creutin and Borga, 2003; O’Connor and Costa, 2004]. Such basins respond rapidly to intense rainfall because of steep slopes and impermeable surfaces, saturated soils, or because of human (i.e., urbanization) or fire-induced alterations to the natural drainage. Causative events are generally excessive storms, but can also be the sudden release of water impounded by a natural jam (i.e., formed by ice or rock, mud, and wood debris) or human-made dam or levee. This thesis focuses on flash flood events associated with heavy rainfall. Europe experienced several catastrophic flash floods in the last decades. Data concerning a number of these floods occurred during the last 15 years have been reported in Marchi et al. (2010). Examination of these data and references therein shows that: Flash floods occur in any of the hydroclimatic regions of Europe, even though three regions appear to be characterized by high flash flood potential: Mediterranean, Alpine Mediterranean, and Inland Continental Europe; Heavy rainfall accumulation is a necessary but not sufficient condition for flash floods, since hydrology critically controls flash-flood-triggering. Without hydrological analysis, it is impossible to evaluate the flood potential of storms, particularly in the fringe of the flood/no flood threshold; Flash flood hazard is related to both stream response (flood) and landscape response (landslide and erosion). The intense erosion and solid transport associated with these extreme events add to the hazard and strongly influence the quality of soils, waters and ecosystems. The twofold consequence of the above observations is that forecasting of flash-floods: Depends critically on meso-scale storm forecasting, with a specific attention to the processes leading to slow movement of the precipitation system; Necessitates real time hydrological modelling, with a specific attention to the runoff generation processes over a wide range of scales. Although they are seldom all deployed at the same time, the technical requirements for a hydrometeorological flash flood forecasting system include: A numerical weather prediction (NWP) model, capable to provide short-range Quantitative Precipitation Forecasts (QPF); A remote sensing based (radar, satellites) precipitation detection system, for storm monitoring and for the possible initialization and conditioning of the NWP model, and A hydrological-hydraulic forecasting model, capable to forecast the stream response from the rain input. These requirements are similar to those of more common riverine flood forecasting systems. However, some features characterise flash flood forecasting with respect to riverine flood forecasting and point out to their larger uncertainty. These are: The short lead time, which implies both the integration of meteorological and hydrologic forecast, and the difficulties of using data assimilation procedures based on real time observed discharges to reduce uncertainty in hydrologic predictions; The need to provide local forecasts, which means that, on one hand, the rainfall must be monitored and forecasted on a wide range of space/time scales, and, on the other hand, every tributary of a monitored basin can be considered as a potential target for flood warning. Estimation of extreme rainfall rates by weather radar at the appropriate time and space scales is the cornerstone of flash flood analysis and forecasting. A large body of research work has greatly improved in the last two decades radar technology and algorithms for rain quantification. This work has shown that well maintained conventional radar systems can estimate rainfall at ground level provided that a number of precautions are taken, and in particular: The siting of the instrument and its scanning protocol must be carefully selected and analysed; The quality of the instrument must be routinely checked; The signal processing must take into account the physics of the instrument as well as the properties of the atmospheric and ground targets. A downstream control of the radar rainfall processing can rely on rain-gauge measurements at ground level using a variety of methods. When these precautions are taken, different studies have shown that radar-based rainfall estimates are reliable and may be used as input in rainfall-runoff models for flood modelling and forecasting [Borga et al., 2000; Delrieu et al., 2005; Borga et al., 2002]. These very positive results must not be hiding some weaknesses: Most of these results never had the opportunity to be coherently validated over a significant number of flash floods events. The use of specific experiments or of limited operational radar data sets is insufficient to test complex combinations of algorithms, especially if high rain intensities are of interest. Very few results have been translated into operational hydrologic applications. This thesis aims to investigate the use of weather radar for the purpose of understanding the hydrometeorological mechanisms leading to flash floods, and then for flash flood forecasting. The outline of the thesis work is as follows. Chapter 1 provides a literature review of the rainfall estimation by weather radar for flash flood-generating storms. Chapter 2 describes a number of procedures for the rainfall estimation at the ground during flash flood events in mountainous catchments. A metric for the analysis of the rainfall field spatial patterns is proposed in Chapter 3, in the context of the analysis of a number of Romenian flash floods. This metric is used for the analysis of two flash flood events, respectively occurred in 2003 in the Eastern Italian Alps (Chapter 4) and in Western Slovenia (Chapter 5). Major conclusions from the work are reported in Chapter 6.
Una piena improvvisa è una piena che segue l’evento precipitativo che la ha causata entro un breve periodo di tempo. Il termine “improvvisa o flash” riflette una risposta rapida, con il picco di piena che si verifica nella rete di drenaggio nel volgere di alcuni minuti fino a poche ore dopo l’inizio dell’evento di pioggia. Questo fatto lascia intendere quanto poco tempo ci sia per l’allerta [Creutin and Borga, 2003; Borga et al., 2008]. Questo tipo di bacini rispondo rapidamente ad una precipitazione intensa a causa di pendii ripidi e superfici impermeabili, terreni saturi, o a per fattori determinati dall’uomo (vedi per esempio l’urbanizzazione) o a causa di alterazioni del drenaggio naturale del terreno dovuto ad incendi. Gli eventi scatenanti le piene improvvise sono generalmente precipitazioni che portano all’eccesso di drenaggio, ma questo tipo di piene possono anche essere scatenate dal rilascio improvviso di acqua trattenuta da impedimenti naturali (per esempio formati da ghiaccio e roccia, fango e detriti di legno) o di tipo artificiale come dighe e argini. Questa tesi si concentra su eventi di piena improvvisa associati a precipitazioni intense. L’Europa ha conosciuto diverse inondazioni catastrofiche negli ultimi decenni. I dati relativi un certo numero di queste inondazioni che si sono verificate nel corso degli ultimi 15 anni sono riportati da Marchi et al. (2010). Dall’analisi di questi dati e di queste fonti risulta che: Una piena improvvisa si può verificare in qualsivoglia regione idroclimatica dell’Europa, anche se tre regioni sembrano essere caratterizzate da una grande incidenza di di piene improvvise: l’area Mediterranea, quella Alpino-Mediterranea, e quella Continentale; Una gran quantità di pioggia accumulata è una condizione necessaria ma non sufficiente al verificarsi di una piena improvvisa, dal momento che l’idrologia controlla in modo decisivo l’innesco della piena improvvisa. Senza un’analisi di tipo idrologico, risulta impossibile valutare la probabilità che una data precipitazione scateni una piena, in praticolare in termini di una soglia oltre la quale si verifica la piena; La pericolosià delle piene improvvise è collegata sia alla risposta del fiume (la piena) che alla risposta del terreno (fenomeni di tipo franoso ed erosivo). L’intensa erosione ed il trasporto solido associati a questi fenomeni estremi si aggiungono alla pericolosità ed influenzano in modo significativo la qualità dei terreni, delle acque e degli ecosistemi. La duplice conseguenza delle osservazioni appena fatte è che la previsione di piene improvvise: Dipende in modo determinante dalle previsioni delle precipitazioni che si sviluppano alla meso-scala, con una attenzione specifica ai processi che frenano la circolazione del sistema di precipitazione; Richiedone modelli idrologici che lavorino in tempo reale, con una particolare attenzione ai processi du generazione del deflusso a vasta scala. Anche se raramente sono tutti utilizzati contemporaneamente, i requisiti tecnici per un sistema di previsione idrometeorologica per le piene improvvise comprendono: Un modello numerico di previsione (NWP2), in grado di fornire previsioni quantitative di pioggia a corto raggio (QPF3); Un sistema di rilevamento in remoto per la pioggia (radar, satellite), per il monitoriraggio dei fenomeni temporaleschi e la possibilie inizializzazione e condizionamento del modello NWP, e Un modello di previsione idrologico-idraulico, in grado di prevedere la risposta del corso d’acqua all’input pioggia. Tali requisiti sono simili a quelli più comuni utilizzati per la previsione delle alluvioni dei sistemi fluviali. Tuttavia, alcuni elementi caratterizzano la previsione delle piene improvvise rispetto alla previsione delle alluvioni e ne sottolineano la grande incertezza. Questi sono: Il breve periodo durante il quale questi processi si sviluppano, che implica sia l’integrazione di un sistema di previsione di tipo meteorologico e idrologico, che la difficoltà nell’utilizzo di procedure di assimilazione di dati basate sull’osservazione in tempo reale delle portate al fine di ridurre l’incertezza nelle previsioni idrologiche; La necessità di fornire previsioni a scala locale, il che significa da una parte che la pioggia deve essere monitorata e prevista su una vasta scala spazio-temporale, all’altra che ciascun tributario del bacino monitorato può essere considerato come un bersaglio potenziale per un allarme di piena. La stima di fenomeni precipitativi estremi tramite l’utilizzo del radar meteorologico alla appropriata scala spazio-temporale è una pietra miliare dell’analisi e della previsione delle piene improvvise. Una grande branca della ricerca in questo campo ha favorito un notevolmente migliorato, negli ultimi due decenni, delle tecnologie radar e degli algoritmi per la stima di pioggia. Questo lavoro ha dimostrato che anche utilizzando sistemi radar convenzionali si possono ottenere stime di precipitaziona a livello del suolo, a condizione che vengono adottate una serie di precauzioni, in particolare: L’ubicazione dello strumento e del suo protocollo di scansione devono essere attentamente selezionati ed analizzati; La qualità dello strumento deve essere sottoposta a controlli ordinari; L’elaborazione del segnale deve tener conto della fisica dello strumento così come delle proprietà atmosferiche e dei bersagli di terra. Un controllo a valle del trattamento delle precipitazioni radar può essere fatto tramite misurazioni da pluviometro a livello del suolo utilizzando una varietà di metodi. Quando si sono prese queste precauzioni, diversi studi hanno dimostrato che le stime di precipitazione basate su radar meteorologico sono affidabili e possono essere utilizzate come input di modelli afflussodeflusso per la modellazione e la previsione delle piene [Borga et al., 2000; Delrieu et al., 2005; Borga et al., 2002]. A fronte di questi risultati molto positivi non devono però essere nascosti alcuni punti deboli: La maggior parte di questi risultati non hanno mai la possibilità di essere coerentemente convalidati su un numero significativo di eventi di piena improvvisa. L’utilizzo di esperimenti specifici o di una banca dati limitata di dati radar è insufficiente a testare la combinazione complessa degli algoritmi utilizzati, specialmente se si è interessati ad intensità di pioggia elevata. Un numero molto limitato di risultati positivi è stato tradotto in applicazioni idrologiche operative. Questa tesi si propone di esaminare l’uso del radar meteorologico ai fini della comprensione dei meccanismi idrometeorologici che portano alla formazione di piene improvvise, e quindi alla loro previsione. L’organizzazione del lavoro di tesi è la seguente. Il Capitolo 1 fornisce una revisione della letteratura sul tema della stima di precipitazione tramite radar meteorologico per le precipitazioni che causano la formazione di piene improvvise. Il Capitolo 2 descrive una serie di procedure per la stima delle precipitazioni al suolo durante gli eventi di piena improvvisa in bacini montani. Una metrica per l’analisi spaziale del campo di pioggia viene proposta nel Capitolo 3, nel contesto dell’analisi di una serie di piene improvvise verificatesi in Romania. Questa metrica è utilizzata per l’analisi di due eventi di piena, accaduti rispettivamente nel 2003 nelle Alpi Italiane friulane e nella parte ovest della Slovenia (Capitolo 5). Le conclusioni principali del lavoro di tesi sono riportate nel Capitolo 6.

Design flood estimates have traditionally been based on records of past events. However, there is a need for a method of estimating peak flows without these records. The Estimated Parameter Flood Forecasting Model (EPFFM) has been developed to provide such a method for small water resource projects based on a 200 year or less design flood. This "user friendly" computer model calculates the expected peak flow and its standard deviation from low, probable, and high estimates of thirteen user supplied parameters. These parameters describe physical characteristics of the drainage basin, infiltration rates, and rainstorm characteristics. The standard deviation provides a measure of reliability and is used to produce an 80% confidence interval on peak flows. The thesis briefly reviews existing flow estimation techniques and then describes the development of EPFFM. This includes descriptions of the Chicago method of rainfall hyetograph synthesis, Horton's infiltration equation, inflow by time-area method, Muskingum routing equation, and an approximate method of estimating the variance of multivariate equations since these are all used by EPFFM to model the physical and mathematical processes involved. Two examples are included to demonstrate EPFFM's ability to estimate a confidence interval, and compare these with recorded peak flows.
Applied Science, Faculty of
Civil Engineering, Department of
Graduate

Floods have potential destructive effects on socioeconomic facilities and cause serious risks for people. During the last decades lots of efforts have been carried out 10 overcome the difficulties caused by this natural phenomenon. In the past, most of the studies have been focused on developing mathematical models to forecast flood events in real -time to provide precautionary activities. The models are various from simple structures to models with high complexity and according to the climate conditions of the catchment under study, most appropriate model must be selected to predict flood events by using the existing recorded data from the catchment Rainfall-runoff model is the main component of a real-time flood forecasting model and transforms rainfall to runoff. The model commonly consists of a number of mathematical equations and parameters which are interconnected together for simulating runoff over a catchment. Since a model is a simplification of the real hydrological system, errors In simulation are unavoidable and influence on the simulation accuracy. Hence. the model should be selected properly and requires to be updated continuously to cope with probable hydrological changes which could create errors on model simulations. The current research focus on real-time flood forecasting by improving and developing rainfall-runoff models and indicating solutions to update the model to cope with frequent hydrological changes which can reduce the model performance. The research was started by evaluating optimisation schemes to derive the model parameters and an optimisation method was proposed based on Genetic algorithm concept. On the second stage, a new rainfall -runoff model called ERM, was introduced and suggested as a reliable model to use In rainfall -runoff modeling. Moreover, the adaptability of the ERM model parameters to cope with different errors occurred in terms of modeling was considered. Finally, in the last part of the thesis, the ERM model was coupled with a well-known numerical filter called the Kalman Filter and a real-time flood forecasting model was introduced.

The need for reliable, easy to set up and operate, hydrological forecasting systems is an appealing challenge to researchers working in the area of flood risk management. Currently, advancements in computing technology have provided water engineering with powerful tools in modelling hydrological processes, among them, Artificial Neural Networks (ANN) and genetic algorithms (GA). These have been applied in many case studies with different level of success. Despite the large amount of work published in this field so far, it is still a challenge to use ANN models reliably in a real-time operational situation. This thesis is set to explore new ways in improving the accuracy and reliability of ANN in hydrological modelling. The study is divided into four areas: signal preprocessing, integrated GA, schematic application of weather radar data, and multiple input in flow routing. In signal preprocessing, digital filters were adopted to process the raw rainfall data before they are fed into ANN models. This novel technique demonstrated that significant improvement in modelling could be achieved. A GA, besides finding the best parameters of the ANN architecture, defined the moving average values for previous rainfall and flow data used as one of the inputs to the model. A distributed scheme was implemented to construct the model exploiting radar rainfall data. The results from weather radar rainfall were not as good as the results from raingauge estimations which were used for comparison. Multiple input has been carried out modelling a river junction with excellent results and an extraction pump with results not so promising. Two conceptual models for flow routing modelling and a transfer function model for rainfall-runoff modelling have been used to compare the ANN model's performance, which was close to the estimations generated by the conceptual models and better than the transfer function model. The flood forecasting system implemented in East Anglia by the Environment Agency, and the NERC HYREX project have been the main data sources to test the model.

ABSTRACT: “Flash flood analysis and modelling in mountain regions” Flash flood are rare and localized phenomena, triggered by meteorological event with a pronounced spatial variability, with a precipitation gradient, at event scale, up to 20-50 mm/km. The consequences of these features is that the scientific and operational communities working on flash flood analysis have to deal with an impressive lack of data. Even a dense raingauge network may not be able to represent spatial variability of rainfall patterns associated with convective storms that trigger flash floods. Radar rainfall estimations, when correctly elaborated, are able to represent spatial patterns, but quantitative precipitation volume estimations need to be validated. In addition, concerning discharge data, the majority of the upstream and larger catchments affected by flash floods are not gauged and stream gauges, where present, are often damaged, so that peak discharge distribution along main and secondary river network is even less known than precipitation fields. This study aims at covering the gap between needed and available data for flash flood event analysis, combining different methodologies. An Intense Post Event Campaign (IPEC) may be very useful to collect peak discharge estimations and time sequence of the flood in ungauged sections. Simplified hydrological model, based on rough runoff excess computation and set velocity propagation, can be used to cross validate quantitative distributed precipitation data from weather radar and peak discharge estimations collected during an IPEC. More complex and detailed model may help to improve the knowledge about flash flood associated phenomena, like debris flow. Another objective of this thesis is to investigate the role of rainfall spatial variability in flash flood triggering. First a standard procedure to describe the variability catchment scale is needed. It will be so possible to study the relationship between rainfall input distribution and flood propagation dynamics. Then a simplified hydrological model is used to investigate the role of spatial variability in precipitation patterns: systematic studies are carried to describe the accuracy of rainfall volumes at basin scale and the effect of spatial variability within the basin. Often the studies about flash flood dynamics are slow down or stopped because no measured data are directly to hand, or, if so, because they are not considered sufficiently accurate. This work shows the possibility to combine together data with an assured degree of uncertainty, the only available or collectable existing data, and processed them with simple statistical and hydrological tools to obtain a more precise knowledge about past flash floods. The remainder of this dissertation is organised as follows. Chapter 1 “Introduction”. The work starts with the aim to define what a “flash flood” is, underlying the importance to characterize such event according spatial and temporal proprieties. From this definition it follows that a generic flood can be classified according its own spatial and temporal proprieties and located in a specific point of a segment delimitated by the two ideal cases of “flash flood” and “flood at large scale”. Chapter 2 “Literature review”. Spatial and temporal characterization lead to describe typical features of flash flood in different climates. Meteorological conditions able to trigger this kind of events are described and analyzed, with particular care about convective cells system organized in mesoscale structures. Finally some literature examples are reported to show different possible approach and to underline usual uncertainty when dealing with flash flood. Chapter 3 “Materials an methods”. This chapter summarizes and describes the tools used in this thesis to carry on flash flood analysis. 3.1 Weather radar data are used to describe rainfall spatial distribution and obtain quantitative estimations of rainfall patterns. Data acquiring and processing are described and most common errors are summarized along with most common procedures and algorithms to avoid and correct them. It is finally shown how merging radar and conventional raingauge network information can provide a more exhaustive description of rainfall fields, with quantitative estimation. This data processing is very useful for further characterization and analysis of past flash flood events. 3.2 Post event surveys are presented as an essential tool to collect the richest possible documentation. Measure campaigns are valorised to obtain qualitative and quantitative description of past floods. The goal is to complete the spatial and temporal precipitation knowledge and dynamic description, focusing on discharge estimation along hydrological network in term of peak values and timing. 3.3 Hydrological models can be routed for a better comprehension of flood dynamics at event scale. Two hydrological models, then used for flash flood analysis, are described in detail. The first one is applied at large event scale and starts from a distributed precipitation input. Hortonian runoff generation is applied punctually and superficial flood propagation is computed basing on fixed hillslope and channel velocity. The second model is built to be applied at very small catchment scale and simulate infiltration and transport processes for surface and subsurface flow through uniform hypothesis equations. Chapter 4 “Analysis of past flash flood events”. Some specific post flood analysis are collected in three section. 4.1 Five flash flood events occurred in Romania are analysed with HYDRATE European project contribution. This study shows that even if the conventional hydrometeorological data are poor, weather radar information and hydrological modelling can help in understanding specific past flood dynamics. 4.2 HYDRATE project was also involved in the analysis of a flash flood occurred in Slovenia in September 2007, including radar processing and post event surveys. It is shown how this approach, characterize by time and cost significant efforts, is a precious tool to collect data and information for a detailed description that would be not possible through traditional hydrometeorological network. 4.3 A detailed model is used to describe surface and subsurface flow dynamics during the debris flow occurred in two small subcatchments in Fella river valley (North Est of Italy), hit by a flash flood on August 29, 2003. The study mainly consists on liquid and solid mass balance during the different phases of the event. Chapter 5 “Spatial variability in flash flood events”. An analysis on rainfall spatial distribution is carried with the same tools on two different basin interested by flash flood event. The studies includes a fist detailed analysis on rainfall spatial variability within selected subcatchments at different scales: spatial variability is described through time distance calculated in base of hydrological network. Then a simplified hydrological model is used to investigate spatial aggregation effects on mean areal rainfall and peak discharge value at subcatchment scale. 5.3 For Fella river basin (in Friuli Venezia Giulia region), ten subcatchments from 10.5 and 623km² are choosen. 5.4 For Cervo River (Piomente region, North West Italy) the study is applied to three flood events characterized by different rainfall spatial variability, and focused on four subcatchments (from 75 to 983km²). Chapter 6 “Conclusions”. Are here reported and summarized the main observations coming from the specific studies describe in the two previous chapters as long as recommendation for future research.
RIASSUNTO: “Analisi e modellazione di piene improvvise in zone montane” Le piene improvvise sono fenomeni rari e localizzati, causati da eventi meteorologici caratterizzati da una spiccata variabilità spaziale, con gradienti di precipitazione che possono raggiungere, a scala di evento, i 20-50 mm/km. La conseguenza di ciò è che la comunità scientifica e gli enti operativi interessati nell’analisi dei fenomeni di piena si relazionano quotidianamente con una carenza di dati. Anche una fitta rete di pluviometri non è in grado di rappresentare la variabilità spaziale dei campi di precipitazione associati a fenomeni convettivi che innescano piene improvvise. Le stime di precipitazione ottenute attraverso il radar meteorologico, opportunatamente elaborate, sono in grado di rappresentare i pattern spaziali, ma i valori di volumi di pioggia necessitano di essere validati. Inoltre, per quanto riguarda i dati di portata, la maggior parte dei bacini colpiti da piene improvvise non sono strumentati e gli strumenti, dove presenti, risultano spesso danneggiati, cosicché la conoscenza della distribuzione delle portate al picco, lungo la rete idrologica principale e secondaria, è persino più approssimativa di quella della distribuzione spaziale della precipitazione. Questo studio si prefigge di colmare la distanza tra i dati disponibili e quelli richiesti per un’analisi a scala di evento con riferimento a fenomeni di piena improvvisa. Un’approfondita campagna di rilievi post evento (in inglese Intense Post Event Campaign, IPEC) può risultare estremamente utile per raccogliere le stime di portate al picco e la sequenza cronologica dello svilupparsi della piena in sezioni non monitorate. Modelli idrologici semplificati, dotati di metodi elementari per la separazione dei deflussi e predeterminate velocità di propagazione, possono essere utilizzati per una validazione incrociata tra una descrizione quantitativa della distribuzione di precipitazione ottenuta attraverso il radar meteorologico e le stime di portate al picco raccolte durante un IPEC. Modelli più complessi e dettagliati possono migliorare il livello di conoscenza riguardo fenomeni associati alle piene improvvise, come le colate detritiche. Un altro obiettivo di questa tesi è quello di investigare il ruolo della variabilità spaziale della precipitazione nei fenomeni di piena improvvisa. In primo luogo è necessario impostare una procedura che permetta di caratterizzare tale variabilità all’interno di un particolare bacino idrografico, mettendo in relazione la distribuzione degli apporti meteorici con le modalità di propagazione della piena. In secondo luogo si vuole indagare, attraverso l’applicazione di modelli idrologici semplificati, il ruolo della risoluzione spaziale della precipitazione. A questo fine è necessario separare due aspetti: l’accuratezza della stima dei volumi piovuti a scala di bacino e l’influenza della variabilità spaziale all’interno del bacino stesso. Spesso gli studi che si concentrano sulle dinamiche delle piene improvvise sono rallentati o resi impossibili per il fatto che nessun dato misurato risulta utilizzabile così come disponibile, oppure perchè i dati di partenza non sono ritenuti sufficientemente accurati. Questo lavoro si prefigge di mostrare come sia possibile, partendo dai soli dati esistenti, disponibili o recuperabili, caratterizzati da un certo grado di incertezza, passare attraverso un’elaborazione tramite semplici strumenti statistici e idrologici al fine di ottenere una conoscenza più precisa riguardo passati eventi di piena improvvisa. Si riporta una breve descrizione del contenuto dei capitoli della tesi, che sarà elaborata in lingua inglese. Capitolo 1 “Introduction”. Introduzione alla tematica che comprende una definizione del termine “piena improvvisa”, convenendo sulla necessità di caratterizzare tali eventi in termini di proprietà spazio-temporali. Si nota che, a partire da questa definizione, è possibile classificare una generica piena in un punto di un segmento ai cui estremi ci sono i casi ideali di “piena improvvisa” e “piena a larga scala”. Capitolo 2 “Literature review”. Partendo dalla caratterizzazione spazio temporale si descrivono le caratteristiche tipiche delle piene improvvise nei diversi tipi di clima, si individuano le condizioni meteorologiche in grado di innescare tali fenomeni, quali le celle convettive organizzate in strutture di mesoscala. Si riportano, infine, alcuni esempi di studi in letteratura che mostrano diverse tipologie di approcci e che sono indicativi dell’incertezza in cui si è soliti lavorare quando si approfondiscono questi temi. Capitolo 3 “Materials an methods”. In questo capitolo vengono presentati i principali strumenti comuni a tutte le analisi di fenomeni di piena improvvisa presentati in questa tesi. 3.1 L’utilizzo del radar meteorologico per studiare, dal punto di vista quantitativo, la distribuzione spaziale della precipitazione. Vengono approfondite la modalità di acquisizione del dato, sottolineando le possibili fonti di errore ed i metodi più comuni per ovviare a questi inconvenienti. Viene anche mostrato come l’utilizzo combinato di radar e tradizionali pluviometri renda più completa la caratterizzazione della precipitazione ai fini di un analisi di una piena improvvisa. 3.2 Le indagini post evento, necessarie per raccogliere la maggior documentazione possibile, sono valorizzate al fine di una ricostruzione, anche qualitativa, delle dinamiche caratteristiche di una specifica piena. Queste, attraverso diverse metodologie, devono aiutare a descrivere la struttura spazio temporale della precipitazione e la stima di portata, distribuita lungo la rete idrica, in termini di valore al picco e di tempistica 3.3 L’uso della modellistica idrologica applicata ad una miglior comprensione delle dinamiche a scala di evento. In particolare vengono descritti i due modelli idrologici utilizzati. Il primo, da applicare a larga scala, parte da un input di precipitazione spazialmente distribuito e, attraverso un meccanismo hortoniano di separazione dei deflussi applicato puntualmente, propaga la piena in base a fissate velocità di versante e di canale. Il secondo, da applicare a bacini di piccolissima dimensione, simula i processi di trasporto superficiale e sottosuperficiale integrando le note equazioni di moto uniforme. Capitolo 4 “Analysis of past flash flood events”. Vengono qui presentate alcune analisi di eventi, distinte in tre sezioni. 4.1 Analisi di cinque eventi di piena improvvisa avvenuti in Romania nell’ambito del progetto europeo HYDRATE. Da questo studio risulta che, pur in presenza di scarsi dati provenienti dalle tradizionali fonti di monitoraggio idro-meteorologico, l’informazione proveniente da radar meteorologico e la modellistica idrologica possono aiutare nella ricostruzione delle dinamiche dell’evento preso in considerazione. 4.2 Analisi di una piena improvvisa avvenuta in Slovenia nel settembre 2007 per la quale, attraverso il progetto HYDRATE si è condotta un indagine post evento. La ricchezza di questo approccio, pur dispendioso in termini di tempo, mostra un possibile percorso per recuperare le maggior informazioni possibili per eventi di piena che non sono ricostruibili solo attraverso le normali reti di monitoraggio idrometeorologico. 4.3 Analisi attraverso un modello dettagliato di deflusso superficiale e sottosuperficiale della colata detritica avvenuta in due piccoli sottobacini nella valle del fiume Fella, colpita da una piena improvvisa il 29 agosto 2003. Lo studio consiste essenzialmente nel bilancio di massa liquido e solido durante le diverse fasi dell’evento. Capitolo 5 “Spatial variability in flash flood events”. Questa analisi sulla distribuzione spaziale della precipitazione è stata condotta con le medesime metodologie in due diversi bacini. Gli studi comprendono un primo approfondimento della variabilità spaziale della precipitazione all’interno di sottobacini di diversa estensione: la variabilità è descritta in funzione del reticolo idrografico del bacino preso in considerazione. Successivamente, attraverso un modello idrologico semplificato, si è valutata l’influenza della variabilità spaziale della precipitazione analizzando gli effetti dell’aggregazione spaziale in termini di precipitazione media su bacino e di portata al picco simulata. 5.3 Per l’analisi nel bacino del fiume Fella (FVG), colpito da una piena improvvisa il 29 agosto 2003, si sono scelti dieci sottobacini di dimensione variabile tra i 10.5 e i 623km². 5.4 Nel caso del fiume Cervo (Piemonte) lo studio ha riguardato tre eventi di piena con diversa variabilità spaziale della precipitazione e si è concentrato su quattro sottobacini (tra i 75 e i 983km²). Capitolo 6 “Conclusions”. Vengono riassunte le principali osservazioni ricavate dalle analisi descritte nei due capitoli precedenti e indicazioni per possibili future linee di ricerca.

In an effort to improve upon rainfall forecasts produced by simple storm advection methods (nowcasts) and to broach the gap between them and the forecasts of complex Numerical Weather Prediction (NWP) models, in terms of the spatial detail and length of lead-time each provides, the research presented explores the possibility of combining elements of each into a physically-based algorithm for rainfall forecasting. It is an algorithm that uses as its foundation the rainfall prediction model of Mark French and Witold Krajewski, developed in 1994. Their model was designed to take advantage of the high resolution rainfall observations and tracking abilities provided by weather radar and to achieve a rainfall forecast by augmenting extrapolation techniques with a representation of storm dynamics in the form of "rising parcel" theory. The new algorithm/model retains those features but incorporates NWP data to assist with forecasting, using it as a means to enable an informed choice of algorithm pathways and, more specifically, to identify the ingredients of precipitation, namely ascending air of high moisture content. A case study application of the new rainfall forecasting model to storms in Northern England shows its performance, at a lead-time of one hour, compares favourably with respect to extrapolation and persistence techniques and also NWP forecasts, and that it is able to provide more assured forecasts than persistence and nowcasts at longer lead-times. The robustness of the model is tested and confirmed by way of another case study, this time using Mediterranean storms and with predictions made in the context of urban hydrology. The case studies help to identify aspects of the model that need improvement, with representation of orographic forcing being a key one. Both the model's encouraging performance and its pinpointed weaknesses provide impetus for further research in the area of integrated mesoscale-hydrometeorological modelling for flood forecasting.

Aus der Einleitung: The calculation and prediction of river flow is a very old problem. Especially extremely high values of the runoff can cause enormous economic damage. A system which precisely predicts the runoff and warns in case of a flood event can prevent a high amount of the damages. On the basis of a good flood forecast, one can take action by preventive methods and warnings. An efficient constructional flood retention can reduce the effects of a flood event enormously.With a precise runoff prediction with longer lead times (>48h), the dam administration is enabled to give order to their gatekeepers to empty dams and reservoirs very fast, following a smart strategy. With a good timing, that enables the dams later to store and retain the peak of the flood and to reduce all effects of damage in the downstream. A warning of people in possible flooded areas with greater lead time, enables them to evacuate not fixed things like cars, computers, important documents and so on. Additionally it is possible to use the underlying rainfall-runoff model to perform runoff simulations to find out which areas are threatened at which precipitation events and associated runoff in the river. Altogether these methods can avoid a huge amount of economic damage.

Aus der Einleitung: The calculation and prediction of river flow is a very old problem. Especially extremely high values of the runoff can cause enormous economic damage. A system which precisely predicts the runoff and warns in case of a flood event can prevent a high amount of the damages. On the basis of a good flood forecast, one can take action by preventive methods and warnings. An efficient constructional flood retention can reduce the effects of a flood event enormously.With a precise runoff prediction with longer lead times (>48h), the dam administration is enabled to give order to their gatekeepers to empty dams and reservoirs very fast, following a smart strategy. With a good timing, that enables the dams later to store and retain the peak of the flood and to reduce all effects of damage in the downstream. A warning of people in possible flooded areas with greater lead time, enables them to evacuate not fixed things like cars, computers, important documents and so on. Additionally it is possible to use the underlying rainfall-runoff model to perform runoff simulations to find out which areas are threatened at which precipitation events and associated runoff in the river. Altogether these methods can avoid a huge amount of economic damage.:List of Symbols and Abbreviations S. III 1 Introduction S. 1 2 Process based Rainfall-Runoff Modelling S. 5 2.1 Basics of runoff processes S. 5 2.2 Physically based rainfall-runoff and hydrodynamic river models S. 15 3 Portraying Rainfall-Runoff Processes with Neural Networks S. 21 3.1 The Challenge in General S. 22 3.2 State-of-the-art Approaches S. 24 3.3 Architectures of neural networks for time series prediction S. 26 4 Requirements specification S. 33 5 The PAI-OFF approach as the base of the system S. 35 5.1 Pre-Processing of the Input Data S. 37 5.2 Operating and training the PoNN S. 47 5.3 The PAI-OFF approach - an Intelligent System S. 52 6 Design and Implementation S. 55 6.1 Design S. 55 6.2 Implementation S. 58 6.3 Exported interface definition S. 62 6.4 Displaying output data with involvement of uncertainty S. 64 7 Results and Discussion S. 69 7.1 Evaluation of the Results S. 69 7.2 Discussion of the achieved state S. 75 8 Conclusion and FutureWork S. 77 8.1 Access to real-time meteorological input data S. 77 8.2 Using further developed prediction methods S. 79 8.3 Development of a graphical user interface S. 80 Bibliography S. 83

This thesis describes research into remotely sensed weather radar information systems and specifically addresses three problems; 1) Weather radar data processing; 2) Real-time flood forecasting models and 3) Computer system design for the realisation of the real-time flood forecasting system using radar data. Quantitative rainfall measurements utilising weather radar is of high temporal and spatial resolution when compared with traditional rainfall measurements. Analysis was carried out to assessth e type of radar datap roductsr equired for operational use in flood forecastings ystem. This includes issues of data processing such as quantisation, temporal sampling and spatial sampling. The influence of the data process on hydrological applications is also addressed. Theoretical analysis was carried out to probe the characteristics of Transfer Function Models and robust flood forecasting modelling procedure is proposed. The proposed model is always stable and physical realisable and is described as PRTF (Physical Realisable Transfer Function model). Algorithms and software for the identification of PRTF are presented. It was found that such a model is easy to identify and more importantly it can be updated robustly in real time. By changing the impulse response of the PRTF, it has been found that significant improvements can be observed in river flow simulation. A RST (Rainfall S eparation Tank) model was developed and incorporated into the PRTF model. The adaptivity of the PRTF also has the potential to make use of high spatial resolution radar rainfall data and could be further incorporated into an Expert System suitable for real-time application. Finally, the thesis includes the development of the WRIP system (Weather Radar Information Processor). Such a system can process weather radar information and use it for the real-time flood forecasting. The system design consists of database design, user interface design and program design. An object-oriented computing concept is used in the program design. The final system is currently in test operation within the N. R. A Wessex Region, including the man machine interface (MMI) incorporating a portable computer based data acquisition and display system known by the acronym `STORM' (System To Obtain Radar Rainfall Measurements).

Kyoto University (京都大学)
0048
新制・課程博士
博士(工学)
甲第12267号
工博第2596号
新制||工||1366(附属図書館)
24103
UT51-2006-J260
京都大学大学院工学研究科都市社会工学専攻
(主査)教授 小尻 利治, 教授 池淵 周一, 教授 中北 英一
学位規則第4条第1項該当

The dangerous hazard posed by flash flooding to upland communities is likely to increase due to climate change. The flood risk management policy approach has become predominant since the 1990s, with an emphasis on the public awareness of, and responses to, flood risks; however, the unpredictable nature of upland flash flooding means that responses to such hazards are uncertain. This thesis uses an integrated analysis of social and physical science datasets to study responses by local residents and the Environment Agency to flash flooding, using a case study of a major upland flood in North Yorkshire. Responses to flash flooding within upland communities were found to be mostly present as changes to individual behaviour and awareness. However, physical, damage reducing modifications were limited. Flash flood hazard perception was found to be linked to knowledge and experience of local flooding. Major flash flood events occurring in areas which have not experienced other recent floods are unlikely to increase perceptions or provoke responses. Although local awareness of changing weather patterns was found, supporting analyses of rainfall records, local flood risks were frequently framed in the context of river management, rather than climate change. The implementation of policy changes and responses to flash flooding by the Environment Agency will prove difficult at the local level, due to the nature of attitudes and perceptions encountered at the local level, including important differences in the perception of the flash flood hazard between local residents and representatives from nationwide organisations. Encouraging property-level modifications following flash floods, in accordance with national policies, is very difficult. In order to increase local perceptions of the flash flood hazard, the use of participatory work, focusing on long-term awareness raising and the sharing of locally held flood knowledge may be beneficial, alongside the support of existing resilience in upland communities.

This study was conducted at the mountainous catchment part of Batinah Region of the Sultanate of Oman called Al-Awabi watershed which is about 260km2 in area with about 40 Km long Wadi main channel. The study paper presents a proposed modeling approach and possible scenario analysis which uses 1D - hydraulic modeling for flood routing analysis; and the main tasks of this study work are (1) Model setup for Al-Awabi watershed area, (2) Sensitivity Analysis, and (3) Scenario Analysis on impacts of rainfall characteristics and transmission losses. The model was set for the lower 24 Km long of Al-Awabi main channel (Figure 13). Channel cross-sections were the main input to the 1D-Hydraulic Model used for the analysis of flash flood routing of the Al-Awabi watershed. As field measurements of the Wadi channel cross-sections are labor intensive and expensive activities, availability of measured channel cross-sections is barely found in this study area region of Batinah, Oman; thereby making it difficult to simulate the flood water level and discharge using MIKE 11 HD. Hence, a methodology for extracting the channel cross-sections from ASTER DEM (27mX27m) and Google Earth map were used in this study area. The performance of the model setup was assessed so as to simulate the flash flood routing analysis at different cross-sections of the modeled reach. And from this study, although there were major gap and problems in data as well as in the prevailing topography, slope and other Hydro Dynamic parameters, it was concluded that the 1D-Hydraulic Modelling utilized for flood routing analysis work can be applied for the Al-Awabi watershed. And from the simulated model results, it was observed that the model was sensitive to the type of Boundary Condition chosen and taken, channel cross sections and its roughness coefficient utilized throughout the model reach.

The height of an overflow dam must be designed low enough to prevent the reservoir water level from exceeding a flood plain during flooding conditions. Because of this constraint, much of the available water storage area is wasted and the available pressure head for power generation will be less than maximum during normal conditions. Crest control gates alleviate this problem by providing a variable spillway height. The Flash Gate Board is a passive automatic crest control gate. Its purpose is to regulate flood water while providing increased water pressure for power generation or for additional water storage for a municipality. The governing equations for the Flash Gate Board system are derived and used to formulate models of the system. Computer simulations are used to examine the system response in a variety of operating conditions. The results of these simulations are presented and discussed. The results include an investigation which developed an optimum gate height to maximize the potential of the Flash Gate Board. An experimental model was developed to verify analytical results and to provide additional insight. Conclusions from the study, recommendations for future work, and modifications for a trouble-free design are discussed.
Master of Science

This diploma work has been done as a part of the EC funded projects, MUSIC VK1- CT-2000-00058 and SmartDoc IST-2000-28137. The objective was to create an intuitive and easy to use visualization of flood forecasting data provided in the MUSIC project. This visualization is focused on the Visual User Interface and is built on small, reusable components. The visualization, FloodViewer, is small enough to ensure the possibility of distribution via the Internet, yet capable of enabling collaboration possibilities and embedment in electronic documents of the entire visualization. Thus, FloodViewer has been developed in three versions for different purposes.

Analysis and report generation (FloodViewer ) Collaborative analysis (FloodViewerNet ) Presentation and documentation (FloodViewerX).

Traditionally, hydrologists had a relatively minor role in rainfall data processing; they usually simply took data from meteorologists. However, meteorological organisations usually provide weather service over a larger area and scale (i.e. country level); the applicability of this large-scale information to urban hydrological applications is therefore questionable. This work tries to provide a local view on rainfall processing, aiming to improve the suitability (in terms of accuracy and resolution) of operational rainfall data for urban hydrological uses. This work explores advanced downscaling and adjustment techniques to address the identified issues in urban hydrology: accuracy and resolution. On the basis of a a review and the testing of state of the art techniques, the Bayesian-based adjustment technique and the newly-developed cascade-based downscaling techniques are found to be suitable tools to improve respectively the accuracy, and the resolution of operational radar (and raingauge) rainfall estimates. In addition, a combined application of these two techniques is tested; the results suggested that, although extra uncertainty may appear, this combination demonstrates a clear potential for providing accurate and high-resolution (street-scale and 5-min) rainfall estimates.

The UK weather radar network and telemetry system for the raingauges and river level gauges provided the solid physical base which produce the large amount of data in real time and a large variety of operational flood forecasting models were supplied from SW Region of the Environment Agency. Data processing, the selection of a suitable model, model calibration and parameters updating have played a more and more important role in real time forecasting and this thesis focuses on many of the key issues involved in the emerging area. Within this context, surface fitting, interpolation and cluster analysis were used for adjustment of the weather radar data and comparison between the raingauge data and radar data. As the core of the forecasting system the rainfall runoff model and river routing model were investigated in a wide-ranging manner, the key model utilised is the Transfer Function model. Potential misinterpretation of the TF model was explained by distinguishing between the "Black Box" model and the "White Box" model. The physically based Genetic Cascade Transfer Function (GCTF) model was introduced and shown to be consistent with the Gamma function and Muskingum model which were based upon the three common assumptions: linear, time-invariant and Single Input Single Output (SISO) system. The calculation formula for the moment parameters and the geometry coefficients (t-peak time, volume parameter) create the initial model and a genetic algorithm provides the basic tool to global search for the parameters. An expert system plus the genetic algorithm are combined to provide a real time updating capability. A dentritic model composed of the SISO rainfall runoff model at several tributaries and the Multi-Input Single Output (MISO) routing model in the mainstream were developed and applied to the Bristol Avon catchment. As a Weather Radar Information Processor, WRIP(II) was extended and implemented on a SPARC 10 workstation and communions at Environment Agency South West Region (Exeter) with a graphical user interface based on X/Motif.

This thesis examines Artificial Neural Networks (ANNs) for rainfall-runoff modelling. A simple ANN was first developed to predict floods in the city of Rome, located in the Tiber River basin. A rigorous comparison of the ensemble ANN and the conceptual TEVERE model were undertaken for two recent flood events in 2005 and 2008. Both models performed well but the conceptual model was better at overall hydrograph prediction while the ANN performed better for the initial part of the event at longer lead times. Further experimentation with the ANN model was then undertaken to try to improve the model performance. Additional upstream stations and rainfall inputs were added including hourly totals, effective rainfall and cumulative rainfall. Different methods of normalisation and different ANN training algorithms were also implemented along with four alternative methods for combining the ensemble ANN predictions. The results showed that the ANN was able to extrapolate to the 2008 event. Finally, Empirical Mode Decomposition was applied to the ANN to examine whether this method has value for ANN rainfall-runoff modelling. At the same time the impact of the random initialisation of the weights of the ANN was investigated for the Potomac River and Clark Fork River catchments in the USA. The EMD was shown to be a valuable tool in detecting signal properties but application to ANN rainfall-runoff modelling was dependent on the nature of the dataset. Overall uncertainty from the random initialisation of weights varied by catchment where uncertainties were shown to be very large at high stream flows. Finally, a suite of redundant and non-redundant model performance measures were applied consistently to all models. The value of applying a range of redundant and non-redundant measures, as well as benchmark-based methods was demonstrated.

This thesis presents a new approach to streamflow forecasting. The approach is based on specifying the probabilities that the next flow of a stream will occur within different ranges of values. Hence, this method is different from the time series models where point estimates are given as forecasts. With this approach flood forecasting is possible by focusing on a preselected range of streamflows. A double criteria objective function is developed to assess the model performance in flood prediction. Three case studies are examined based on data from the Salt River in Phoenix, Arizona and Bird Creek near Sperry, Oklahoma. The models presented are: a first order Markov chain (FOMC), a second order Markov chain (SOMC), and a first order Markov chain with rainfall as an exogenous input (FOMCX). Three forecasts methodologies are compared among each other and against time series models. It is shown that the SOMC is better than the FOMC while the FOMCX is better than the time series models.

This research assesses possible methods for extending the Streamflow Prediction Tool from a streamflow forecasting model to a flood extent forecasting model. This new flood extent forecasting model would allow valuable and easy to understand information be disseminated in a timely manner for flood preparation and flood response. The Height Above Nearest Drainage (HAND) method and AutoRoute method were considered for flood extent models but the HAND was the better option for its simple and quick computation as well as its viability on a global scale. Due to the importance of Digital Elevation Models (DEMs) in these flood extent models, an analysis was performed on the sensitivity and response of different DEMs with the HAND method. The HAND method with the differing DEMs was also analyzed using the Streamflow Prediction Tool for model boundary conditions against Sentinel-1 SAR generated flood extent images from August 24, 2017. The MERIT DEM performed the best in this analysis and is recommended for future research in creating a global forecasting flood extent model. The HAND method covered about 25% of the generated flood extent images and more complex flood extent models may need to be considered in areas where HAND underperforms. Finally, a proof of concept flood extent model was created and deployed as a web application for easy accessibility and distribution of flood information. Additional research to consider is flood impact based on affected population or an economic analysis, as well as optimizing model parameters for increased accuracy and performance. Additional research is also needed for HAND DEM analysis in other parts of the world.

Flood warning systems have been researched and discussed for several decades and there is a high degree of consensus in the literature that the most effective structure for a flood warning system is that of an integrated system. Experience suggests however, that few, if any, operational systems are designed in an integrated way and that few practitioners fully appreciate the benefits of integration. Through an analysis of arrangements in the Thames Basin, this research addresses this issue by identifying the necessary criteria and actions required to introduce an integrated system. The limited number of models that attempt to conceptualise flood warning systems in an integrated way have been critically examined and have found to focus too narrowly on selective integrative criteria. It is concluded that there is a need for a wider and multidimensional perspective. This study rectifies this deficiency by presenting a conceptual model that is derived from a more comprehensive assessment of the most relevant integrative factors. A two-staged process is adopted with an initial identification of a wide range of issues and variables, leading to a more focused set of factors presented under four main headings that are used to structure the substantive chapters of this thesis. These integrative factors can be conceptualised as crosscutting strands running through and drawing together the main components of a flood warning system (detection, forecasting, dissemination and response) that help ensure that these components work together collaboratively towards a common aim. Few of the integrative factors identified in this research were found in operational flood warning practices in England and Wales prior to 1996. A number of improvements were made with the establishment of the Environment Agency as the lead authority in both flood forecasting and flood warning dissemination, but a number of weakness still prevail. Through the use of case studies the plausibility of introducing a fully integrated approach to future arrangements has been tested and found to be both practical and feasible.

The movement of computational power and communications capabilities onto networks of sensors in the environment through the concept of pervasive or ubiquitous computing has initiated opportunities for the delivery of ground-based data in real-time and the development of adaptive monitoring systems. Measurements of water level taken by a network ofwireless sensors called 'FloodNet' were assimilated into a one-dimensional hydrodynamic model using an ensemble Kalman filter, to create a forecasting model. The ensemble Kalman filter led to an increase in forecast accuracy of between 50% and 70% depending on location for forecast lead times of less than 4 hours. This research then focused on methods for targeting measurements in real-time, such that the power limited but flexible resources deployed by the FloodNet project could be used optimally. Two targeting methods were developed. The first targeted measurements systematically over space and time until the forecasting model predicted that the probability of the water level exceeding a pre-defined threshold was less than 5%. The second method targeted measurements based on the expected decrease in forecasted water level error variance at a validation time and location, quickly calculated for various sets of measurements by an ensemble transform Kalman filter. Estimates of forecast error covariance from the ensemble Kalman filter and ensemble transform Kalman filter were significantly correlated, with correlations ranging between 0.979 and 0.292. Targeting measurements based on the decrease in forecast error variance was found to be more efficient than the systematic sampling method. The ensemble transform Kalman filter based targeting method was also used to estimate the 'signal variance' oftheoretical measurements at any computational node in the hydrodynamic model. Furthermore, time series data, different sensors types and measurements of floodplain stage could all be taken into account either as part of the targeting process or prior to measurement targeting.

This thesis reports the use of a time-series analysis approach to study the catchment hydrological system of the River Ribble. Rain gauge records, radar rainfall estimates and flow data are used in the analysis. The preliminary study consists of the flow forecasting at Reedyford, Pendle Water (82 km2). Flow forecasts generated from the rain gauge records are better than the radar rainfall estimates over this small catchment. However, the catchment response to rainfall is quick and no clear advantages in extending the lead-time of the forecast can be introduced by using an artificial time delayed rainfall input. A non-linear rainfall-flow relationship has been studied using the rain gauge rainfall and flow records at the River Hodder catchment (261 km2). A calibration scheme is used to identify the non-linear function of the catchment as well as the rainfall-flow system model. Although a better time-invariant system model can be identified, the non-linear rainfall-flow process cannot be fully explained by a power law function of effective rainfall. Assuming the dynamic, nonlinear system characteristics of the catchment can be reflected by a time-varying model gain parameter, relationships between the parameter and the flow, and between the parameter and the rainfall can be evaluated. These relationships have been used to improve the flow forecast during storm events. The results indicate, however, that the approach failed to improve the flow forecast near the peak flow condition. Radar data have been incorporated to forecast the flow at Jumbles Rock (1053 km2) and Samlesbury (1140 km2), River Ribble. The radar data calibrated by the Lancaster University Adaptive Radar Calibration System appears to produce better flow forecasts than the standard radar data product calibrated by the Meteorological Office. The proposed flow forecasting scheme generates better forecasts than the current system operated by the National Rivers Authority, North West Region.

Floods are considered the most damaging of natural hazards, and their frequency and damage is predicted to increase in the future. This research aims to develop an automated methodology using statistical and machine learning approaches that can perform a probabilistic monthly flood forecast. The methodology was tested to handle multiple variables as predictors. The significance of the spatial variability of the predictors was determined through model maps using 222 hydrological reference stations in Australia. Variable screening to forecast the upper 10th percentile of flow was based on the ten best scoring variables using Random Forests (RF), and flexible forecast models were developed using Generalized Additive Models (GAM). Results showed that the methodology can be used to sort through many variables (i.e. past streamflow, rainfall, Southern Oscillation Index (SOI), El Niño/ Southern Oscillation Modoki Index (EMI), and Pacific Sea Surface Temperatures (SST)) as predictors. It can be easily updated and it can vary spatially. The basic conceptual model assumed that Flow was a function of Antecedent conditions (=Lag rainfall), Flow memory (=Seasonality + autocorrelation), Climate effects (=SST indices) and random noise. Lagged flow, lagged rainfall, and lagged NIÑO 1+2 were the most important predictors using a monthly one-out cross-validation (OOCV) process and a forward cross-validation process (FCV). The Gilbert Skill Score indicated that using transformed flow data performed better than using non-transformed flow data or binary data. Model performance was affected by unsupervised variable selection in the RF model; and the employed threshold (10%) which defines a flood event. Overall skill scores based on OOCV process were in the range of 0.2-0.5 indicating reasonable forecast skill.

The hydrologic risk (and the hydro-geologic one, closely related to it) is, and has always been, a very relevant issue, due to the severe consequences that may be provoked by a flooding or by waters in general in terms of human and economic losses. Floods are natural phenomena, often catastrophic, and cannot be avoided, but their damages can be reduced if they are predicted sufficiently in advance. For this reason, the flood forecasting plays an essential role in the hydro-geological and hydrological risk prevention. Thanks to the development of sophisticated meteorological, hydrologic and hydraulic models, in recent decades the flood forecasting has made a significant progress, nonetheless, models are imperfect, which means that we are still left with a residual uncertainty on what will actually happen. In this thesis, this type of uncertainty is what will be discussed and analyzed. In operational problems, it is possible to affirm that the ultimate aim of forecasting systems is not to reproduce the river behavior, but this is only a means through which reducing the uncertainty associated to what will happen as a consequence of a precipitation event. In other words, the main objective is to assess whether or not preventive interventions should be adopted and which operational strategy may represent the best option. The main problem for a decision maker is to interpret model results and translate them into an effective intervention strategy. To make this possible, it is necessary to clearly define what is meant by uncertainty, since in the literature confusion is often made on this issue. Therefore, the first objective of this thesis is to clarify this concept, starting with a key question: should be the choice of the intervention strategy to adopt based on the evaluation of the model prediction based on its ability to represent the reality or on the evaluation of what actually will happen on the basis of the information given by the model forecast? Once the previous idea is made unambiguous, the other main concern of this work is to develope a tool that can provide an effective decision support, making possible doing objective and realistic risk evaluations. In particular, such tool should be able to provide an uncertainty assessment as accurate as possible. This means primarily three things: it must be able to correctly combine all the available deterministic forecasts, it must assess the probability distribution of the predicted quantity and it must quantify the flooding probability. Furthermore, given that the time to implement prevention strategies is often limited, the flooding probability will have to be linked to the time of occurrence. For this reason, it is necessary to quantify the flooding probability within a horizon time related to that required to implement the intervention strategy and it is also necessary to assess the probability of the flooding time.

The hydrologic risk (and the hydro-geologic one, closely related to it) is, and has always been, a very relevant issue, due to the severe consequences that may be provoked by a flooding or by waters in general in terms of human and economic losses. Floods are natural phenomena, often catastrophic, and cannot be avoided, but their damages can be reduced if they are predicted sufficiently in advance. For this reason, the flood forecasting plays an essential role in the hydro-geological and hydrological risk prevention. Thanks to the development of sophisticated meteorological, hydrologic and hydraulic models, in recent decades the flood forecasting has made a significant progress, nonetheless, models are imperfect, which means that we are still left with a residual uncertainty on what will actually happen. In this thesis, this type of uncertainty is what will be discussed and analyzed. In operational problems, it is possible to affirm that the ultimate aim of forecasting systems is not to reproduce the river behavior, but this is only a means through which reducing the uncertainty associated to what will happen as a consequence of a precipitation event. In other words, the main objective is to assess whether or not preventive interventions should be adopted and which operational strategy may represent the best option. The main problem for a decision maker is to interpret model results and translate them into an effective intervention strategy. To make this possible, it is necessary to clearly define what is meant by uncertainty, since in the literature confusion is often made on this issue. Therefore, the first objective of this thesis is to clarify this concept, starting with a key question: should be the choice of the intervention strategy to adopt based on the evaluation of the model prediction based on its ability to represent the reality or on the evaluation of what actually will happen on the basis of the information given by the model forecast? Once the previous idea is made unambiguous, the other main concern of this work is to develope a tool that can provide an effective decision support, making possible doing objective and realistic risk evaluations. In particular, such tool should be able to provide an uncertainty assessment as accurate as possible. This means primarily three things: it must be able to correctly combine all the available deterministic forecasts, it must assess the probability distribution of the predicted quantity and it must quantify the flooding probability. Furthermore, given that the time to implement prevention strategies is often limited, the flooding probability will have to be linked to the time of occurrence. For this reason, it is necessary to quantify the flooding probability within a horizon time related to that required to implement the intervention strategy and it is also necessary to assess the probability of the flooding time.

Effective prediction of localized flash flood regions for an approaching rainfall event requires an in-depth knowledge of the land surface and stream characteristics of the forecast area. Flash Flood Guidance (FFG) is currently formulated once or twice a day at the county level by River Forecast Centers (RFC) in the U.S. using modeling systems that contain coarse, generalized land and stream characteristics and hydrologic runoff techniques that often are not calibrated for the forecast region of a given National Weather Service (NWS) office. This research investigates the application of embedded geographic information systems (GIS) modeling techniques to generate a localized flash flood model for individual small watersheds at a five minute scale and tests the model using historical case storms to determine its accuracy in the FFG process. This model applies the Soil Conservation Service (SCS) curve number (CN) method and synthetic dimensionless unit hydrograph (UH), and Muskingum stream routing modeling technique to formulate flood characteristics and rapid update FFG for the study area of interest. The end result of this study is a GIS-based Flash Flood Forecasting system for ungaged small watersheds within a study area of the Blacksburg NWS forecast region. This system can then be used by forecasters to assess which watersheds are at higher risk for flooding, how much additional rainfall would be needed to initiate flooding, and when the streams of that region will overflow their banks. Results show that embedding these procedures into GIS is possible and utilizing the GIS interface can be helpful in FFG analysis, but uncertainty in CN and soil moisture can be problematic in effectively simulating the rainfall-runoff process at this greatly enhanced spatial and temporal scale.
Master of Science

The frequent occurrence of natural disasters; especially flash floods, are resulting the significant threats to many countries around the world. The truth that cannot be ignored, is that the effects of flash floods on the developing countries’ societies and economies are massive, compared with developed countries. Petra region; which is located in Jordan, is exposed to flash flood risks, which led to losses in lives, public and private properties. While the frequencies and impacts of flash floods might not be controlled easily, the need for more effective early warning systems has become extremely important. United Nations International Strategy for Disaster Reduction (UNISDR) and many researchers assumed that if an effective tsunami early warning system had been in place in the Indian Ocean region on 26 December 2004, thousands of lives would have been saved. Accordingly, Petra Region’s communities have experienced the impacts of flash floods in recent years due to absence of early warning systems and the lack of knowledge among communities about flash flood risks. These problems provide the context and demonstrate the significance of this study. Research aims to develop a responsive Flash Flood Early Warning System Guidance (FFEWSG) to enhance resilience in Petra Region. This research takes the social constructivism (interpretivism) stance in the continuum of philosophy and adopts a case study research strategy with qualitative method of research techniques. The research data collection was conducted in three phases. During the first phase, pilot semi structured interviews were conducted among people in Petra Region while the second phase focused on collecting the data from disaster affected communities, and disaster experts using semi-structured interviews. The third phase gathered information from Petra Development and Tourism Region Authority (PDTRA) documents. Data was analysed using content analysis. The research investigated flash floods in developed and developing countries; reviewed previous reports of flash flood events in Petra Region and how they affect the study area; and current early warning systems related to flooding. The research recommended a flash flood early warning system that could empower the local governmental institutions to mitigate flash flood impacts and enhance the resilience in Petra Region. It is expected that the research will add significant empirical evidence on the elements of the guidance within early warning system for flash flood, and will provide a useful tool in Petra Region for stakeholders, particularly for the government or the implementing agencies, helping to ensure the success of reducing the flash flood risks by the development of FFEWSG.