Academic literature on the topic 'Flood nowcasting'

Abstract The predictability of hydrometeorological flood events is investigated through the combined use of radar nowcasting and distributed hydrologic modeling. Nowcasting of radar-derived rainfall fields can extend the lead time for issuing flood and flash flood forecasts based on a physically based hydrologic model that explicitly accounts for spatial variations in topography, surface characteristics, and meteorological forcing. Through comparisons to discharge observations at multiple gauges (at the basin outlet and interior points), flood predictability is assessed as a function of forecast lead time, catchment scale, and rainfall spatial variability in a simulated real-time operation. The forecast experiments are carried out at temporal and spatial scales relevant for operational hydrologic forecasting. Two modes for temporal coupling of the radar nowcasting and distributed hydrologic models (interpolation and extended-lead forecasting) are proposed and evaluated for flood events within a set of nested basins in Oklahoma. Comparisons of the radar-based forecasts to persistence show the advantages of utilizing radar nowcasting for predicting near-future rainfall during flood event evolution.

Abstract. Forecasting river discharge is a very important issue for the prediction and monitoring of ground effects related to severe precipitation events. The meteorological forecast systems are unable to predict precipitation on small spatial (few km) and temporal (hourly) scales. For these reasons the issuing of reliable flood forecasts is not feasible in those regions where the basin's response to rainfall events is very fast and can generate flash floods. This problem can be tackled by using rainfall nowcasting techniques based on radar observations coupled with hydrological modeling. These procedures allow the forecasting of future streamflow with a few hours' notice. However, to account for the short-term uncertainties in the evolution of fine scale precipitation field, a probabilistic approach to rainfall nowcasting is needed. These uncertainties are then propagated from rainfall to runoff through a distributed hydrological model producing a set of equi-probable discharge scenarios to be used for the flood nowcasting with time horizons of a few hours. Such a hydrological nowcasting system is presented here and applied to some case studies. A first evaluation of its applicability in an operational context is provided and the opportunity of using the results quantitatively is discussed.

Abstract This study presents a first attempt to address the propagation of radar rainfall nowcasting errors to flood forecasts in the context of distributed hydrological simulations over a range of catchment sizes or scales. Rainfall forecasts with high spatiotemporal resolution generated from observed radar fields are used as forcing to a fully distributed hydrologic model to issue flood forecasts in a set of nested subbasins. Radar nowcasting introduces errors into the rainfall field evolution that result from spatial and temporal changes of storm features that are not captured in the forecast algorithm. The accuracy of radar rainfall and flood forecasts relative to observed radar precipitation fields and calibrated flood simulations is assessed. The study quantifies how increases in nowcasting errors with lead time result in higher flood forecast errors at the basin outlet. For small, interior basins, rainfall forecast errors can be simultaneously amplified or dampened in different flood forecast locations depending on the forecast lead time and storm characteristics. Interior differences in error propagation are shown to be effectively averaged out for larger catchment scales.

Abstract Flood forecasting in mountain basins remains a challenge given the difficulty in accurately predicting rainfall and in representing hydrologic processes in complex terrain. This study identifies flood predictability patterns in mountain areas using quantitative precipitation forecasts for two summer events from radar nowcasting and a distributed hydrologic model. The authors focus on 11 mountain watersheds in the Colorado Front Range for two warm-season convective periods in 2004 and 2006. The effects of rainfall distribution, forecast lead time, and basin area on flood forecasting skill are quantified by means of regional verification of precipitation fields and analyses of the integrated and distributed basin responses. The authors postulate that rainfall and watershed characteristics are responsible for patterns that determine flood predictability at different catchment scales. Coupled simulations reveal that the largest decrease in precipitation forecast skill occurs between 15- and 45-min lead times that coincide with rapid development and movements of convective systems. Consistent with this, flood forecasting skill decreases with nowcasting lead time, but the functional relation depends on the interactions between watershed properties and rainfall characteristics. Across the majority of the basins, flood forecasting skill is reduced noticeably for nowcasting lead times greater than 30 min. The authors identified that intermediate basin areas [~(2–20) km2] exhibit the largest flood forecast errors with the largest differences across nowcasting ensemble members. The typical size of summer convective storms is found to coincide well with these maximum errors, while basin properties dictate the shape of the scale dependency of flood predictability for different lead times.

Abstract Flash flooding represents a significant hazard to human safety and a threat to property. Simulation and prediction of floods in complex urban settings requires high-resolution precipitation estimates and distributed hydrologic modeling. The need for reliable flash flood forecasting has increased in recent years, especially in urban communities, because of the high costs associated with flood occurrences. Several storm nowcast systems use radar to provide quantitative precipitation forecasts that can potentially afford great benefits to flood warning and short-term forecasting in urban settings. In this paper, the potential benefits of high-resolution weather radar data, physically based distributed hydrologic modeling, and quantitative precipitation nowcasting for urban hydrology and flash flood prediction were demonstrated by forcing a physically based distributed hydrologic model with precipitation forecasts made by a convective storm nowcast system to predict flash floods in a small, highly urbanized catchment in Denver, Colorado. Two rainfall events on 5 and 8 July 2001 in the Harvard Gulch watershed are presented that correspond to times during which the storm nowcast system was operated. Results clearly indicate that high-resolution radar-rainfall estimates and advanced nowcasting can potentially lead to improvements in flood warning and forecasting in urban watersheds, even for short-lived events on small catchments. At lead times of 70 min before the occurrence of peak discharge, forecast accuracies of approximately 17% in peak discharge and 10 min in peak timing were achieved for a 10 km2 highly urbanized catchment.

Modeling and nowcasting of flash floods remains challenging, mainly due to uncertainty of high-resolution spatial and temporal precipitation estimates, missing discharge observations of affected catchments and limitations of commonly used hydrologic models. In this study, we present a framework for flash flood hind- and nowcasting using the partial differential equation (PDE)-based ParFlow hydrologic model forced with quantitative radar precipitation estimates and nowcasts for a small 18.5 km2 headwater catchment in Germany. In the framework, an uncalibrated probabilistic modeling approach is applied. It accounts for model input uncertainty by forcing the model with precipitation inputs from different sources, and accounts for model parameter uncertainty by perturbing two spatially uniform soil hydraulic parameters. Thus, sources of uncertainty are propagated through the model and represented in the results. To demonstrate the advantages of the proposed framework, a commonly used conceptual model was applied over the same catchment for comparison. Results show the framework to be robust, with the uncalibrated PDE-based model matching streamflow observations reasonably. The model lead time was further improved when forced with precipitation nowcasts. This study successfully demonstrates a parsimonious application of the PDE-based ParFlow model in a flash flood hindcasting and nowcasting framework, which is of interest in applications to poorly or ungauged watersheds.

Severe hydrometeorological hazards such as floods, droughts, and thunderstorms are expected to increase in the future due to climate change. Due to the significant impacts of these phenomena, it is essential to develop new and advanced early warning systems for advance preparation of the population and local authorities (civil protection, government agencies, etc.). Therefore, reliable forecasts of extreme events, with high spatial and temporal resolution and a very short time horizon are needed, due to the very fast development and localized nature of these events. In very short time-periods (up to 6 h), small-scale phenomena can be described accurately by adopting a “nowcasting” approach, providing reliable short-term forecasts and warnings. To this end, a novel nowcasting system was developed and presented in this study, combining a data assimilation system (LAPS), a large amount of observed data, including XPOL radar precipitation measurements, the Chemical Hydrological Atmospheric Ocean wave System (CHAOS), and the WRF-Hydro model. The system was evaluated on the catastrophic flash flood event that occurred in the sub-urban area of Mandra in Western Attica, Greece, on 15 November 2017. The event was one of the most catastrophic flash floods with human fatalities (24 people died) and extensive infrastructure damage. The update of the simulations with assimilated radar data improved the initial precipitation description and led to an improved simulation of the evolution of the phenomenon. Statistical evaluation and comparison with flood data from the FloodHub showed that the nowcasting system could have provided reliable early warning of the flood event 1, 2, and even to 3 h in advance, giving vital time to the local authorities to mobilize and even prevent fatalities and injuries to the local population.

Preparing proper disaster prevention measures is important for decreasing the casualties and property losses resulting from floods. One of the most efficient measures in this regard is real-time flood forecasting using quantitative precipitation forecasts (QPFs) based on either short-term radar-based extrapolation or longer-term numerical weather prediction. As both methods have individual advantages and limitations, in this study we developed a new real-time blending technique to improve the accuracy of rainfall forecasts for hydrological applications. We tested the hydrological applicability of six QPFs used for urban flood forecasting in Seoul, South Korea: the McGill Algorithm for Prediction Nowcasting by Lagrangian Extrapolation (MAPLE), KOrea NOwcasting System (KONOS), Spatial-scale Decomposition method (SCDM), Unified Model Local Data Assimilation and Prediction System (UM LDAPS), and Advanced Storm-scale Analysis and Prediction System (ASAPS), as well as our proposed blended approach based on the assumption that the error of the previously predicted rainfall is similar to that of current predicted rainfall. We used the harmony search algorithm to optimize real-time weights that would minimize the root mean square error between predicted and observed rainfall for a 1 h lead time at 10 min intervals. We tested these models using the Storm Water Management Model (SWMM) and Grid-based Inundation Analysis Model (GIAM) to estimate urban flood discharge and inundation using rainfall from the QPFs as input. Although the blended QPF did not always have the highest correlation coefficient, its accuracy varied less than that of the other QPFs. In addition, its simulated water depth in pipe and spatial extent were most similar to observed inundated areas, demonstrating the value of this new approach for short-term flood forecasting.

Abstract. Heavy precipitation events and subsequent flash floods regularly affect the Mediterranean coastal regions. In these situations, forecasting rainfall and river discharges is crucial especially up to 6 h, which is a relevant lead time for emergency services in times of crisis. The present study investigates the hydrometeorological skills of two new nowcasting systems: a numerical weather model AROME-NWC and a nowcasting system blending numerical weather prediction and extrapolation of radar estimation called PIAF. Their performance is assessed for 10 past heavy precipitation events that occurred in southeastern France. Precipitation forecasts are evaluated at a 15- min time resolution and the availability times of forecasts, based on the operational Météo-France suites, are taken into account when performing the evaluation. Rainfall observations and forecasts were first compared using a point-to-point approach. Then the evaluation was conducted from an hydrological point of view, by comparing observed and forecast precipitation over watersheds affected by floods. In general, the results led to the same conclusions for both evaluations. On the very first lead times, up to 1 h 15 min and 1 h 30 min of forecast, the performance of PIAF was higher than AROME-NWC. For longer lead times (up to 3 h) their performances were generally equivalent. An assessment of river discharges simulated with the ISBA-TOP coupled system, which is dedicated to Mediterranean flash flood simulations and driven by AROME-NWC and PIAF rainfall forecasts, was also performed on two exceptional past flash flood events. The results obtained for these two events show that using AROME-NWC or PIAF rainfall forecasts is promising for flash flood forecasting in terms of peak intensity, timing, and first wave of discharge, with an anticipation of these phenomena that can reach several hours.

Rainfall is not only one of the most natural processes on the earth, but also an important factor of flood generation. Precise rainfall nowcasting can give an effective warning before hazards occur. This paper presented an ensemble nowcasting methodology which combined two deterministic nowcasting methods: PIV_Semi-Lagrangian and PIV_Lagrangian-Persistence and the spatial correlated random error field. For the deterministic nowcasting methods, the past velocity fields were estimated by Particle Image Velocimetry (PIV) method and the advection fields were extrapolated by Semi-Lagrange and Lagrange-Persistence schemes separately, then the forecasted errors at former time step were simulated by the spatially correlated random error field and were added to the next forecasting steps. Additionally, a predicting method for rain field property was proposed and a Kalman filter algorithm was also implemented for rain field’s centre of mass prediction. The methodology was applied to 8 historical rainfall events occurred in North Rhine Westphalia (NRW), Germany by using high-resolution rainfall data acquired from C-band Essen radar belonging German Weather Service (DWD). Results showed that the promoted ensemble nowcasting methods and the rain field property predicting methods improved the forecasting accuracy obviously which confirmed their effectiveness.

Floods resulting from intense rainfall are one of the most disastrous hazards in many regions of the world since they contribute greatly to personal injury and to property damage mainly as a result of their ability to strike with little warning. The possibility to give an alert about a flooding situation at least a few hours before helps greatly to reduce the damage. Therefore, scores of flood forecasting systems have been developed during the past few years mainly at country level and regional level. Flood forecasting systems based only on traditional methods such as return period of flooding situations or extreme rainfall events have failed on most occasions to forecast flooding situations accurately because of changes on territory in recent years by extensive infrastructure development, increased frequency of extreme rainfall events over recent decades, etc. Nowadays, flood nowcasting systems or early warning systems which run on real- time precipitation data are becoming more popular as they give reliable forecasts compared to traditional flood forecasting systems. However, these kinds of systems are often limited to developed countries as they need well distributed gauging station networks or sophisticated surface-based radar systems to collect real-time precipitation data. In most of the developing countries and in some developed countries also, precipitation data from available sparse gauging stations are inadequate for developing representative aerial samples needed by such systems. As satellites are able to provide a global coverage with a continuous temporal availability, currently the possibility of using satellite-based rainfall estimates in flood nowcasting systems is being highly investigated. To contribute to the world’s requirement for flood early warning systems, ITHACA developed a global scale flood nowcasting system that runs on near-real-time satellite rainfall estimates. The system was developed in cooperation with United Nations World Food Programme (WFP), to support the preparedness phase of the WFP like humanitarian assistance agencies, mainly in less developed countries. The concept behind this early warning system is identifying critical rainfall events for each hydrological basin on the earth with past rainfall data and using them to identify floodable rainfall events with real time rainfall data. The individuation of critical rainfall events was done with a hydrological analysis using 3B42 rainfall data which is the most accurate product of Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) dataset. These critical events have been stored in a database and when a rainfall event is found in real-time which is similar or exceeds the event in the database an alert is issued for the basin area. The most accurate product of TMPA (3B42) is derived by applying bias adjustments to real time rainfall estimates using rain gauge data, thus it is available for end-users 10-15 days after each calendar month. The real time product of TMPA (3B42RT) is released approximately 9 hours after real-time and lacks of such kind of bias adjustments using rain gauge data as rain gauge data are not available in real time. Therefore, to have reliable alerts it is very important to reduce the uncertainty of 3B42RT product before using it in the early warning system. For this purpose, a statistical approach was proposed to make near real- time bias adjustments for the near real time product of TMPA (3B42RT). In this approach the relationship between the bias adjusted rainfall data product (3B42) and the real-time rainfall data product (3B42RT) was analyzed on the basis of drainage basins for the period from January 2003 to December 2007, and correction factors were developed for each basin worldwide to perform near real-time bias adjusted product estimation from the real-time rainfall data product (3B42RT). The accuracy of the product was analyzed by comparing with gauge rainfall data from Bangladesh and it was found that the uncertainty of the product is less even than the most accurate product of TMPA dataset (3B42).

Chaque année, les crues soudaines provoquées par des pluies intenses sur de petits bassins versant à réaction rapide sont la cause de pertes humaines et économiques considérables. Pour réduire ces impacts, il est nécessaire d’avoir recours à des systèmes de prévision combinant prévision météorologique et hydrologique à une échelle spatio-temporelle suffisamment fine. Etant données les nombreuses difficultés liées à cet exercice, ces systèmes doivent être capables de bien appréhender les principales sources d’incertitude susceptibles d’influencer leurs prévisions. Les incertitudes associées aux pluies observées ou prévues sont souvent considérées comme celles qui ont un impact le plus important sur les prévisions hydrologiques, surtout lorsqu'il s'agit de crues rapides et très localisées.Le but principal de cette thèse est d'étudier le potentiel d’une méthode de simulation conditionnelle géostatistique pour générer un ensemble de scénarios de pluies qui peuvent être utilisés par un système de prévision des crues soudaines. Pour ce faire, nous cherchons à générer un ensemble de champs de pluie fiable (au sens statistique), tout en exploitant au mieux les points forts des mesures souvent disponibles pour la prévision immédiate : les propriétés spatio-temporelles fournies par les données radar et les intensités pluviométriques mesurées par les pluviomètres au sol. Nous nous appuyons pour cela sur les données de pluie (radar et pluviomètres) de 17 événements intenses observés entre 2009 et 2013 dans le département du Var (sud-est de la France).La première étape de cette thèse est dédiée à la prise en compte des incertitudes sur les observations de pluie. Pour cela, le simulateur SAMPO-TBM développé à l’Irstea de Lyon est adapté pour fournir des simulations des champs de pluie alternatifs au champ de pluie radar observé, tout en respectant, à travers le conditionnement des simulations, les valeurs de pluie observées par les pluviomètres. L’évaluation de ces champs générés montre que la méthode mise en place est capable de générer des scénarios de pluie fiables et ainsi proposer une quantification des incertitudes sur les champs de pluie observés.Dans la deuxième étape de cette thèse, nous évaluons la capacité de notre méthode à être utilisée pour la prévision immédiate de pluies. Plusieurs méthodes sont testées pour la paramétrisation du simulateur et pour l’ajustement des sorties. Ces méthodes sont évaluées en considérant les principaux attributs souhaités pour une prévision d’ensemble : la précision, la fiabilité, la justesse, la discrimination et la performance globale des prévisions. La méthode la plus performante est celle estimant les paramètres du simulateur sur une fenêtre glissante de 4h, mais également en donnant un poids prépondérant à la dernière heure d’observation pour le paramètre lié à la moyenne des pluies non-nulles, associée à une correction des sorties basée sur la dernière erreur de prévision.Enfin, dans la dernière étape de cette thèse, les prévisions d’ensemble de pluie sont utilisées en entrée de la méthode AIGA d’avertissement aux crues rapides développée à Irstea Aix-en-Provence. Cette approche permet d’estimer la période de retour (en débit) de l’événement en cours sur des bassins non jaugés. L’événement du 3 au 7 novembre 2011 dans le département du Var est utilisé pour illustrer le potentiel de notre méthode. Des cartes probabilisées indiquant à différentes échéances et sur l’ensemble du réseau hydrographique du département le risque de dépassement d’une certaine période de retour sont générées. Celles-ci sont comparées à la localisation de dégâts relevés sur le terrain après l’événement démontrant un réel intérêt pour la gestion de crise en temps réel
Each year, flash floods, generated by small fast-responding catchments hit by intense rainfall, are responsible for huge human and economic losses. To mitigate these impacts, it is necessary to use forecasting systems combining meteorological and hydrological forecasts at small temporal and spatial scales. Because of the underlying difficulties, these systems have to be able to communicate the uncertainties of their forecasts. Uncertainties associated to observed or future rainfall are often seeing as those having the most important impact, in particularly in the case of flash floods localised on small areas.The main aim of this thesis is to study the potential of a geostatistical conditional simulation method to generate an ensemble of rainfall scenarios that can be used by a flash flood warning system. We seek to generate a reliable ensemble of rain fields by making the best use of the strengths of the measurements often available for nowcasting: the spatial and temporal properties of rainfall fields provided by the radar data and the rainfall intensities measured by rain gauges. In order to achieve our objectives, we use radar and rainfall data from 17 intense rainfall events observed in the Var region (south-east France) between 2009 and 2013.The first part of this thesis was devoted to taking into account the uncertainties on the observations of rainfall. For this purpose, the SAMPO-TBM generator developed at Irstea-Lyon is adapted to provide simulations of alternative rain fields to the observed radar rain field, while respecting the rainfall values observed by the rain gauges through a conditioned simulation. The evaluation of the generated fields shows that the method implemented is able to generate a reliable ensemble of rain fields and thus to propose a quantification of the uncertainties on the observed rain fields.In the second part of this thesis, the capacity of our method to be used for the nowcasting of rainfall is evaluated. Several methods are tested for the parameterization of the rainfall generator and for the adjustment of the outputs. These methods are evaluated by considering the main attributes of forecast quality, such as accuracy, reliability, precision, discrimination and overall forecast performance. The best method is the one estimating generator parameters over the last four hours, but also using only the last hour for the parameter related to the mean of the non-zero rainfall distribution, combined by the adjustment of the outputs based on the last forecast error.Finally, in the final part of this thesis, ensemble rainfall forecasts are used as inputs of the flash flood forecasting method AIGA developed at Irstea Aix-en-Provence. The AIGA method enables return period of the ongoing event to estimate at ungauged catchments. The 3th-7th November 2011 event in the Var region is used to illustrate the potential of our method. Nowcasting maps indicating, for different lead times and for the whole hydrological network of the region, the probability to exceed a given return period are produced. They are compared to the localization of observed damages collected from field surveys, illustrating a real interest for the real time crisis management

Les régions méditerranéennes sont régulièrement exposées aux événements fortement précipitants et aux crues rapides. En situation de crise, disposer d'une prévisionhydrométéorologique sur quelques heures d'échéance peut s'avérer crucial, notamment pour l'intervention des services de secours. La prévision des conséquences hydrologiques des épisodes méditerranéens de pluie intense aux échéances de la prévision immédiate, jusqu'à 6h d'échéance, est le sujet de cette thèse. Deux aspects ont été étudiés : la modélisation des débits et du ruissellement à fine échelle, et l'utilisation de prévisions immédiates de pluie, notamment des nouveaux systèmes de prévision de Météo-France, pour anticiper les crues. L'apport d'une représentation plus fine de l'occupation et de la texture des sols pour la simulation des débits des cours d'eau et du ruissellement sur des zones urbaines et péri-urbaines a d'abord été examiné. Une étude de sensibilité a été menée sur l'impact des descripteurs de sol et de la résolution spatiale (1km et 300m) avec le modèle hydrologique ISBA-TOP sur 12 cas de crues passées, incluant celui des inondations meurtrières dans la région de Cannes en 2015. Une analyse approfondie de ce cas a été conduite en exploitant les estimations de débits obtenues dans le cadre des enquêtes post-évènement du programme HyMeX et en explorant le potentiel dedonnées d'impacts pour évaluer le ruissellement simulé. La résolution spatiale du modèle s'estmontrée avoir le plus d'impact, suivie de la description de la texture du sol et enfin de ladescription de l'occupation du sol. Le potentiel des prévisions immédiates de pluie pour laprévision des crues méditerranéennes jusqu'à 6h d'échéance a ensuite été étudié. En particulier,les prévisions de pluie issues d'une version dédiée à la prévision immédiate du modèle deprévision numérique du temps AROME (AROME-PI), et d'une méthode de fusion entreextrapolation des lames d'eau radar et prévision numérique du temps (PIAF) ont été examinées.L'évaluation a été conduite en considérant les temps de mise à disposition réelle des prévisions,qui est un paramètre important à prendre en compte pour la prévision jusqu'à quelques heuresd'échéance. Ces pluies ont été évaluées d'un point de vue hydrologique, en comparant les lamesd'eau observées et prévues sur des bassins versants affectés par des crues passées, puis àl'échelle régionale sur un grand domaine Sud-Est. Les résultats sont globalement les mêmes pourles deux évaluations. PIAF est de très bonne qualité jusqu'à 1h de prévision, mais cette qualité sedégrade très rapidement, pour atteindre une qualité comparable voire inférieure à AROME-PI audelà de 1h15/1h30 de prévision. Entre 2h et 3h d'échéance, AROME-PI est de qualité supérieure ou équivalente à PIAF. Des ensembles formés par les prévisions successives de AROME-PI et dePIAF respectivement, ont aussi été étudiés. La sensibilité des ensembles à leur taille et à l'ajout d'une tolérance temporelle sur la prévision de chacun des membres a été examinée. Les résultats indiquent que plus un ensemble contient de membres, plus il est performant. Il en est de même pour les ensembles avec une tolérance temporelle de 15 ou de 30min. Une évaluation des débits aux exutoires simulés avec ISBA-TOP et MARINE forcés par les prévisions immédiates de pluie AROME-PI et PIAF, utilisées seules ou en tant qu'ensembles, a aussi été effectuée sur deux cas de crues exceptionnelles (l'Aude en 2018 et Cannes en 2015). Pour les meilleurs scenarii basés sur les prévisions AROME-PI, l'anticipation de la bonne intensité des pics de crue et de la décrue peut atteindre jusqu'à 5h, et un peu plus pour la montée. Pour ceux basés sur les prévisions PIAF, l'anticipation varie entre 20min et 4h, selon le phénomène, le bassin et le modèle hydrologique étudié
The Mediterranean regions are regularly exposed to heavy precipitating events and flash floods. Hydrometeorological forecasts up to a few hours are crucial for planning the intervention of emergency services in these situations. The prediction of the hydrological consequences of Mediterranean events of intense rainfall at the nowcasting ranges (few minutes to 6h) is the topic of this Ph. D. thesis. Two areas were studied: modelling of river flows and runoff at a fine scale, and the use of rainfall nowcasting, and particularly those from Météo-France new forecasting systems, to anticipate floods. The sensitivity to a more detailed representation of land use and texture in ISBA-TOP for simulating river flows and runoff over urban and peri-urban areas was first studied. The influence of terrain descriptors and spatial resolution (1km and 300m) has been analyzed for 12 flood events, including the major flood event in 2015 over the Cannes region. A more detailed analysis of this case was conducted using streamflow estimates at fine scale obtained from the HyMeX post-event survey and exploring the potential of impact data to evaluate simulated runoff. The results reveal that the spatial resolution has the largest impact on the hydrological simulations, larger than soil texture and land cover. Then, the potential of rainfall nowcasting for forecasting Mediterranean flash floods up to 6h was studied. The rainfall forecasts from the nowcasting suite based on the numerical weather prediction system AROME (AROME–PI), and from the nowcasting system blending numerical weather prediction and extrapolation of radar estimation (PIAF) were examined. The availability times of forecasts, based on the operational Météo-France suites, are taken into account when performing the evaluation. The evaluation of rainfall has adopted a hydrological point of view, by comparing observed and forecast rainfall over watersheds affected by past floods. A more classical evaluation comparing rainfall observation and forecast at the same location over Southeastern France has been also carried out. The results generally led to the same conclusions for both evaluations. The performance of PIAF is very good over the first hour of forecasting, but it deteriorates very quickly, to reach about the same or even a lower skill than AROME-PI beyond about 1h15/1h30 of forecasting. Between 2 and 3 hours of forecasting, AROME-PI performs better or at the same level as PIAF. Time-lagged ensembles based on AROME-PI and on PIAF forecasts respectively, were also studied. The sensitivity of the ensembles to their size and to the addition of a time tolerance on the forecast for each member was examined. The results indicate that the more members an ensemble has, the better it performs. The same applies to the ensembles with a time tolerance of 15 or 30 minutes. An assessment of river discharges simulated with ISBA-TOP and MARINE forced by AROME-PI and PIAF rainfall forecasting, used alone or as an ensemble, was also confducted on two exceptional past flash flood events (Aude in 2018 and Cannes in 2015). For the best scenarios based on AROME-PI, the anticipation of the flood peak intensity and of the instant of recession can reach up to 5h, and a little more for the first increase of flow. For those based on PIAF, the anticipation varies between 20 minutes and 4h, depending on the phenomenon, the watershed and the hydrological model studied

Floods cannot be prevented, but their devastating effects can be minimized if advance warning of the event is available. The South African Disaster Management Act (Act 57 of 2002) advocates a paradigm shift from the current "bucket and blanket brigade" response-based mind set to one where disaster prevention or mitigation are the preferred options. It is in the context of mitigating the effects of floods that the development and implementation of a reli able flood forecasting system has major significance. In the case of flash floods, a few hours lead time can afford disaster managers the opportunity to take steps which may significantly reduce loss of life and damage to property. The engineering challenges in developing and implementing such a system are numerous. In this thesis, the design and implement at ion of a flash flood forecasting system in South Africa is critically examined. The technical aspect s relating to spatio-temporal rainfall estimation and now casting are a key area in which new contributions are made. In particular, field and optical flow advection algorithms are adapted and refined to help pred ict future path s of storms; fast and pragmatic algorithms for combining rain gauge and remote sensing (rada r and satellite) estimates are re fi ned and validated; a two-dimensional adaptation of Empirical Mode Decomposition is devised to extract the temporally persistent structure embedded in rainfall fields. A second area of significant contribution relates to real-time fore cast updates, made in response to the most recent observed information. A number of techniques embedded in the rich Kalm an and adaptive filtering literature are adopted for this purpose. The work captures the current "state of play" in the South African context and hopes to provide a blueprint for future development of an essential tool for disaster management. There are a number of natural spin-offs from this work for related field s in water resources management.
Thesis (Ph.D.Eng.)-University of KwaZulu-Natal, Durban, 2007.

Flood risk assessment, as essential part of flood risk management, is a useful tool for the indication of economic damages and for the identification of the most vulnerable cities worldwide. In most cases, cities with a high concentration of people and goods are vulnerable to floods (Kubal et al., 2009). As a consequence there is a need to assess the flooding risk in all its entirety. Risk is the outcome of the interaction between a hazard phenomenon and the elements at risk within the community (e.g. people, buildings and infrastructure) that are vulnerable to such an impact (Jacks et al., 2010). In risk assessment, one has to consider the probabilities of hazardous events affecting the community and the consequent harm to the community. Probability is a concept and skill that most people have problems understanding, as many cannot handle statistical concepts or effectively factor probabilities into their decision-making (“PWS Guidelines on Communicating Forecast Uncertainty” (PWS-18), WMO/TD No. 1422). The flooding risk assessment consists on different phases: prevention, forecasting, real time monitoring and finally post-event. In the prevention phase, all the actions that allow to reduce the risk of flooding for the most sensitive areas are implemented. These actions are different and the mains are intended to: a) introduce advance flood warning and pre-planning can significantly reduce the impact of flooding; b) modify homes and business to help the population to withstand the floods; c) Construct buildings above flood levels; d) tackle climate change; e) protect wetlands and introduce trees strategically; f) restore rivers to their natural courses; g) introduce water storage areas; h) put up more flood barriers; and at least l) improve soil conditions. All these phenomena are investigated through post event flooding maps. However, because of the ongoing climate change these flooding maps based on past extreme events are obsolete. In fact, the return period (intended as an average time or an estimated average time between events such as floods) of such floods is completely different from the ones happened in the past and therefore there is a need to update these maps. The forecasting and nowcasting phases (the term forecast is referred to a very short time, generally from zero to six hours) are very important from the point of view of effective warnings and response in order to reduce the disaster risk. The forecast phase generally is based on four components: Observational Data and Monitoring Systems, Numerical Weather Prediction, Conceptual Models and Situational Awareness. For all of these components higher temporal and spatial resolution of the data (temperature, humidity pressure and wind data) are required to lead to a better weather diagnosis. These analyses allow to plan a better warning response. To date both these two phases are fundamental for the civil protection in providing information to the citizens (from e.g. media, governance institutions, etc.), understanding the hazard and, at least, in the emergency response plans. Especially the nowcasting phase, that is characterized by shorts times, requires a wide range of tools that allow to evaluate the hazard immediately. However, in some cases it is not possible to give these technologies to the civil protection and therefore, sometimes they entrust themself to the experiences of the operators or/and to their knowledge based on the experiences of past extreme events. The monitoring phase, instead, is managed by collecting information on the territory thanks to the ground-based systems linked to the local databases available or through the rescue teams. The ground stations (i.e. sensors measuring precipitation and / or water levels at relevant sites in local waterways) often are not available or capable to monitor well the extreme event. In fact, it is of extreme importance for this phase that the spatial disposition of the stations is such as to cover the whole area of interest. Because of this usually does not happen, the local rescue emergency team are employed, hence, exposing them at risk. Finally, the post-event phase deals with the using of flooding maps obtained with numerical models and / or via photographic evidence of the damages happened. These maps, as mentioned above, usually are obsolete considering the ongoing climate change and, often, do not make possible to localize all the area affected by the flood. Moreover, often these maps are implemented with hydraulic modelling using old input data (Digital Elevation Model, land cover, etc.) that do not represent the real status of the environment at the time of the extreme event of interest. Finally, usually these tools are validated considering maps based on past extreme events. As a consequence, this methodology is not useful for the reconstruction of post event maps with high accuracy, because of the land and climatic changings lead to consider new flooding areas that were safety in the past. Regarding the post event analysis with photographic evidences, the goodness of the maps depends directly from the area considered for the analysis. In fact, if the zone is in a developed city it is simple to find pictures that show the situation and that allow to rebuild a flooding map, while it is very difficult if the area is in a developing country. In flood risk assessment, satellite remote sensing constitutes a very useful tool in all above described phases. In fact, through the satellites it is possible to have, with regard to the prevention phase, a record of the floods that occurred in the past and the consequent location of the areas most exposed to the risk of flooding. For the forecast phase, information on the distribution and intensity of the rains that is about to fall can be obtained. In the monitoring phase, the satellites allow to follow the evolution of the extreme event and detect the flooding. Finally, in the post-event phase, satellites allow to a rebuilding of the flood maps and to the identification of flooded areas and to a consequently better organization of the securing of the areas most at risk. Remote sensing from satellite, therefore, for all the phases of flooding risk assessment, allows obtaining important information in the detection of soil moisture, subsidence, precipitation and the extent of flooding. Satellite technology is an excellent solution for obtaining information for flooding risk assessment for several reasons: a) allows analysis on a much larger scale compared to those made with ground instruments; b) involves less risk for rescue teams in action during floods; c) some sensors, such as radar, allows to get information in correspondence with any atmospheric configuration and during the night; finally d) in the near future satellites will be launched capable of providing ever better spatial resolutions and revisit time. At the same time, this technology also has some rather important limitations that lead us to integrate it with other existing techniques. Some of the most important limits are: a) of an instrumental nature (i.e. the radar is not able to capture flooding in the urban area); b) the spatial and temporal resolution not always able to capture the peak of flood during the extreme event; and finally c) that not all satellite missions are free and / or accessible; For this reason the satellite instrument alone is not sufficient and must necessarily be associated with other existing and compatible technologies for flooding risk assessment. Specifically, in this thesis we explore the possibility to improve flood risk assessment by the integration of hydraulic models with satellite data with reference at the two phases of post-event analysis and nowcasting. Regarding the post event, the aim was to understand if the remote sensing from satellite, considering its limitation, is a useful tool for the reconstruction of very accurate flooding maps that can perform the flooding risk assessment. Furthermore, also the possibility to integrate this technology with other available tools (social media marker and hydraulic modelling) was explored. Drawbacks arose about open source shallow water model (HEC-RAS) used in the case studies have suggested to develop a new 2D hydraulic model, that has been used in the carrying out of nowcasting phase. Concerning the nowcasting phase, the possibility to integrate data of precipitation measured by radar with real time flood forecast model was explored. A flood model, following an artificial intelligence approach, was carried out. Such model was trained by a number of simulations carried out by the 2D hydraulic model. The development of the thesis has seen three main moments: 1) Analysis of the capabilities of remote sensing in flooding risk assessment by applying it to real cases and other available tools (hydraulic modeling and social media markers); 2) Application to the post-event phases in which it was first rebuilt the extreme event of Hurricane Harvey in Houston and then an assessment of the need to build more flexible numerical models than the one used for the simulations was done; 3) Application to the analysis of flooding risk assessment in the nowcasting phase through the construction of a real time artificial intelligence model. In the first point, first of all was studied how remote sensing from satellite was used in the evaluation of flooding risk assessment for: detection of soil moisture and subsidence phenomenon; flood detection; and finally for estimating precipitation. For each of these applications the sensors and missions currently in orbit that are mostly used for these purposes have been presented. Such applications have been reported as case studies. Finally, other existing tools were investigated, such as hydraulic modeling and social media markers, useful for the flooding risk assessment. In the second point, instead, considering the applications developed in point 1, in particular that of the reconstruction of the flooding emerged in Greece (river Strymon) and in Vietnam (Quang Ngai), remote sensing techniques from satellite with hydraulic modeling and social media markers were integrated. The numerical code used in these simulations was HEC-RAS 5.0.3. This hydraulic model was considered because suggested by FEMA and because of its excellent results in terms of robustness. The peculiarity of this hydraulic model is that it exploits the subgrid approach (Casulli et al., 2008). This method allows using large meshes to perform hydraulic simulations in large areas with reasonable computational times. The Hurricane Harvey, occurred in Houston, was adopted as a test case. In this study the technique of remote sensing from satellite was integrated with hydraulic modeling and social media markers. This methodology allowed us to reconstruct, using the information obtained from ground stations, extremely accurate flood maps for all days of the extreme flood event. Subsequently, tests were carried out on the robustness of the HEC-RAS code on a portion of the urban area of Houston. These analysis, conducted in correspondence of complex geometries (such as buildings, roads, slope changes, etc.), have allowed us to understand the goodness of the approach to the subgrid in returning the results of depth and velocity of the flow to the different spatial resolutions of the computational grid. The results of the tests led to the creation of a new numerical code that was able to overcome the limitations found in HEC-RAS and, in general, those that afflict the various models present in the literature. The model created was first of all object of a bench marking test and then of an application to a real case study. Twelve extreme events occurred in correspondence of Saint Lucia Island were simulated and the results were validated with social media collected by the local population. Finally, the third and final point allowed us to explore the possibility of overcoming the main limitation in the application of hydraulic models in the nowcasting phase. In fact, the numerical codes for simulations require much longer times than those useful for the nowcasting phase (a few seconds). For this reason, an approach based on artificial intelligence has been used that allows, for a given atmospheric configurations that occurs, to obtain in a short time flooding maps based on past events similar to the extreme event that is coming. The surrogate model, in this thesis, has been trained with the simulated maps for the extreme events happened on Saint Lucia in past.

The flash flood forecasting remains one of the most difficult tasks in the operative hydrology worldwide. The torrential rainfalls bring high uncertainty included in both forecasted and measured part of the input rainfall data. The hydrological models must be capable to deal with such amount of uncertainty. The artificial intelligence methods work on the principles of adaptability and could represent a proper solution. The application of different methods, approaches, hydrological models and usage of various input data is necessary. The tool for real-time evaluation of the flash flood occurrence was assembled on the bases of the fuzzy logic. The model covers whole area of the Czech Republic and the nearest surroundings. The domain is divided into 3245 small catchments of the average size of 30 km2. Real flood episodes were used for the calibration and future flood events can be used for recalibration (principle of adaptability). The model consists of two fuzzy inference systems (FIS). The catchment predisposition for the flash flood occurrence is evaluated by the first FIS. The geomorphological characteristics and long-term meteorological statistics serve as the inputs. The second FIS evaluates real-time data. The inputs are: The predisposition for flash flood occurrence (gained from the first FIS), the rainfall intensity, the rainfall duration and the antecedent precipitation index. The meteorological radar measurement and the precipitation nowcasting serve as the precipitation data source. Various precipitation nowcasting methods are considered. The risk of the flash flood occurrence is evaluated for each small catchment every 5 or 10 minutes (the time step depends on the precipitation nowcasting method). The Fuzzy Flash Flood model is implemented in the Czech Hydrometeorological Institute (CHMI) – Brno Regional Office. The results are available for all forecasters at CHMI via web application for testing. The huge uncertainty inherent in the flash flood forecasting causes that fuzzy model outputs based on different nowcasting methods could vary significantly. The storms development is very dynamic and hydrological forecast could change a lot of every 5 minutes. That is why the fuzzy model estimates are intended to be used by experts only. The Fuzzy Flash Flood model is an alternative tool for the flash flood forecasting. It can provide the first hints of danger of flash flood occurrence within the whole territory of the Czech Republic. Its main advantage is very fast calculation and possibility of variant approach using various precipitation nowcasting inputs. However, the system produces large number of false alarms, therefore the long-term testing in operation is necessary and the warning releasing rules must be set.