Skip to main content
European Commission logo print header

Systematic, palaeoflood and historical data for the improvement of flood risk estimation

Deliverables

The conceptual model of the SPHERE GIS was designed by P1 (CSIC) with contributions from all partners regarding their needs of the system. The spatial data model within the GIS includes the following covertures: rivers, watersheds, political borders (e.g. municipalities), topography (contour lines) and a digital elevation model (DEM), where each cell within the system is assigned an elevation. The structure of the spatial elemental data has been achieved using two models: the vectorial data model holds all the covertures, except the DEM, which has been introduced using a raster data model. The GIS implementation provides an effective way of displaying past flood alphanumerical information within a geographical scenario as well as simple analysis resulting from any particular query (e.g. magnitude and frequency of flood events during a particular time period). Following the latest concepts of software components, SPHERE-GIS is ready to incorporate other software applications on flood frequency analysis (e.g. FRESH Software, developed within this SPHERE Project) or hydraulic computation software such as HECRAS (Hydrologic Engineering Centre). The system includes two Relational Database Management Systems (RDBMS): Histo and Paleo, the tables of which were normalized until the third normal form. Both consist of a Basic Data Table which includes the main information of a flood event and complementary data tables containing detailed information related to the main data fields. The Histo Database comprises documentary flood information obtained from official and ecclesiastic historical archives, whereas Paleo database contains palaeoflood records obtained from geological-geomorphologic techniques. Spatial data structure has been deployed under two models: vectorial data model and raster data model. Vectorial GIS application requires a data model based on topological models, whereas the raster data requires a GRID approach. The vectorial cover (IntRioMuniMas) results from the graphical intersection between the rivers (River) identified by a unique code (RiverID) and the municipality (Municipality) with its unique code (MunicipalityID). This cover is linked by a unique code (HydroCode = RiverID & MunicipalityID) with the related-table (Hydro), which relates the spatial intersectioned cover with the basic table of each RDBMS, Histo and Paleo. Hydrologic data (basin, sub-basins, rivers, reservoirs, gauging stations), geographical data (administrative limits, topography, rail ways, roads, urban centres, parks), digital elevation models of different resolutions, and some orthophotos are also integrated. The SPHERE-GIS has been implemented with the package GIS ArcView 8.1, GIS tools were built using Arc Objects and Visual and the data format used for these components is the shapefile format. Custom tools extend the functionality of ArcMap to perform tasks specific to a user�s need. The GIS functions programmed include: select, buffer, counting, statistics functions, map and data display. A user-friendly interface has been developed using Visual Basic for Applications (VBA), allowing the final user (scientist or technician) to obtain all the information required. The point is focused on detecting which is the functionality that is considered useful for the final user and to implement it on the system. The system is designed to present menus through a toolbar in which to manage the graphical and alphanumerical information and easily obtain the required information and its geographical location though queries, operations and analysis. All through the application, the menus will allow the user to add some graphical information to the map, to access the alphanumerical information, to perform some queries to the databases or to perform some analyses. The data available on the SPHERE GIS comprises only the study areas included in the SPHERE Project, although the structure and the graphical coverage is designed to extend the database to other European countries. The database includes transcriptions of the documentary descriptions of the floods. Copies of original documents were included in a few examples. Scanning historical documents are limited by archive rules, and it is not very operative to overload the GIS with large documents. However, there is a good bibliographic secondary table where the user is referred to the original document.
To develop methods of treating historical data (mainly level marks) and its reliability in order to increment the amount of data for the flood frequency analysis. We have developed a double objective: - Estimate rank of peak discharges of flood events in the city of Girona (because of the Onyar river); - Estimate the threshold level of perception for catastrophic discharges in Girona (because of the Onyar river) and Lleida (because of the Segre river). The tools we have used to estimate (rank of) peak discharges are two different hydraulic models (one of each objective) 1. 2D Model: For developing the study about floods we have used the SMS (Surface water Modelling System) model. In this case the work done was: a) Generate different meshes, one for each event. It has only been analysed events with more than two water level marks registered because the error we would make in other case could be more than 50%. b) Run several simulations with different combinations of discharges (in case of more than one river for the same mesh) c) Choose the simulation that fits the best the flood and adjusts the best in all level marks. 2. 1D Model: In order to analyse perception threshold of discharges it has been used the model HEC-RAS 3.0.1 The work done was to run different simulations, one simulation for each urban morphology, for each roughness and downstream boundary conditions. In each simulation there is a cross section, which is the one that we use like perception threshold. There has been developed, two different 1D methodologies: - In order to analyse the perception threshold of discharges in Girona we had enough historical and topographic data, so the work done was to run different simulations, one simulation for each urban morphology, roughness and downstream boundary conditions. - In case of Segre river there wasn't enough information in order to develop a complete 1D analysis. We have had to considered some hypothesis and analyse carefully all historical information. Finally there was chosen one reference cross section in which one where considered different geometry (one for each historical period with historical and geometry data). As in case before, there were run different simulations, one simulation for each cross section geometry, roughness and downstream boundary conditions. It was possible to obtain the flood peak discharge and its error (using a flood peak rank) for the historical events with enough inundation flow level information using the 2D model. In this case it is very important to have good topographic and historical data. In order to estimate the evolution of the historical threshold of perception its a good tool a 1 dimensional model, but don't forget that it is necessary to decide which one could be the best reference cross section. Nowadays it is beginning 2D analysis and although in some cases the information available is not so large, taking care of the results obtained and considering an error (it depends on the information quality), the output obtained could be very useful in order to study, for example, the flood risk in areas near the rivers and the most important in this way: take precautions against floods. In the case study of Onyar river, although the methodology applied in 2D analysis is good, it has been considered that the results obtained are not so because of the lack of information; there is many uncertainty about river cross section geometry, changes in historical topography after developing buildings, roads, etc. and also the roughness. This situation makes that the error committed in the results obtained will be larger than 40% and will decrease the reliability. However, the results obtained in 1D analysis have been considered like very good results, this could be because of although the level of uncertainty is important, it is less than committed in 2D analysis, because the model calibration is easier. With 1D hydraulic modelling it is possible to estimate a range of peak discharge of threshold of perception of historical food events and also, with lots of historical information it is possible to estimate a range of peak discharges of these flood events developing 2D analysis by mean hydraulic modelling.
First approach to proxy and instrumental data for flood events prior to the existence of public meteorological services provide an interesting point of view concerning this specific natural hazard. Study of basic frequency patterns for the past 500/750 years and different degrees of severity for every flood event can provide a better knowledge about the phenomena. The analysis of flood events carried out within the SPHERE project has focused on two different topics: 1. Climatological analysis of the flood chronologies obtained into the SPHERE project, and other flood chronologies and annual precipitation series to compare the results. The analysis has been developed in order to evaluate trends, periodicities (spectral analysis), seasonal shifts (temporal evolution) and anomalous periods. The flood chronologies do not show any important trends, excepts for the extraordinary floods produced in the Maresme Basin. There is not any important trend in the oldest available annual precipitation series in Catalonia, NE Spain, (Barcelona, 1786-2002). Other long series corroborate this result. Then, the increasing of extraordinary floods in the little basins near the sea is related with the human activity. Spectral analysis shows an important problem of noise. It is possible to found some signals related with the quasi-biennial oscillation, the solar cycle and a period of 375 years. Although the number of floods does not seem to increase, the damages produced by the extraordinary and catastrophic floods have increased in the second part of the last century. More interdisciplinary analysis should be developed in order to have a regional knowledge and to discriminate the human intervention in front of the climatic patterns (Llasat et al, 2002). Finally, It is possible to find a good correlation between the temporal evolution of some Swiss glaciers (Pfister, 1980) and the extreme floods in the Spanish SPHERE basins. 2. Meteorological analysis of an anomalous period in which an important increase of flood frequency was recorded (1840-1870) according to the features of the end of the Little Ice Age. This period has been analysed flood by flood and a classification of weather types have been done. A methodology has been created to apply this analysis via synoptic reconstructions. Finally, a comparison between the synoptic weather types found in this period with these other ones for the period 1948-2000 has been done and it does not show any significant difference, although a more detailed study is necessary. Therefore, the experience in those analysis and cumulated data could arrive to water authorities and other administrations services (Civil Protection, Education, Prevention & Warning Systems, land use & urbanization authorities). Then, society could benefit of the information collected by our ancestors related to natural manifestations that exceed by large the magnitude of events recorded during more recent instrumental period (in any occasion, no more than 100 years).
An inventory of 172 historical floods has been collated on the Drac and Isère rivers in Grenoble area between 1600 to 1900. A specific scale, base on flood damages (3 stages), has been built to better describe the spatial and temporal variation of floods through the centuries in the valley. The last century data are known by continuous direct measurement of flow levels since 1870 decade. Now, the 10 biggest events since 4 centuries can be describe with details. These results could be used for the administrative flood mitigation plan. At the present time, discussions about flood hazard prevention are just beginning in Grenoble urban area and the SPHERE conclusions are timely. In a general point of view, the historical methodology developed along the program could be used in many other places in France or in Europe to know the large floods impact. In this objective, a specific knowledge tool has been created to manage all the historical data about flood. The databases allow disseminating the information for each type of end users (scientists, engineers, public). The main innovative points of the first prototype are to propose a Java interface using Applet technology. It allows data requests on Web access regarding three dimensions: time, space and event magnitude. The next stage will be to develop the prototype inside a local vulnerability communication plan: socio-economic aspects and space/time flood definition (PPR document in France for example).
Generation of a definitive chronology for floods in the three basins under study (Ter, Llobregat and Segre) with definitive data and classification, following the system developed under criteria of water levels identified and severity of impacts on human communities. Available chronologies have the following characteristics: Ter Basin: Girona (Ter/Onyar) AD1322-1987, 172 floods; Ripoll (Ter/freser) 1577-1987,18 floods; Camprodon (Ter/Ritort) 1617-1992,10 floods.Segre Basin: Seu d’Urgell (Segre/Valira) 1453-1982, 25 floods; Balaguer (Segre) 1617-1982, 21 floods; Lleida (Segre) 1306-1982, 51 floods. Llobregat Basin: El Prat (Deltaic sector) 1315-1992, 158 floods. Taking into account only the extraordinary and catastrophic floods, other scattered information for other places during the XXth century, all the floods identified could joint in flood meteorological events. The total number of flood event for each studied basin is the following: Ter Basin (1322-2000): 131 events, resulting an average of 19.3 events per century; Segre Basin (1306-1982): 70 events with an average of 10.0 events per century; Llobregat basin (1315-2000): 119 events with an average of 17.3 events per century. Other scattered floods are available for the period 11-13th centuries. All data and material was prepared following the criteria established for other partners of the project involved in the hydraulic modelling to generate the quantified reconstruction of historical events. Dossiers with all data in electronic format were prepared for Girona city and Lleida city (largest chronologies for Ter and Segre basins respectively). Llobregat basin did not produce long chronologies because severe problems of destruction of documentary heritage. Only a previous chronology is available but in the mouth of the river. This information because of the location in a deltaic zone cannot be transformed so easily to water discharges by means of hydraulic modelling. In all the case, the chronology defined allows us the climatic analysis of main frequency patterns.
The principal objective of WP3-Partner 6 (Universitaet Stuttgart) within the SPHERE Project was to provide a better understanding of the links between climate, meteorology and hydrology within the distinct river basins. In order to achieve this task, WP3 was responsible for linking atmospheric circulation to floods and thus analyse the hydrological extremes to define the flood-prone weather situations. Due to the fact that only discharge-time series in the studied areas were available, P6 suggested to link atmospheric circulation to discharge. The difficulty of this task is that a high discharge might correspond not only to a day with heavy rainfall, but also to a dry day following a flood peak. Therefore a direct link is not possible. Instead of investigating the discharges, P6 focused on the investigation of discharge increments on daily time scale. It is assumed that increases in discharge are caused by precipitation, which is a result of atmospheric circulation, and decreases in discharge are the natural reaction of the catchments to conduct excess water out of the watershed. For the identification of flood causing weather situations, the atmospheric circulation patterns have to be classified in order to assign one CP to each day, which is called classification of circulation patterns (CPs). Existing classifications were developed to explain intensive rainfall events, but not extreme floods. Therefore, WP3 proposed a new methodology to identify flood-producing CPs in mesoscale catchments. The advantage of the classification technique applied here is that it is assessed to explain the increases in discharge, thus is more appropriate for the investigation of extreme floods. The classification of the circulation patterns is a fuzzy-rule based optimisation approach. It is based on the concept of fuzzy sets, which enable dealing with imprecise statements. The expert does not identify the fuzzy rules, but instead uses an optimisation algorithm called simulated annealing. The classification consists of three steps: (1) data transformation, (2) definition of the fuzzy rules, (3) classification of observed data. Each circulation pattern is defined by a fuzzy rule. Five different possibilities are considered: a) large positive, b) medium positive, c) medium negative, d) large negative, e) arbitrary. While the first classes describe the locations of the pressure centres, the fifth indicates locations whose anomalies are irrelevant to CP. These five possible classes of anomalies appear to be adequate to describe the main CP features. The classification of the SLP map of a given day is done as follows: 1. The daily SLP map is transformed to a daily anomaly map 2. For each rule, the degree of fulfilment (DOF) is calculated. 3. The rule with the highest DOF is selected, and the corresponding index is assigned as CP of the day. In order to find the optimal rules for the description of flood producing weather situations, the performance of the classification has to be defined. Two measures have been used within this task: the occurrence of a positive increments and a wetness index. The optimisation is done over all the possible rule matrices. Due to the complexity of the problem, an algorithm based on simulated annealing was used. The performance of the classification was measured using a split sampling approach. 10 years of discharge data (1981-1990) were used in the optimisation algorithm to identify the rule matrix. Then the rule system was applied to a control period (1951-1980) to see, whether the rule system captured the main features correctly. The objective function values for this control period are compared to those of the learning period. In all cases, there were no significant differences between the statistics corresponding to the learning and the validation period. According the evaluation of the CPs, it was concluded that in the Ardèche catchments three distinct CPs cause stronger increases in discharge than normal. The occurrence of CPs does not necessarily lead to floods, but all-important floods were related to them. The wet CPs in Ardèche basin contributes 70% of all discharge increases. The wettest CP (CP01) is causing 434% increase compared to a normal day. A summary of statistics of discharge increases corresponding to wet and dry CPs is available where CP01, CP04 and CP05 are wet CPs and CP03, CP08 are dry CPs. The normalised anomalies of the SLP for the wettest CPs (CP01, CP04 and CP05) are available. A low-pressure anomaly on the Atlantic northwest from the Iberian Peninsula and a high-pressure anomaly over the eastern Mediterranean region characterise CP01. Due to the fact that CPs might cause precipitation leading to a discharge increase on the following day, the contribution of days following wet CPs was also calculated. These days contribute 10% of the total increase, so 80% of the total increase can be considered as consequence of those wet CPs.
Hydraulic modelling of historical floods of the Ardeche River: A list of historical flood levels from 1644 to the present has been drawn up and converted into discharge using hydraulic modelling. A sensitivity analysis provides error intervals on discharge estimates taking into account uncertainties on water level, roughness coefficient and channel geometry, and the impact of a non-permanent discharge or the backward effect. Two locations have been investigated, upstream and downstream the Ardeche canyon, at Vallon (1930 km2) and St Martin (2240 km2). Detailed hydraulic modelling on the downstream reach of the Ardèche (St Martin) was carried out. Roughness coefficients have been fitted using the water surface profile from the 1992 flood and field measurement of discharge. A sensitive analysis relative to the effect of downstream conditions, upstream hydrograph and topographic riverbed variations provided discharge and uncertainty estimations for historical floods. A detailed hydraulic modelling study focussed on the Vallon area, based on additional topographic survey, in order to take into account the possible by-passing of the natural arch of Vallon through a secondary branch. Some additional data from French hydrometric Office on Rhone basin (CNR) have been used to check the computed rating curve. Two discharge series are available at Vallon and St-Martin, providing a continuous series of annual maximum discharges during 1892-2000 and the selection of 15 historical floods from 1644 to 1890. A range of value has been associated to each discharge estimate, taking into account the uncertainties related to the historical water levels and the rating curves.
The palaeoflood hydrology of the Llobregat River identifies two periods of increased flood frequency: namely the Late Bronze Age (ca. 2800-2700 years BP) and the Little Ice Age (16-18th century). The magnitude of flooding during these periods was higher than that observed during the instrumental period. For example the flood radiocarbon dated to cal. AD 1516-1642 had an estimated discharge of 4860 m3/s compared to that of 2500 m3/s recorded for the 1971 flood, at a gauge station located a few km upstream of the study reach. The methodological advances applied in the research of this study focused on improvements in understanding the sedimentology of slackwater flood deposits, centred on identifying the number of flood events preserved at a site. Also the application of Cs-137 dating was developed to identify the deposits of modern floods recorded in the instrumental record so as not to count these events twice in the subsequent flood frequency analysis. A further advance was provided by the combined methodological approach of using palaeoflood and historical food evidence. In terms of the historical record of the Llobregat River, the largest event was the AD 1617 event that falls within the one-sigma age range of the largest palaeoflood dated to AD 1516-1642. The historical evidence therefore enables a more precise date to be assigned to the palaeoflood deposits. Furthermore, the site of the historical evidence is on the Llobregat delta, an area too complex for hydraulic modelling, therefore, the palaeoflood evidence provides a more reliable discharge evidence from a stable gorge reach (4860 m3/s). The palaeoflood hydrology of the Llobregat River will provide benefits in flood risk prevention in the Llobregat Basin as it provides direct evidence of large magnitude flooding in the past that exceeds that of the instrumental period. The dissemination of the results will be through peer reviewed scientific journals and reports aimed directly at the end-users within the region.
The paleoflood hydrology study was conducted at the upper end of the Ardèche River gorge. Four flood slackwater deposit sites were found at different elevations, which constitute three different thresholds for flood events. The sites are an abandoned meander; an alcove; and a cave, which recorded 6, 9, and 19 Holocene events, respectively. The sites were analyzed, and dated by radiocarbon and OSL methods. Dating the deposits enabled a correlation with the historical record. The dated deposits were correlated to the 1890, 1827, 1644 and 1522 AD historical events, which were the largest events on record. A hydraulic 1-D stationary model was conducted and the results showed the threshold discharges of 3600-4000 m3s-1, 4900-5400 m3s-1, and 5200-5700 m3s-1 for the cave, the alcove and the abandoned meander. Although the abundance of extreme flood events during 19th century, our study shows that this was not the case throughout the Holocene. Distinct flood clustering occurred during the Little Ice Age (~1500-1900 AD) and was preceded by long gap of ~1500 years with no extreme events. The conclusion from this observation is that stationary cannot be assumed for long-term statistical calculations of flood occurrences. A similar paleoflood study has been conducted on the Gardon River, during which an extreme flood occurred. The study shows that larger floods have occurred in the past, and as many as five events have been identified and dated. The results show that these five events have occurred during the Little Ice Age. This strengthens our conclusion that floods are not randomly distributed in time, but are clustered as a result of climatic influences.
The methodological guide was produced in order to describe the methodological advances developed by the SPHERE project for the use of palaeoflood and historical flood data in flood risk estimation. The particular aim of the guide was for the dissemination of the methodology to the potential end-users throughout Europe (not solely those within SPHERE). 300 copies of the book are currently in press and will be distributed to the relevant public bodies, professionals and research scientists. The guide is divided into six chapters, each focusing on a particular methodological theme of the SPHERE project methodology: - Palaeoflood data collection - Documentary flood data collection - Discharge Estimation of past flood data- Flood Frequency Analysis using non-systematic data - Floods and Climate - Living with floods: Lessons from the past. The guidelines will also be made available on the web in PDF format.
The collection and organization of daily meteorological records from the Early Instrumental Period (EIP) has a great interest for scientific research. Nowadays, the concern about climatic change and its possible impact over extreme events, point to the necessity of knowing past climatic scenarios related with those situations. Although it is possible to find some instrumental data since the XVIIth century (like data temperature in Central England), the first systematic information for a European grid does not start until the end of the XVIIIth century. Particularly, in the SPHERE project, the starting year is 1780, starting data of the daily series of Barcelona (the oldest Spanish series located until this moment). Barcelona is located at the NE of Spain, at the Llobregat basin. Other meteorological series starting during the EIP and included in the SPHERE project have been obtained into the framework of previous European projects, like ADVICE, IMPROVE and personal initiatives (data sent by A. Bardossy, P. Jones, C. Pfister). Those series are: Reykjavik, Uppsala, Stockholm, Bern, Basel, Prague, Padova, Milan, Barcelona, Madrid and Cadiz-San Fernando. Finally, the daily series of Paris has been collected and digitalized by the team of the University of Barcelona, into the SPHERE framework. Besides the previous data from the EIP, the research work carried out in the SPHERE project about the collection and analysis of meteorological records shows the information obtained for the period AD1300-2000. Such information is not homogeneous, however, for which reason it is advisable to distinguish between: a) Episodes recorded after 1950 and for which pluviometric, hydrological, synoptic and thermodynamic information is available. It is possible to distinguish between two sub-intervals: 1) 1950-1995. Meteorological data are mainly synoptic and thermodynamic, and rain-gauge data are from conventional stations. 2) 1996 up to the present. Besides the previous information, radar data, mesoscale data and precipitation and flow data from automatic networks are available. b) Episodes recorded between 1780 and 1950, for which mean daily surface pressure and total rainfall is available for various places in Europe, together with technical and handwritten reports: 1) 1780-1880. Early Instrumental Period. Scattered locations available. 2) 1880-1950. Modern Instrumental Period. Grid Databases available. c) Episodes recorded prior to 1780 for which only information from archives and handwritten papers is available. This work with proxy data has been done by collecting the continuous records of floods in municipal and private documentary sources, from various places in Catalonia. Starting from historical document sources, early instrumental data (basically, rainfall and surface pressure) and the most recent meteorological information, it is possible to analyse the meteorological patterns producing floods in NE Spain since the 14th century. Besides this, different research fields could use those values for their own research (meteorology, hydrology, climatology, public works engineering, medicine, biology). But also administrations (water authorities, civil protection) could improve their protocols of intervention if they would take into consideration the reconstruction of the severest weather events for the past 220 years.
The main purpose is to improve the flood-quantile for high return periods and the PMF estimation reliability, developing new methodologies that make use of systematic and additional historical flood and palaeoflood information (non-systematic data), taking into account the stationery or not of the flood population. Three distribution functions with upper bound have been tested. The Elíasson Transformed Extreme Value type distribution (ETEV), the Slade-type four parameter LogNormal distribution (LN4), and the Extreme Value with 4 parameters distribution (EV4). In order to compare results obtained with these upper bounded models, two additional unbounded distribution functions have been tested. The Two Component Extreme Value (TCEV), and the General Extreme Value (GEV). The parameter set for each distribution was estimated used the Maximum Likelihood Estimation (MLE) method. This method allows incorporating easily non-systematic information. The flood data can be classified in known (EX), lower bounded (LB), upper bounded (UB) and double bounded (DB). Depending on the type of data, each year will contribute to the likelihood function, where the random variable is described each year by its probability density function. The final likelihood function will be computed as the product of the likelihood of the information of each year. To test the statistical model performance the fitted CDF and the plotting positions are compared graphically. The probability plotting positions with Systematic and Censored information are calculated with the E formula. The models mentioned above, were applied at five sites in four Spanish rivers. In each of them, a flood frequency analysis was performed, and the PMF was estimated. First case in Llobregat River is at ‘Pont de Vilomara’, where palaeoflood information LB type, was available, with a first reconstructed flood dated at 798 bc. The non-systematic period starts at the middle year between the oldest floods. The other case study in the same river is at ‘Monistrol de Monserrat., where the type of paleoflood information available was DB. In the other three rivers was available only systematic and historical information. This last classified like LB type and founded in historic documents of Lleida city, for the Segre River, and in Gerona city for the Onyar case. For the Jucar River, historical information was available in a site located close to its sea mouth. Because the most important source of non-stationary in the systematic series was the dam construction, relationship between natural and reservoir affected flow peaks was obtained using hydrological simulations by a distributed hydrological model (called TETIS), assuming different historical reservoir scenarios. Concerning the non-systematic periods, its stationary was checked using the Lang’s test. From this test applied to all case studies, potentially two different scenarios arise for the flood frequency analysis. In the first one it can be assumed that in the next future the climate will be similar only to the recent past climate. In this case, the non-systematic period must start at the end of the Little Ice Age in order to be sure the same Contemporary climate is producing the data sample. In the second scenario, it is open the possibility to any past Holocene climate into the next future, without conditioning it. Consequently, all non-systematic information can be introduced into the analysis. In both scenarios the same systematic record is used. The distribution functions were fitted to all case studies data in both climate scenarios The ETEV does not fit well, in most cases. In Mediterranean rivers the EV4 model performs better. Also, its bigger sensitivity to the PMF value makes this model preferable to others. If the interest is in high return period quantile and because the actual existence of a flood upper bound called PMF, the predictive ability of an upper bounded distribution function must be exploited. The estimated PMF by the upper bounded models, in all cases, are in the same range in the Holocene climate scenario, but only the EV4 gives a sensible different value in the Contemporary climate. The 90% PMF confident limits were obtained by Monte Carlo simulation. Comparing the results for the different cases studies, we have observed a bigger amplitude of this PMF confident interval, when the information is only LB or DB, and type EX is not included. The RMSE have been estimated by Monte Carlo simulations, for the EV4, and we can conclude that with longer non-systematic periods the RMSE is reduced, but the reduction is not proportional to the period length increment.
The palaeoflood hydrology of the Ter River was focused on a study reach downstream of Gerona. Methodological advances centred on the combined use of palaeoflood slack water flood deposits located upstream of a bedrock narrowing in the river valley and the palaeohydraulic bounds approach from floodplain and terrace levels located in the wider valley floodplain upstream. The site was particularly effective for this as the river channel flowed over bedrock ensuring limited vertical channel change. The results of TL dating of pottery found in the terrace sequences provide dating evidence of flooding from the late Bronze Age through to the Little Ice Age. The palaeohydraulic bounds indicate a threshold discharge of 2600 m3/s for the upper surface of the terrace. The palaeoflood sediments preserve 3 individual flood events post-dating AD 1650 (the age of a coin found buried by the flood deposits). The estimated palaeodischarge associated with the deposits is 2900 m3/s. This is particularly important as it provides an accurate discharge estimate for the Ter River. The long flood chronology from Gerona also includes the influence of the Onyar River, with its tributary junction with the Ter located at the city. Nevertheless the Gerona historical record identifies 4 catastrophic flood events since AD 1650, including the events of 1678 and 1940. The results provide valuable long-term flood data on which to base flood risk estimation within the Ter basin. The dissemination of the results will be through peer reviewed scientific journals and reports aimed directly at the end-users within the region.

Searching for OpenAIRE data...

There was an error trying to search data from OpenAIRE

No results available