Interactive comment on “ The “ Prediflood ” database of historical floods in Catalonia ( NE Iberian Peninsula ) AD 1035 – 2013 , and its potential applications in flood analysis ” by M . Barriendos

"Prediflood" is a database of historical floods that occurred in Catalonia (NE Iberian Peninsula), between the 11th century and the 21st century. More than 2700 flood cases are catalogued, and more than 1100 flood events. This database contains information acquired under modern historiographical criteria and it is, therefore, suitable for use in multidisciplinary flood analysis techniques, such as meteorological or hydraulic reconstructions.

technological resources, made impossible the systematic collection and analysis of large quantity of historical information and data. Consequently most of these works do not have minimum conditions, hence, many of these works do not meet the minimum historiographical rigor. Nevertheless, some Spanish compilators took as a reference the work of the 5 French historian Maurice Champion (Champion, 1858(Champion, -1864. In this line, two local studies stand out: one in the town of Girona (Chía, 1861) and one in town of Murcia (Hernández, 1885), this last one including an analysis of the causes and effects of floods. A remarkable synthesis of all the basins in the Iberian Peninsula was also published (Bentabol, 1900). 10 The first half of the 20th century saw a stop in flood compilations due to a movement of rejection towards historiographical determinism. However, highly destructive floods occurred in this period reignited the interest in this area of research and several local works with an increasing methodological rigor appear, such as those of the Turia River (Almela, 1957), the Ebro River (Blasco, 1959), the Segura River (Couchoud, 1965), 15 the Llobregat River (Codina, 1971) and the junction of the Ter and the Onyar rivers at Girona (Alberch et al., 1982). At the same time, analytical studies began to appear, focused either on single events (Iglésies, 1971) or on the general characteristics of floods (López Gómez, 1983). 20 In the last thirty years, the Spanish administration has made several attempts to gather historical floods information and to render it useful. More specifically, two types of organisms have led the way: basin authorities (called in Spanish "Confederaciones Hidrográficas") and civil protection authorities ("Dirección General de Protección Civil y Emergencias"). 25 On the one hand, basin authorities early began to search and use information about historical floods as a complement of instrumental data, with the objective of better assess floods' frequencies, flows, duration and behaviour. To this end, they launched 7938 Introduction several initiatives of historical floods data collection. Unfortunately, the personnel involved in those projects were civil engineers and, therefore, with a poor background on historiographical methods; they looked for information in ill-organized compilations of uneven quality, which did not allow a clear identification of the documentary sources. Furthermore, the information thus found was only used in comparing some extreme 5 historical flood to those of the instrumental period and in creating flood chronologies that lacked any methodological criteria of exhaustiveness and, hence, had a mere informative objective. Besides all this, over the years, basin authorities have been placed under different ministries due to their diverse competencies on water (irrigation, drinkable water, waste water, infrastructures, taxes), and this hampered long-term 10 projects, such as historical floods compilations. Civil protection service is relatively recent in Spain (Law 2/1985, 21 January 1985. This new concept of emergency prevention and management gives a new challenge to collection and analysis of historical information. Indeed, this service needs greats amounts of reliable information in order to perform the multidisciplinary analysis 15 required both in emergency planning (prevention, rescue, evacuation, safe and vulnerable areas) and in urban planning.

The involvement of the administration (since 1980)
Membership of the European Union also places new demands. Water Framework Directive (2000/60/EC of the European Parliament and of the Council, 23 October 2000) defined new work elements on water resources management and their 20 severe manifestations, as droughts and floods. But it was the EU Floods Directive (EU Floods Directive on the assessment and management of flood risks, 2007/60/EC of the European Parliament and of the Council, 26 November 2007), that, for the first time, specifically commanded the EU State members to assess flood hazard and risk. In Spain, this task was already underway with the mapping of flooding areas ("Sistema Introduction However, the results of the actions ordered by the EU Floods Directive are uneven in Spain. Instrumental information has been successfully catalogued and homogenised. But available historical information has not been thoroughly confirmed by systematically consult of reliable documentary sources. Indeed, in what regards historical floods information, the work has been reduced to organize and digitize the 5 data from the previous flood compilations done by the basin authorities (Catálogo Nacional de Inundaciones Históricas, 2006. This is a mere accumulation of information, but not an improvement in its quality, quantity or applicability, because the source compilations are fragmentary and they lack both flood selection criteria and references to primary documentary sources.

10
Unfortunately, this has been the usual procedure in the treatment of historical floods information until now. Therefore, although powerful software programmes support these modern compilations, their applicability in flood analysis is very limited and they are seen as mere collections of anecdotes for informational pieces into calendars or yearbooks.

Scientific approaches
Apart from the administration efforts, there have also been scientific approaches to collect historical floods information these last years. In this case, with the aim to create flood compilations of European-homologable quality.
The first doctoral thesis on the subject was by Grimalt on the island of Mallorca 20 (Grimalt, 1988), which was later published as a book (Grimalt, 1992). Since then, several research projects acquired historical information with specific criteria in order to produce consistent and reliable data series. However, these projects, which were costly and lasted from two to four years, were limited to a scarce number of chronologies in small areas; some examples are the compilations of Maresme County (Barriendos 25 and Pomés, 1993), of the Spanish Mediterranean coast (Barriendos and Martín-Vide, 1998) -CT-1999-CT- -00010, 2000-CT- -2004. Besides, the Geological Institute of Catalonia started a campaign of systematic collection of information between 2008 and 2010 in order to map natural risks, but budget limitations stopped the survey and only the Pyrenees area was completed. These compilations done with scientific purposes, although scarce and modest, 5 follow the methods of European research. This allows complex analyses such as: the improvement of climate behaviour estimations from multi-centennial flood chronologies (Llasat et al., 2005;Barriendos and Rodrigo, 2006); the study of flash floods (Llasat et al., 2003;Barrera et al., 2006;Balasch et al., 2010Balasch et al., , 2011; or the reconstruction of the peak flows and the impacts of one of the worst floods in the Iberian Peninsula, that 10 of November 1617 (Thorndycraft et al., 2006).

Characteristics of the "Prediflood" database
The "Prediflood" database contains information about historical floods occurred in Catalonia (NE Iberian Peninsula) between AD 1035 and AD 2013. Catalonia is a quite mountainous region of 32 114 km 2 on the east Mediterranean 15 coast of the Iberian Peninsula ( Fig. 1). Due to both its location and its relief, it is prone to several flood-causing weather phenomena: severe thunderstorms, long frontal rain events, and massive snow thaw. Furthermore, it is a quite populous area and it has recently undergone a period of massive construction, sometimes in flooding-prone areas, promoted by an effect of 20 speculative bubble in the building trade sector. Therefore, exposition and vulnerability have grown in the last few decades. documentary source, from which the printed sources derive: monographs, articles, reports. A flood record is reliable only when its sources are completely traceable. In addition, this allows the maximum access to generated information.
Because of these previous factors and future needs, the information organization structure has two different parts. On the one hand, all the found materials in 10 documentary and bibliographical sources are stored in their original formats. The minimal transformation and reduction permits the use of the information in successive improvements and corrections that would come up after new material gathering. On the other hand, a spreadsheet records the basic information required for all kinds of queries but, at the same time, allowing quick changes in the created categories and 15 items.
In order to do that, the "Prediflood" database information is organized in three areas: 1. Digital archive, which stores, in digital format, publications, technical reports, academic works, as well as instrumental data, and material and references available on line from public administrations, social networks and non-specialized Introduction

Location and codification system
The "Prediflood" project's research area is Catalonia as administrative unit, which is divided in two group of basins: a part of the Ebro River basin (corresponding to the Segre River basin, a tributary of the Ebro's) and the "Catalan Interior Basins", which flow directly into the Mediterranean sea. 5 The studied period is the last 500 years, which is the usual length that law requires to define flooding areas under extreme magnitude events. Nevertheless, strict time limits are unadvisable in historiographical research. Historical events information is not always complete and detailed but sometimes has cross references to previous events and, therefore, an extension of the studied period contributes to an improvement of the 10 first information.
The information has been singularized to the locations where a flood is described or documented. For the geographical location, the ACA (Catalan Water Agency) procedure has been used: 1. Basin 15 2. River 3. Town 4. Element A full identification up to level 3 is the most usual, using the official name of municipalities, the basic local administrative unit in Spain. The use of smaller units has 20 not been envisaged due to the great diversity of the descriptive level of the different flood records. It is preferable to keep this information in a raw state for eventual specific analyses when needed.
Time location is not excessively complex. The consulted documentation is usually precise with dating. Curiously, the worst indeterminations are found in bibliographical period events. In contrast, the local press provides rich information, even allowing hour resolutions, very useful in hydrological and meteorological reconstructions. The only issue that deserves attention is the possibility to record the duration of some events. In larger rivers, the precise dating of the beginning of floods and of their peak flows can be very helpful. 5 Dates are the key-element proposed to identify every flood record, because of high reliability of them. Every record will have a code composed of the complete date (YYYYMMDD) and an order number. When only one record is available for a flood event, this order number is "01". When different flood cases have the same date, order number simply shows the order in which records have entered the database.

10
After this identification of "Case Code", when a group of records are suspected, according to hydrological or meteorological evidences, to correspond to a same event, an independent code for the event is also generated (YYYY-MM):  15 The collected floods require a minimal common characterization in order to be classified. Most of the flood records are still to be completed with more precise and reliable information search, but, for the moment, the most evident traits can be used. The more common elements to an event of any time are those referring to its basic hydrological behaviour and the impacts it caused. The combination of the two criteria 20 has been used in many studies at a European level. In the case of Spain, the first proposal had three levels of classification (Barriendos and Pomés, 1993;Barriendos and Martín-Vide, 1998;Llasat et al., 2005):

Classification system by assessment of impacts
HESSD 11,2014 The "Prediflood" database of historical floods in Catalonia AD 1035-2013 The analysis of many and very diverse floods during project SPHERE led to refining of the classification system, hereby presented with the latest improvements: ERR Erroneous information : The flood never existed 0.
Overbank flood + damages : Catastrophic flood + destructions In general, the basic criteria are the occurrence of flood and whether it is an overbank 5 flood or not. Besides, there are two levels more. First, the capacity to damage nonpermanent elements (vehicles, cattle, stored goods) or light structures (catwalks or temporary wooden structures). Second, the capacity to destroy completely or partially permanent structural elements, either in an urban or in a rural environment: stone bridges, walls and other defensive elements, watermills, buildings, irrigation systems, 10 or roads and railroads. Regarding agriculture, a flood is considered destructive if it has rooted out large fields, or if it has destroyed the harvest or the productive plants (grapevines, fruit trees), removing the productive soil and leaving large fluvial deposits of any kind. In summary, catastrophic situations that will need important economic resources and several years for a full recovery, or that mean the abandonment of the HESSD 11,2014 The "Prediflood" database of historical floods in Catalonia AD 1035-2013 The classification system does not take into account human fatalities due to occurrence of this kind of impact being random in relation with the severity of the flood. To fix the evaluation of impacts on permanent structural elements is a more objective approach and more adequate for this task. The effects on population is recorded but only used in specific studies. 5 A last issue to take into account is the lack of a criterion of severity classification according to the number of affected catchments. Due to the characteristics of the Mediterranean regions, with intense torrential but not always extensive rainstorms, and with a complex orography, this territorial affection criterion would be not much representative of the magnitude of the floods. Nevertheless, the accumulation of 10 information will lead to the application of these kind of criteria in the near future, and they will be useful in identifying and classifying large floods.

Meteorological and hydrological information
Historical accounts usually have complete information about time and space location of the flood and the most relevant damages. However, information on meteorological 15 and hydrological issues is scarcer, only frequent in the most recent accounts. Because of that, it is convenient to identify and singularize the information that can be of special interest in the reconstruction of those issues.
The database have cells to confirm the presence of meteorological information, such as duration and behaviour of the precipitation, previous rain events, or any 20 other described variable, as pressure or wind speed and direction and associated phenomena. Regarding the hydrological behaviour, the data to be taken into account are: maximum water height, flood behaviour and other hydrological information such as changes in the channel, sediment accumulation, landslides, etc. of information about an unsuspectedly high number of events that have been detected. The work will be gradual and it will go beyond the initial "Prediflood" project itself, but it is the only way to acquire the historical floods information truly useful in meteorological and hydrological reconstruction of severe events. Thus, the results hereby presented are a mere starting point, which keeps open to future campaigns 10 of improvement and applied research.
As of April 2014, the "Prediflood" database has the following structure and contents: -2711 flood cases (flood records) in Catalonia, organized in 1103 flood events.
-Accumulation of textual materials: 1246 pages. The great number of events with only one documented flood case highlights the typical regime of torrential precipitations, very intense but not large, which cause serious but localized overbank floods. But it highlights, as well, an insufficient historiographical research that has not more completely defined flood events. A single-cased flood event is a stimulus to deepen the research in those area and date.

10
The greatest events, with ten or more documented cases, are optimal starting points to deepen the research. They occurred in a relatively recent period, thus their study will be more efficient. Besides, their already proved severity can be definitely characterized and bring more information for the meteorological and hydrological reconstruction. The detail study of these high-impact events can be one of the immediate applications of 15 the "Prediflood" database (see Tables A1 and A2).
Finally, the results of the "Prediflood" database can be compared to those of the compilations of the competent institutions: basin authorities and civil protection service (see Table 2 basins outstand with about 500 events, a number very much due to their large areas. Catalonia, with an average area compared to other basins, reaches 1103 flood events. The use of an objective criterion to compare the general results in Spain with those of the "Prediflood" project in Catalonia (Gaume et al., 2009), show a space and time coverage obviously greater for the whole of Spain compared to Catalonia 5 (413.8 years · surface)/10 6 km 2 and 31.4 years · surface)/10 6 km 2 , respectively). However, considering the number of events in Spain (2579 events) and Catalonia (1103 events), the density of events in relation to their space and time coverage reaches a value of 6.2 events/coverage in Spain and 35.1 events/coverage in Catalonia, which is almost six times greater.

Justification for a historiographical research
Historical floods information has specific sources, documentary and bibliographical, traditional area of research of historians. However, natural events are not, in general, appealing to this collective. Floods simply appear as mere anecdotes in local 15 historiography, and only deserved some systematic effort during the positivist period.
The present context of natural risks in their interaction with human activities makes interesting this research field. In a few years, historical climatology has shown its development capacity in scientific literature from information exclusively collected in historical documentary sources on the issue of floods (among others: Camuffo and 20 Enzi, 1996;Glaser, 1996;Pfister, 1998;Brázdil et al., 1999Brázdil et al., , 2006Wetter et al., 2011).
Situation in Spain is optimal to this kind of research thanks to the great Documentary Heritage preserved. However, historiographical research has focused in political and social issues. Until present, only 3 % of the documentary sources of specific interest to floods have been explored. In Catalonia, this percentage can be 5 % approximately. Introduction Local historiography has accessed a greater number of documentary sources, but just to generate lists of flood dates. The administrations competent with managing basins and emergency situations have used these bibliographical sources and the results have been scarce and limited despite the potentiality of the available documentation. The solution to this situation 5 can come from historiographical research itself, and the results can be as positive as those of previous European experiences.

Conclusions References
The majority of flood events in Spain are based on an insufficient exploitation of historiographical sources. Reaching a complete identification of these sources is, by itself, a study with multiple positive aspects (see Fig. 3).

Proposal of classification of information sources
The development of a study on so large and diverse historical source requires a good classification of them. The following proposal is based in their reliability levels and formats of contents: The level of the sources defines their proximity to the events. Besides, every level of 10 sources has some objective ones, with which data gathering is almost complete, and some subjective ones, which offers incomplete information.

Proposed procedures
The first analysis of the compiled floods shows the levels of the sources of information. Its exploitation can be immediate, but the classification of sources can highlight as well 15 the reliability and quality of used sources and, therefore, of the available information.
If required, the origin of the information can be investigated until arrive to the primary level sources. According to the present state of references on flood cases, the research effort should focus in finding the primary sources for most of them ensuring, at least, one 20 reliable and objective source of information. Application of this principle of traceability would suppose positive aspects: 1. Starting point would be already available information, thus not limiting its availability but consolidating and improving its reliability. 5. The accumulation of the maximum available description of impacts and quantifiable information about hydrological and meteorological information, up to an acceptable degree of exhaustiveness, would be reached. It would not be all the desiderable information but, at least, all information kept until now. 15 6. This studies, besides detecting unknown floods information, could detect as well information about other infrequent natural risks (earthquakes, landslides and rare meteorological phenomena).

Reconstruction methodology
Our multidisciplinary reconstruction of historical floods consists of three parts: 20 1. Hydraulic reconstruction, the objective of which is the calculation of the peak flow (or, when possible, the whole hydrograph) of the flood.
2. Hydrological reconstruction, the objective of which is the calculation of the hyetograph of the rain event that caused the flood. 3. Meteorological reconstruction, the objective of which is to analyse the meteorological processes before and during the rain event that caused the flood.
These three parts are linked between them, in that the results of the hydraulic reconstruction (flood's peak flow or hydrograph) are needed in the hydrological one, and that the results of the hydrological reconstruction (hyetograph) should agree with 5 the results of the meteorological one (Fig. 4). Besides, the three reconstructions occur in very different spatial scales: typically, the hydraulic reconstruction takes place along a river reach (up to a dozen km 2 area); whereas the hydrological one takes into account the whole catchment (from some dozens to thousands of km 2 ); and the meteorological reconstruction is done, depending on the meteorological phenomenon causing the event, from a local (hundreds of km 2 ) to a regional scale (1 million km 2 ). Whatever the case, all of them need historical information in order to feed the models used with the required input data and initial and boundary conditions. 15 The objective of the hydraulic reconstruction is to calculate the flood's peak flow from the maximum water height observed or flood mark, recorded in a plaque or in a written document.

Hydraulic reconstruction
This calculation could be quickly done (although with a high uncertainty) with Manning's empirical equation, which relates, in one section of the stream, the flow 20 of water with the geometrical and friction characteristics of the section, summarized in only four values: the section's area and wet perimeter, the longitudinal slope and a roughness coefficient.
However, the precision of a peak flow calculation is improved with the use of hydraulic models. Typically, these models use physically-based equations (e.g. Bernouilli, one- Simple hydraulic models (e.g. WSPRO, QUICK-2, CAUCES) can only operate in steady flow conditions (that is, no variation in time is allowed: they calculate the situation of a still instant), while others (e.g. HEC-RAS, DAMBRK, SWMM, Mike 11 HD) can calculate in unsteady flow conditions, thus obtaining more accurate results, especially in river reaches with floodplains with a great water-storing capacity. 5 Similarly, some simpler models do their calculations in one dimension only (all flow lines are perpendicular to the cross-section), while more sophisticated and accurate ones (e.g. Iber, Sobek, Mike 21 and FLO-2D) do them in two dimensions (flow lines can be oblique to the cross-section). The difference in accuracy between 1-D and 2-D models increases in winding stretches, in those in which the water velocities in the 10 channel and on the floodplain are very different, and in those were the flow is clearly not unidirectional.
However, the gain in accuracy with the use of unsteady flow conditions or 2-D models comes at a higher effort in input data acquisition and, especially, in computation time, which can even make the use of complex models impractical in historical floods 15 reconstruction, because they have to be applied iteratively. Besides, a high standard of accuracy in the calculations is not essential in reconstructing historical flows, because the input data have themselves a high degree of uncertainty. Because of this, and for the sake of homogeneity between data-rich and data-poor sites, we systematically use the 1-D hydraulic model HEC-RAS in steady flow conditions (USACE, 2008), which 20 gives accurate enough results (Balasch et al., 2010(Balasch et al., , 2011. Nevertheless, we also apply the 2-D model Iber (Iber, 2010;Ruiz-Villanueva et al., 2013) in some cases that would produce excessive inaccuracy: highly urbanized or very sinuous reaches or with large floodplains.
It must be noted that models calculate hydraulic parameters (water velocities and 25 depths) from a given peak flow, whereas we need the opposite: to calculate the peak flow from a given water height. Thus, the hydraulic model has to be applied iteratively, feeding it with tentative peak flows until the observed water height is approached enough (Fig. 5). The input data that the model needs, besides the tentative peak flow, are the stretch's geometry and friction it shows against water flow, the former given by the digital elevation model, and the latter given by Gauckler-Manning roughness coefficients, found in tables that relate friction with type of surface, sinuosity, vegetation, obstacles and cross section's contractions or expansions (Chow, 1959). Besides, a hydraulic 5 model has to be given boundary conditions, which link what happens inside the modelled river reach with what happens upstream and downstream.

HESSD
All these input data have to be adequately adapted to be as close as possible to their values at the time when the historical flood to be reconstructed took place. Therefore, old maps and documents are essential in reconstructing the channel and floodplain morphology at the time of the flood (obstacles, meanders, islands) and in hypothesizing the roughness coefficients. It must be noted that since they are acquired by estimating and hypothesizing from old documents, the input data have a high degree of uncertainty. Again, this process adds a high degree of uncertainty to the input data.
In those rare cases where measured flow data are available, they should be used 15 in calibrating the hydraulic model in that reach, that is, in estimating more accurately roughness coefficients and boundary conditions (Lang et al., 2004). As said above, hydraulic reconstruction involves a great deal of assumptions about input data; therefore, sensitivity analyses should be performed to delimit the effect of a given variation in input data on the results, that is, to estimate the error of the results.

Hydrological reconstruction
The objective of the hydrological reconstruction is the hyetograph of the rainfall that caused the flood.
A hydrological model summarizes the characteristics of the catchment that conform its hydrological response, that is, the way how it transforms rainfall into runoff and, eventually, into river flow. In other words, a model tries to quantify the hydrological processes occurring between the rain precipitation and the water exiting the catchment through its outlet. There are three main types of hydrological models: the stochastic ones, the empirical ones and the physics-based ones. Firstly, the stochastic models use large amounts of paired rainfall-flow data to calculate non-dimensional parameters that describe the catchment's hydrological response; an example is GR4J (Perrin et al., 2003). Secondly, the empirical models use simplified empirical equations and methods (such as the 5 Curve Number method; NRCS, 2007). Finally, the physics-based models use complex physics equations and need a lot of precise field measurements; an example is InHM (VanderKwaak and Loague, 2001).
Hydrological models can also be classified according to their treatment of space as well: lumped models calculate processes at the catchment or subcatchment scale (e.g. HEC-HMS), whereas distributed models do it in smaller areas and afterwards aggregate the results (e.g. r.water.fea, Vieux et al., 2004).
Due to the scarcity of data typically found outside heavily instrumented catchments and for the sake of simplicity, we use HEC-HMS, an empirical, lumped hydrological model (USACE, 2010). HEC-HMS allows the user to choose among an array of 15 different empirical methods for each one of these three hydrological processes: runoff generation, transformation of runoff into river flow, and river flow routing. For each of these processes we chose, systematically and respectively, the SCS Curve Number, the Synthetic Unit Hydrograph and the Muskingum-Cunge methods, because of their simplicity of use, their moderate requirements in input data and their being generally 20 accepted and commonly used (NRCS, 2007).
Similarly as in the hydraulic reconstruction, the calculation procedure is iterative, because the result (that is, the hyetograph) is, actually, an input datum required by the model (Fig. 6). Therefore, a tentative hyetograph must be built using the available historical information about the rain event, such as, its duration, the affected area (in 25 which subcatchments it rained and in which it did not), or indications that can lead to a rough estimation of the rainfall volume. Besides this tentative hyetograph, the model needs input data describing the catchment (or subcatchments) hydrological characteristics, such as soil type, land use, antecedent soil moisture and the stream's slope.
The result of the hydrological model (the peak flow) is then compared to the one calculated in the hydraulic reconstruction; if the two are similar enough, the tentative hyetograph is provisionally accepted. If this provisional hyetograph agrees with the 5 meteorological processes found in the meteorological reconstruction, it is definitely accepted.
The kind of inputs variables and empirical methods used have a great degree of uncertainty (Willems, 2001), all the more in the case of historical floods, because the data have to be adapted from present-day values to the estimated ones at the time of 10 the studied flood. Thus, a calibration of the model should be made whenever measured data are available. For the same reason, a sensitivity analysis should be performed once the hydrological reconstruction is done in order to estimate the real amount of uncertainty in the results.
Thus, a sensitivity analysis is performed once the hydrological reconstruction is done 15 in order to estimate the real amount of uncertainty in the results.

Meteorological reconstruction
The objective of this reconstruction is the analysis of the meteorological processes before and during the rain event that caused the flood. This analysis has two direct applications: the estimation of the antecedent soil moisture condition (an input required 20 in the hydrological reconstruction) and the classification of floods according to their meteorological causes, which can, eventually, become a useful tool in flood forecasting. The meteorological reconstruction is done in three different levels depending on the data availability or, more specifically, on the horizontal, vertical and temporal resolution of the available data, which decreases as we move back in time. Indeed, there are three different periods according to the quality of the available data, and a different level of reconstruction is applied to the floods in each one of them: Oceanic and Atmospheric Administration (NOAA) (Kalnay et al., 1996) are used to estimate the synoptic conditions of each episode: temperature, atmospheric circulation at different vertical levels, and precipitation estimates.
Additionally, the reanalysis data allow us to calculate several parameters related to the convection intensity, such as the Vertical, Cross and Total Totals indexes (Miller, 1972), the K index (George, 1960), the Humidity index (Litynska et al., 1976), the Ko index (Andersson et al., 1989), the Lifted Index LI (Galway, 1956), the Integrated Convective Available Potential Energy -ICAPE- (Mapes, 1993;Doswell and Rasmussen, 1994), the Vorticity Generating Parameter (Rassmussen et al., 1998), the difference between the LCL and LFC, the wind 20 shear between surface and 1, 3 and 6 km height, among others. In addition, the reanalysis data can be used to obtain information about wind field, moisture, and column of precipitable water, among others.
3. Events occurred since ca. 1960 (available data global models with larger resolution and mesoscale numerical simulations): finally, for more recent events, 25 version 3.3 of the WRF-ARW mesoscale model (Skamarock et al., 2008) is used at high horizontal resolution (up to approximately 1 km) to analyse synoptic, mesoscale and local conditions during the floods. The initial and boundary HESSD 11,2014 The "Prediflood" database of historical floods in Catalonia AD 1035-2013

Concluding remarks
The Prediflood database meets the internationally accepted scientific standards. It is, therefore, a repository of reliable and contrasted information that allows accurate flood 5 analysis. Actually, some of its data have already been successfully used in several flood reconstructions; at the same time, the density of the information in both the space and the time scales gives this database a great potentiality in time series analysis. The Prediflood database is in a permanent state of data incorporation. The presentday information comes from the search of about 5 % of documentary sources with 10 interesting information in Catalonia. Consequently, definition of any kind of conclusions is premature. Firsts steps are showing that this research with an interdisciplinary framework is possible in Spanish context and may be fruitful.
This effort is focused not only in quantity of flood events detected, but also in qualitative aspects, putting especial effort to increase reliability and detail of information 15 collected to be subjected to hydraulic, hydrological, meteorological reconstructions, as is made for climatic reconstruction during recent years. It produces a substantial improvement of quality and quantity of obtainable results: quality because results are more credible; quantity because spatio-temporal scales covered by reconstructions can be enlarged. Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | be easily identified and classified. Information quantified allows basic reconstruction of hydraulic and hydrological processes involved. Atmospheric conditions producing strong rainfall events and floods will be also better analyzed with enlargement of number of cases for NE Iberian Peninsula. Detection and definition of patterns of the synoptic conditions, compared between different flood 5 events will improve comprehension of atmospheric processes producing floods.
When long data series be available, after homogenizations needed by different demographic and social contexts existing for different flood events, an improved climatic variability analysis related to flood events will be possible. Application for meteorological forecasting services and flood risk managers will be strongly positive.  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Table A1. Relation of flood events selected according to severity (10 or more cases per event). Chronologically sorted.