This work is a contribution to project CS 02012-32367: “Procesos y cambios hidrogeomorfológicos en llanuras de inundación costeras mediterráneas ante la variabilidad climática y la acción humana. Una aproximación multiescalar”. Dirección General de Investigación Científica y Técnica (Spain). We thank the anonymous reviewers for their careful review and helpful comments. The English translation was done by Neil Macowan.

1Increasing concern over the effects of future climate change and the uncertainty of climatic and hydrological predictive models has encouraged many researchers to use documentary sources to determine climate variability at longer timescales. In this sense, research workers have used historical sources whereby different rivers and flood regions were related in order to detect climatic changes or anomalies since the medieval period (CAMUFFO and ENZI, 1992 and 1996; PICHARD, 1995; BARRIENDOS and RODRIGO, 2006). This research shows a broad flooding framework in terms of causality, magnitude, seasonality and frequency (BENITO et al., 2003). The shortage of instrumental records limits an adequate understanding of the relationships between climatology and flood (KNOX, 2000). Archival documents of site and weather patterns are more accurate than natural archives. The record discussed here covers a period of approximately 700 years and provides a safer basis for assessing the natural climate variability than the shorter instrumental period (PFISTER, 1992).

2Many authors have highlighted the difficulties of handling historical documentation, especially due to the discontinuity and heterogeneity of data sources and the need to validate information. Establishing comparisons between past and recent floods due to natural and anthropogenic changes in the drainage basin, channels and floodplains is sometimes difficult. Therefore, it is important to have a detailed understanding of recent flood dynamics of the sites referred to in the documentation.

3The main aim of this study is to analyse flood frequency of two Mediterranean rivers of the Iberian Peninsula (Júcar and Turia), using events of different magnitude (a range of magnitudes from moderate to extreme floods). This region constitutes a transitional zone between sub-humid and semiarid environments, and is characterized by an unusually high frequency of events, including moderate floods, being a better climate change indicator than just the occurrence of extreme or catastrophic events. Moreover, changes in seasonal flooding distribution were analysed. The frequent occurrence of flooding out of the normal season could reflect changes in weather mechanisms responsible for the hydrological responses in the Mediterranean fluvial systems.

1 - Hydrological, climatological and geomorphological setting of the Júcar and Turia floods

4The Júcar and Turia drainage basins are neighbouring catchments that drain into the Mediterranean Sea in the eastern Iberian Peninsula. Both are located between the Iberian Chain and the northern Betic Ranges, with an area of 21.600 km² and 6.393 km², respectively (Fig. 1).

5The Turia River is 280 km in length and the Júcar River is 500 km, and their headwater relief reaches elevations between 1800 and 2000 m asl. Limestone lithology and subterranean drainage, naturally regulating the discharge in the basin headwaters, are predominant. Annual medium discharge at the entrance of the floodplains (before the construction of regulation dams) was 59.7 m³/s and 10.43 m³/s for the Júcar and Turia, respectively (MASACHS ALAVEDRA, 1948). Both river regimes presented a spring rain-snow source peak, a minimum rate in summer, a low one in January due to the rainfall deficit, and a peak in June associated with convective storms.

6The low basin tributaries, of pronounced irregularity, hardly contribute to the average run-off, despite generating inundations of extraordinary magnitude. Gauging flood registers gave values higher than 400 m³/s for the Turia River with exceptional values being recorded during the catastrophic floods of the 28^th September 1949 (2300 m³/s) and the 14^th October 1957. During the last episode, there were two peaks of 2400 m³/s and 3700 m³/s (MATEU, 1988). The Júcar River recorded 15 floods higher than 400 m³/s in 50 years (1949‑2000), but two of them reached maximum estimated values of 12000 and 5000 m³/s, in 1982 and 1987 respectively. The former case was partly caused by the collapsing of the Tous dam. The right‑bank tributaries were responsible for most of the Júcar floods.

7Average annual precipitation in the Júcar and Turia basins ranged from less than 400 mm to more than 1000 mm. Moreover, maximum precipitation during torrential rainfall events in this region is centred in the Valencia Gulf. Precipitation foci were usually coastal, but sometimes affected a broad area and were associated with the most important events. The right‑margin sub‑catchments (Escalona, Sallent and Albaida rivers) of the Júcar catchment are particularly affected due to the orientation of the relief in relation to the opposing E and NE winds. For example, on 3^rd-4^th November 1987, an absolute maximum of 817 mm in 24 h was recorded (ARMENGOT, 2002; PEÑARROCHA et al., 2002).

8Floodplain typology (Fig. 2) and the geomorphical position of the flooded areas referred to in the documents must be considered in order to adequately assess flood descriptions.

9The historical evolution, hydrological and sedimentological characteristics and the dominant geomorphical processes of the Turia and Júcar floodplains are discussed by CARMONA (1990) and RUIZ (2002). The Júcar River is a meandering channel that has developed prominent alluvial ridges on its floodplain with parallel flood basins. Several tributaries (Sallent, Albaida and Magro rivers) close the topography of the plain and provoke a backwater effect during simultaneous floods. In its lower reach, the alluvial ridge buried the coastal plain, whose wetlands were filled in historical times. It could be classified as a medium- to low-energy non-cohesive floodplain according to NANSON and CROKE (1992) with average longitudinal slope values of 0.1%. The Turia River flows along a valley between Holocene terraces until near the marine mouth and has a broad channel of coarse bed load. This is a high-energy environment of non-cohesive sediments (coarse-grained confined floodplain with average longitudinal slope values of 0.2%). The floodplains of both rivers have formed a deltaic plain around the Albufera of Valencia lagoon.

2 - Historical sources of flooding of the rivers Júcar and Turia

10The compilation of palaeohydrological data (frequency, magnitude, levels, duration of floods, sedimentation, etc.) from archives was particularly challenging, both in the process of searching and extracting and the subsequent processing (WOHL and ENZEL, 1995). Historical documents reported on certain sites but not others and the terminological precision of the information was quite poor (CALVO, 1989). If a flood did not cause damage in towns or villages, it may not have been recorded in chronicles. Bearing in mind these limitations, a total of 301 flood events since the 14th century (1300‑2000) were used in this work.

11All the documentation on the Turia River refers to a geomorphical unit, the river terrace over which the city of Valencia is situated. By contrast, the information on the Júcar River refers to different geomorphical positions on the floodplain, more or less confined reaches and of high or low energy. Documentary sources of the Júcar and Turia floods were compiled by scholars or collected in local histories, cited in chronicles or diaries, etc. (ALMELA, 1957; FONTANA TARRATS, 1978; BUTZER et al., 1983, FONT TULLOT, 1988; CARMONA, 1990; MELIÓ, 1990; CARMONA, 1997; PERIS ALBENTOSA, 2001 and 2005; ALBEROLA, 2013). Moreover, original documents (Capitular Acts, Council Manuals, information on damaged irrigation infrastructure, letters soliciting tax exemption, surveys of the river, reports and design projects by engineers, etc.) were consulted in council, irrigation communities and general archives. Flood information (magnitude, frequency, seasonality, fluvial processes, weather data and anthropogenic impacts on rivers) was extracted from the documents.

12However, the heterogeneity of flood series over the studied time period must be assumed; between the 14^th and the 16^th centuries, available information was scarce, and increased at the end of this century for the Turia River. In the case of the Júcar River, sufficient data for statistical analysis has only existed since the 18^th century, although relevant indirect news sources, such as the river surveys examining the causes of the floods in the mid-16^th century, were available.

13A large amount of information was available for some floods regarding affected zones, levels, duration, contributing tributaries, etc., whereas for others the year of the flood was the only data available, even the month being unknown. Data were grouped on a series of floods (sometimes several in one year), and were eliminated when the same floods were cited erroneously by different documents in consecutive years.

14Flood events in this area were usually of short duration (1-4 days), although numerous overflows during one rainy season were occasionally cited. For example, the Júcar River overflowed its banks as many as 22 times between 1589 and 1590 and 11 times between 1671 and 1672 and between 1783 and 1784. At any time, only one monthly flooding event was considered for frequency calculations. In a previous paper, BARRIENDOS and MARTÍN VIDE (1998) selected 17 catastrophic floods for the Júcar River and 31 for the Turia River. In the present work, the process was more selective and included a greater quantity of floods (105 for the Turia River and 196 for the Júcar River), for frequency and seasonality analysis, considering that almost all the overbank flows were a response to floods or torrential rainfall at some sites in the catchment. We think that this approach may be more appropriate for the characteristics of Mediterranean hydrology.

3 - Results: Magnitude, frequency and seasonality of historical flooding of the rivers Júcar and Turia

15The Júcar and Turia rivers have recurrently experienced severe flooding episodes on their coastal floodplains since the emergence of the first cities. The 301 flood events of the rivers Júcar and Turia considered for the period 1300‑2000 were distributed as follows: 8 (14^th century), 19 (15^th century), 39 (16^th century), 45 (17^th century), 54 (18^th century), 84 (19^th century) and 52 (20^th century). With this information, the magnitude, frequency and seasonality analyses were carried out.

3.1 ‑ Flood magnitude

16An important issue when working with historical documentation is establishing the relative magnitudes of floods. Different authors have used distinct methods to define extreme floods (BRÁZDIL et al., 1999). Hydrological parameters such as water levels, flooded area, duration, references to human victims, collapse of bridges and houses, rupture of irrigation dams or walls and farming losses were considered, when cited. However, considering the diversity of environments in the floodplain areas, the magnitude of an event was assessed taking into account the geomorphology of the floodplain cited in the documentation. Bearing in mind these questions, three levels of relative flood magnitude were considered:

• Medium-magnitude floods: floods with local overflows or extensive flooding. Could affect infrastructure in the river such as dams or bridges, without seriously affecting village settlements.

• High-magnitude floods: floodplain or terrace overflow, with severe damage in fields, farm produce and irrigation infrastructure, cattle, also affecting village settlements and cities.

• Extreme-magnitude floods: high levels and/or prolonged duration, destruction of houses, walls, human life in cities, complete destruction of bridges and dams.

17By way of guidance, based on gauged floods, the approximate discharge for each flooding level were reconstructed: the medium flood was equivalent to a discharge between 500-2000 m³/s for the Júcar River, and >300 m³/s for the Turia River. The high magnitude flood could reach 2000-5000 m³/s in the Júcar River and 1000‑2000 m³/s in the Turia River. According to statistical analysis by FRANCÉS (1995), extreme magnitude floods exceeded 5000 m³/s, even reaching 13000 m³/s in the Júcar River and more than 3000 m³/s in the Turia River (maximum measured 3700 m³/s in 1957) until 6000 m³/s. All of these events were due to intense rainfall during easterly flow situations, when several tributary flood waves were superimposed and in general when these events were prolonged between 1 to 4 days. Precise secular information of damage or levels for 2-5 flood events of extreme magnitude was available for the rivers Júcar and Turia (Table 1):

18Although the Júcar and Turia basins are adjacent, simultaneous catastrophic flooding in both rivers is an exceptional event. This could be explained by the different orientation of the basin relief to the easterly flows, E-NE or E-SE, respectively. Therefore, while an extreme flood can occur in the Turia River, it was barely reflected in the Júcar River as an ordinary flood. Nonetheless, highly destructive events have coincided on both rivers, for example in autumn 1328 and in 1517 (September 27).

3.2 - Flood frequency

19In this study, all cited floods, irrespective of their magnitude, were considered for frequency analysis. Different illustrations were prepared to interpret flooding frequency through time. First, the list of annual floods is reflected separately for the rivers Turia and Júcar, three levels of magnitude being distinguished (medium, high and extreme; Fig. 3).

20The annual flood series showed clear differences between the Júcar and Turia, which in part were a consequence of the different hydrological characteristics of the basins, but should mainly be attributed to the particular cultural history of their floodplains. For example, in Valencia, in the early decades of 17^th century, city protection was necessary and the new defence walls of the Turia River were built. In this way, the documents of later periods (18^th-20^th centuries) reflected a lesser flooding incidence for the Turia River. Only a few events were able to breach the walls and affected the city centre (extreme events of 1776, 1897 and 1957).

21Even so, the periods of greater flooding frequency were well reflected for the Turia River, although with less density than for the Júcar River over the last three centuries. By contrast, the Júcar River episodes of the 16^th and 17^th centuries were only partially recorded, although diverse documents noted that the river overflowed its banks frequently in the 1540s, 1550s, 1590s, 1630s and 1670s.

22Fig. 4 shows the frequency distribution for the number of floods grouped by decade and the 30-year moving average, which helps to identify the secular trends.

23Fig. 5 shows the same, but with the sum of the floods of both rivers. Several maximums produced in the 1580s, 1610s, 1780s, 1850s, 1860s, 1870s and 1890s, with 10 or more events observed each decade.

24The tendency line shows grouping of events in three extensive periods since the end of the 16^th century. The first period includes the end of the 16^th century and the early decades of the 17^th century; the second period includes the end of the 18^th century and the first years of the 19^th century (1770-1810); and the third period corresponds to the second half of the 19^th century (1850-1900). Secondary maximums are observed in the middle of the 16^th century (1540‑1559), at the end of the 17^th century (1671‑1690) and during the mid-18^th century (1728‑1737 and 1750‑1754) and the 20^th century (1913‑1924 and 1747‑1758). Moreover, the minima are observed during the 1560s, 1600s, 1650s, 1660s, 1690s, 1720s, 1810s and 1820s (0‑1 event in each decade) since the 16^th century. The trend line shows these inflexions corresponding with the periods of very low frequency during the mid-17^th century, during the first decades of the 18^th century, and during the first half of the 19^th century. Looking at these groups of events, the following periods of high flood frequency can be considered:

• Until the mid-16th century. Fig. 5 shows a very slight increase in the flood frequencies during the 1320s, 1400s, 1470s, 1540s and 1550s. Nevertheless, catastrophic floods left abundant alluvium, which has caused the elevation of the bed and the floodplain of the Turia and Júcar channels since the 14^th century (MATEU, 1983; CARMONA and RUIZ, 2011). Similarly, the Júcar River surveys produced in 1545‑1547 and 1562 mention a rise in flood levels and the unpaid tithes due to flooding between 1553 and 1556, reflecting an increase in the frequency of flooding during the mid-16^th century.

• End of the 16^th century, early decades of the 17^th century. This period shows the greater frequency of floods on the Turia River, centred between 1580 and 1620 (21 floods in 40 years), with serious floods in 1577, 1581, 1589 and 1590. After this last event, authorities decreed the channelisation of the river at the city of Valencia (MELIÓ, 1990). The maximum number of floods in the Turia River was recorded in the 1610s, five of them in 1617. In the Júcar River, floods of great magnitude occurred in 1571 (several villages in the floodplain were depopulated) and 1632 (PERIS ALBENTOSA, 2005). Due to these floods, new surveys were carried out in the Júcar River in 1575 and 1635. An extreme year is noted in 1590 when the river overflowed its banks 22 times, the most damaging event being in July (severe summer floods in summer are extremely unusual in the Júcar basin). During this particular year, the high water level of the Albufera lagoon ruined the saltworks (SANCHIS, 2001). Likewise, reference to the tithes of the fruit crisis during the 1590s resulting from non‑seasonal floods is particularly relevant (BELL, 1980, cited in CASEY, 1979).

• End of the 17^th century, early decades of the 18^th century. After a phase of quiescence, the later decades of the 17^th century showed an increase in flood frequency, notably between 1671 and 1695, with 15 events in the rivers Turia and Júcar. During the prolonged rains of the autumn-winter of 1671-1672, the Júcar River overflowed its banks 11 times. As a result, the excessive water level of the Albufera lagoon hampered fishing and rice-growing (SANCHIS, 2001) activities. During the first half of the 18^th century, floods were irregularly distributed (19 events in 50 years). In the Turia River very few floods are cited during this period, which may be due to the efficiency of the river channelisation (“Fábrica Nova dita del Riu”; CARMONA, 1997).

• End of the 18^th century, early years of the 19^th century. Circa 1770, a new phase of high flooding frequencies began and lasted until 1808 (36 events in 39 years), the most severe years being 1776, 1779 and 1805 (ALBEROLA, 2013). The Júcar River overflowed its banks 11 times between October 1783 and 1784. Recurrent flooding of this decade triggered river avulsions and meander cutoffs on the Júcar floodplain (RUIZ, 1998 and 2002). During the decade of 1780, several village mayors declared their concern regarding the high frequency of floods and even considered the possibility of abandoning crops due to the lack of means to repair the irrigation infrastructure continuously damaged by floods. The following decades, during the first half of the 19^th century (1809-1829), showed a drastic fall in the number of floods with abnormally low frequency (3 events in 20 years).

• From the middle of the 19^th century to the beginning of the 20^th century. The highest peak in our records was recorded between 1850 and 1871 (34 events in 21 years) and the greatest flood occurred in the Júcar River in 1864, with an estimated discharge of 13000 m³/s (FRANCÉS, 1995). These floods were well described by contemporary engineers (BOSCH, 1866; GÓMEZ ORTEGA et al., 1866) and a climatologist (BODÍ, 1881). The frequency once again increased up to the end of the century until 1908 (28 events, during a period of 30 years) and between the years 1913‑1924 (14 floods in 12 years). Other severe floods took place in 1897 in the Turia River and in 1923 in the Júcar River. Since 1912, average daily discharge of the rivers Júcar and Turia from numerous gauging stations are available (MATEU et al., 2013).

• Since the middle of the 20^th century. Many defence works were carried out during the first half of the century, decreasing the impact of ordinary floods in urban areas. After some fluctuations, the flooding frequency increased between 1947 and 1958, with a total of 13 floods in 12 years and two events of severe magnitude in the Turia River (September 1949 and October 1957). The next two decades were characterized by few flood events (in part owing to the effect of construction of regulation reservoirs and the diversion of the Turia River). However, in 1982 and 1987, two catastrophic floods occurred in the Júcar River.

3.3 - Flood seasonality

25In Figs. 6, 7 and 8, the seasonal and monthly distribution of floods in each river for the total number of events in a given month is represented.

26The flooding distribution shows a clear tendency to autumnal floods, when the highest number occur (50% in the Turia River and 57% in the Júcar River). In the latter river, there are a great number of events (25%) with unknown seasonality. Almost all the extreme magnitude floods occurred in autumn, with the exception of the 17^th August 1358 event in the Turia River. The main difference between the rivers Júcar and Turia is that the latter had a greater percentage of non-autumnal floods, the summer being the second season with 21% of events. Heavy storms are very frequent during the summer in the Turia headwaters, although recent floods are very modest (MORELL, 2001). In contrast, the Júcar River had the lowest percentage of summer floods (7.7%).

27The seasonality has been analysed on the basis of irregular periods of increased flood frequency and an obvious change in the characteristic predominance of autumnal flooding during these periods has been verified. Particularly, for the 28 events of the Turia River during the period 1577‑1629, the greater number of floods (62%) was non-autumnal (Fig. 9).

28Moreover, during the maximum of 1855‑1897 there was a considerable increase in the percentage of non-autumnal floods (around 60% of the events). In the case of the Júcar River, two high frequency periods were analysed (1773‑1808 and 1850‑1899). In the first, the normal seasonal flooding distribution did not change, 2/3 of the events occurring in autumn. However, during the second half of the 19^th century there was a clear change in the seasonal trend, favouring non-autumnal floods (55% compared with 45% of the autumnal floods).

29These results led us to consider that a relationship may be expected between the increase in flood frequency and a greater probability of events out of the normal season. Fig. 10 shows the different monthly distribution of floods in the Júcar and Turia catchments.

30A greater probability of events between May and September in the Turia basin during episodes of high frequency (1571‑1631) could be noted. In this climatic region, precipitation during the warm months were related to convectivity in the Iberian mountains (Serranía de Cuenca and Sierras de Javalambre, Albarracín and Gúdar). In the Júcar catchment, monthly distribution also changed during the period between 1850 and 1875. These areas in the headwaters of the rivers Júcar and Turia were prone to the formation of thunderstorms and hailstorms from thermoconvective cells. An increase in flooding during the warm season in this Mediterranean area was probably related with some factor that reinforced atmospheric instability, such as the passing of upper troughs or depressions.

4 ‑ Discussion

31Abrupt flooding responses to global climate change have been both synchronous and diachronous across regions (KNOX, 2000). Some periods of high flood frequency in the Mediterranean during the “Little Ice Age” are coincidental with glacial advances in the Alps and the deglaciation towards the end of the 19^th century (BARRIENDOS and MARTÍN‑VIDE, 1998; GROVE, 2001). Broadly speaking, the three main climatic oscillations (1579‑1633, 1770‑1808; 1850‑1908) identified for the rivers Turia and Júcar are well correlated (Fig. 11) with other regions of the Iberian Peninsula (rivers Segura and Tajo, Catalonia and Andalusia; BARRIENDOS and MARTÍN-VIDE, 1998; BENITO et al., 2003; BARRIENDOS and RODRIGO, 2006), showing similarities and differences with other European rivers (PICHARD, 1995; GLASER et al., 2010; SCHMOCKER‑FACKEL and NAEF, 2010).

32This agreement is observed in phases of high flood frequency, but individual events are not usually simultaneous. In most cases, their behaviour is independent because Mediterranean torrential rainfall has a more localised impact compared to the regional scope of several drainage basins. Only a few large-scale events in the eastern part of the Iberian Peninsula can simultaneously affect several catchments.

33Other Spanish Atlantic catchments also show good agreement in the phases of high flooding frequencies. In the Tajo River, high flood frequencies were recorded in 1540‑1640, 1730‑1760, 1780‑1810, 1870‑1900, 1930‑1950 and 1960‑80 (BENITO et al., 2003). These authors found that although the meteorological flood-producing conditions in the Iberian Peninsula catchments draining into the Mediterranean (autumn and spring flooding) differ from those generating floods in Atlantic catchments (mainly winter conditions), there is a clear coincidence between most periods showing high flooding frequencies. These coincidences can also be found with the precipitation anomalies in southern Spain, although most floods in the Guadalquivir Basin are associated with persistent rainfall caused by winter Atlantic lows (RODRIGO et al., 1999 and 2000). The main phase of the Little Ice Age in Andalusia (southern Spain) corresponds to the period 1590‑1649, characterised by wet conditions and higher flooding frequency. Rainfall shows a maximum positive anomaly reached around 1860. The Júcar River recorded an initial interval of events around 1550, but like others rivers in the Mediterranean basin (BRÁZDIL et al., 1999), it is at the end of the 16^th century and the early decades of the 17^th century (period 1579-1633) when flooding frequency clearly increased. The peak of 1590 in the rivers Júcar and Turia is coincident with Andalusia where the worst floods occurred in the decade of 1590. Likewise, 1617 was the “year of the flood” in Catalonia (NE Spain) and there were five events in different months in the Turia River. The ‘Maldá’ anomaly (1760-1800) in the western Mediterranean studied by BARRIENDOS and LLASAT (2003) in Catalonia is also observed in the Júcar and Turia catchments (delayed to 1770‑1808).

34With regards to the seasonality during phases of high flooding frequency, a shift to non-autumnal events has been verified during 1570-1630 in the Turia River and during 1850‑1899 in the Júcar River. In contrast, BENITO et al. (2003) observed an increase in autumnal floods (Atlantic basins have a predominance of winter situations) between 1350 and 1650 and suggested a relationship between intense and persistent precipitation in the Tajo headwaters and the areas close to the Mediterranean. According to RODRIGO et al. (1999), periods of intense rainfall can be associated with hemispheric meridional circulation, whereas periods of prevailing drought are associated with hemispheric zonal circulation. For their part, PEÑARROCHA et al. (2002) consider that the crucial factor driving torrential rains in the Valencia region is the presence of easterly flows accompanied by upper troughs or depressions. These situations are more likely to occur in autumn but it is possible that, during some periods, the predominance of meridional circulation is responsible for a more frequent passing of upper cold troughs during the warm season occurs. In these conditions, the formation of supercellular storms is possible in areas like the Iberian mountains, including the watersheds of the rivers Júcar and Turia. We suggest that there may be an increase in meteorological situations favourable to thermoconvective rains during the warm season (from May to the late summer). In any case, extreme flooding occurred almost exclusively in autumn, which is related to the maximum instability of the atmosphere during this season owing to the combination of frequent passage of upper troughs and the high Mediterranean temperatures.

5 - Conclusions

35Periods of higher flooding frequency periods were noted (1579‑1598, 1610‑1633, 1671‑1695, 1770‑1808, 1850‑1908, 1913‑1924 and 1947‑1958), with notable peaks at the end of the 16^th century, 1610‑1633, 1770‑1791, 1850‑1899, 1913‑1924 and 1947‑1958. These periods are broadly synchronous with those described in the Iberian Peninsula by other authors and in other Mediterranean regions. The grouping of a great number of floods in periods of several decades alternating with periods of very low frequency implies that this occurrence is not a random phenomenon. It seems that there was a change in flooding seasonality during the high frequency periods of 1579‑1630 and 1850‑1899, increasing non-autumnal events. This is probably associated with a more frequent passing of troughs or cold pools in the upper levels of the atmosphere that favour thermoconvective thunderstorms. This in turn is related to a predominance of hemispheric meridional circulation. Almost all floods (including floods of intermediate to high magnitude) in the Valencia region are caused by heavy Mediterranean rainfall. It is therefore possible to use a range of historic floods of different magnitudes (not only extreme magnitude events) as a climate indicator.

36The greatest historical flood in the Júcar River occurred in 1864 during the highest frequency peak (1850‑1871) in our records. Furthermore, during the last centuries almost all major floods (1590, 1632, 1776, 1779, 1805, 1864, 1897, 1923, 1949 and 1957) occurred during periods of high flooding frequency. Climatic fluctuations and land-use changes partly explain these flooding peaks. Moreover, during decadal periods of high flooding frequency, runoff thresholds were reduced in such a manner as to produce several floods in a year. Coarse sediment transfer during repeated floods raised the height of the channel-bed and reduced channel capacity. Moreover, other factors such as antecedent rain and aquifer water levels probably play a crucial role, explaining the magnitude and frequency of historical floods in this transitional climatic region (semiarid to sub-humid).