Iberian extreme precipitation 1855/1856: an analysis from early instrumental observations and documentary sources

ABSTRACT

Flood events is the natural hazard that originated more damages and fatalities in the Iberian Peninsula in the last decades. While most 20th century extreme precipitation and flood episodes in Iberia have been documented, the same does not hold for most events that took place during the 19th century. This article describes the unusually high precipitation and associated impacts recorded during the 1855/1856 hydrological year. We combine newspaper reports, early instrumental precipitation series and sea level pressure (SLP ) reconstructed gridded fields. The early instrumental precipitation time series includes 11 observatories that were not previously digitized and preceded the implementation of the official meteorological observation network in Spain. We show that high values of precipitation were mostly recorded during the months of September, October 1855 and January 1856, with most of the flooding and damages occurring in the last month. The use of daily circulation weather types and monthly differences of SLP are particularly useful to explain the heavy precipitation in October and January, which were clearly associated with unusual high frequencies of wet weather types. However, SLP patterns cannot explain the September records, which could be associated to upper cold air intrusions in the Iberian Peninsula.

1 Introduction

Iberian precipitation is characterized by intense floods and droughts, whose intensity and frequency could be intensifying due to increasing emissions of greenhouse gases (Sousa et al. , 2011; Hoerling et al. , 2012). Both types of events are highly variable in time and space and have severe societal impacts. According to the Spanish public reinsurance company responsible of compensations for natural disasters (Consorcio de Compensación de Seguros, 2012) floods were responsible of 67.9% of the total compensations during the period 1971–2011. The average economic lost related with floods events represents annually 0.1% of the Spanish Gross Domestic Product. The greater economic compensations were in 1983 (683.5 M€ affecting mainly Northern Spain), 1987 (332.9€, with the main damages in the SE), 1997 (294.0 M€ with, again, the SE mostly affected) and 2010 (328.0 M€ mostly for Andalusia). In Portugal, floods were the deadliest natural disaster during the 20th century, followed by earthquakes, with each death associated to earthquakes being matched by seven fatalities due to floods (Ramos and Reis, 2001). The most damaging events were those that occurred in the Lisbon area in November 1967 (more than 500 deaths) and 16 years later, on November 1983, with circa 10 fatalities, more than 1800 people dislodged people and huge financial losses due to electric power blackouts and road and rail links blocked (Liberato et al. , 2013). Both episodes were aggravated by urban flash flooding and landslides events around the Lisbon area (Zêzere et al. , 2005).

Owing to their high societal impacts flood occurrence is frequently recorded in historical documents and there is a vast literature derived from the abstraction of documentary sources from different archives which helps to infer their frequency and intensity during the pre‐instrumental era (Barriendos, 1997; Benito et al. , 2003; Llasat et al. , 2005; Barriendos and Rodrigo, 2006; Dominguez‐Castro et al. , 2014).

The interpretation of the associated dynamics is more problematic due to the lack of dense observations. However, the number of Iberian stations with long time series (some starting as early as late 18th century, and a larger number debuting in the second half of the 19th century) have increased considerably as a result of projects focused on digitizing data (Barriendos et al. , 2002; Brunet and Jones, 2011; Alcoforado et al. , 2012; Dominguez‐Castro et al. , 2013). Thus, despite presenting a less dense net than some other European regions, both Iberian countries have enlarged considerably the number of long‐term time series of monthly (Trigo et al. , 2009) and daily precipitation and temperature (Brunet et al. , 2006; Fragoso et al. , 2010; Gallego et al. , 2011; Cortesi et al. , 2013). Gridded datases such as the 20th century reanalysis (Compo et al. , 2011) or observational sea level pressure (SLP) such as Ansell et al (2006) can be useful to analyse more objectively the atmospheric circulation associated with several extreme phenomena in the late 19th century.

In this article, we provide new evidences on the extreme precipitation recorded between September 1855 and August 1856 that resulted in severe floods and damages. This hydrological year is peculiar, with several prolonged intense precipitation events occurring during distinct times of the year, and being sufficiently relevant to impinge a strong fingerprint. The strong precipitation and associated floods were concentrated in autumn (September and October) and winter (January) characterized by an affected area covering a large fraction of Iberia. The combination of documentary sources and early instrumental data allow us to produce a comprehensive analysis of the entire hydrological year, including some dynamical features.

2 Dataset

The characterization of this unusual year (including the distinct rainfall spells) has been done considering a range of different data sources.

2.1 Documentary sources

Many newspapers have been abstracted looking for weather information in Iberia during the hydrological year 1855/1856. These include El Balear, El Clamor Público, Diario Oficial de Avisos de Madrid, La Época, La España, La Iberia, La Ilustración, Mercurio de Madrid , Crónica Científica y Literaria, La Gaceta de Madrid and Ilustracao Luso‐Brazileira. The newspapers are a primary source of great quality (Gallego et al. , 2008) since they usually provide eyewitness descriptions, in many cases with subdaily information and, additionally, relevant information on the causes that provoked the weather events.

We have also used a technical report from Spanish National Commission for floods emergencies (Comisión Nacional de Protección Civil‐Comisión Técnica de Inundaciones, 1988, this technical report can be found on http://www.proteccioncivil.org/inundaciones‐ctei), which analysed the more relevant historical floods in the main 10 hydrological basins of Spain. This is a secondary source carefully assembled and supplements the information provided by the newspapers.

2.2 Instrumental data

Instrumental measurements from September 1855 to August 1856 in 14 Spanish locations have been rescued from (Dirección General de Obras Públicas (DGOP), 1856). The monthly record is continuous during the whole period. The original table shows the symbol ‘>>’ instead a value in some months January (Granada), May (Seville), June (Valladolid), July (Malaga, Valladolid, Seville and Madrid) and August (Malaga, Barcelona, Valladolid Seville and Madrid). We have considered this symbol as 0 mm, due to its concentration during the summer months, which in Iberia are usually very dry.

It should be stressed that these observations are prior to the official network in Spain that started in 1858 with the publication of the first Statistical Annals. However, it is quite clear that a proto‐official network had been started around 1850. The Royal Order of 6 October 1850 provided the legal framework to develop a meteorological network in Spain. This made that some meteorological stations started to develop, under the direction of Manuel Rico Sinobas. These time series add to those already recovered for Cadiz (Barriendos et al. , 2002), Barcelona and Madrid (Barriendos et al. , 1997; Brunet et al. , 2006). Unfortunately, the documents of the DGOP only cover 1855/1856 and cannot be expanded longer in time. Thus, it is worth stressing that 11 of the 14 Spanish stations used in this work have not been identified or used previously and had been rescued and digitized specifically for this paper.

Figure 1 shows the locations of the 15 stations considered in this work (14 Spanish and 1 Portuguese). We acknowledge that their spatial distribution is not homogeneous and gives us hardly any information over some regions e.g. the southern Inner Plateau. Unfortunately, very little is known about observers or the methodology of observation due to the lack of metadata, a general problem in Spain. But focus on the precipitation, probably the Spanish observers followed the recommendations that Manuel Rico Sinobas made some year earlier (Rico Sinobas, 1854). According to them the pluviometer ‘should be of easy construction, with a diameter between 9 and 12 inches, made of zinc or copper by any of our artisans and with scales divided in Spanish lines or the decimal system’. To compare with long‐term rainfall series we have selected 15 stations in the same places of the historical ones, but currently active. Figure 1 shows the period covered by each series. All become from the AEMET data base, except Lisbon, where the data become from the Infante D. Luiz Observatory (Fragoso et al. , 2010; Cortesi et al. , 2013). In addition, the Spanish selected long‐term stations were tested for quality control and temporal homogeneity by the AEMET before being handled for this work. Regarding the Lisbon precipitation time series, it was previously examined in detail for temporal homogeneity by Kutiel and Trigo (2014) and no significant change‐points were found.

Figure 1

Location and timespan of meteorological series, and other geographical locations cited in the text.

2.3 Sea level pressure dataset and circulation weather types

We have used the longest gridded dataset of daily mean SLP available for the Atlantic‐European region. This dataset covers the period 1850–2003 on a 5° latitude by longitude grid and was compiled by Ansell et al. (2006) in the framework of the EMULATE project. This product was obtained using 86 continental and island stations distributed over the region 25°–70°N, 70°W–50°E blended with marine data from the International Comprehensive Ocean–Atmosphere Data Set (ICOADS). The EMSLP fields for 1850–1880 are based purely on the land station data and ship observations.

A number of automated classification schemes has been proposed in the last two decades and their different approaches have been summarized recently (Philipp et al. , 2010). In the present work, we make use of an automated version of the Lamb circulation type's methodology. This classification of atmospheric circulation into circulation weather types (CWTs) was first developed for the British Isles by Lamb, (Lamb, 1972). Originally, it was a manual classification, entirely based on the visual inspection of daily SLP fields over the British Isles. Later, it was adapted by Jenkinson and Collison (1977) for an objective automated classification for the British Isles that would be fully automated with the widespread use of personal computers (Jones et al. , 1993). This automated methodology was later adapted for different areas in the Iberian Peninsula (Goodess and Palutikof, 1998; Trigo and DaCamara, 2000). In this work, daily circulation weather types affecting Iberian Peninsula are characterized by the use of a set of indices adopted by Trigo and DaCamara (2000). Therefore we have computed the daily CWTs for the 1850–2003 period by means of the daily SLP from Ansell et al. (2006). The atmospheric circulation conditions were determined using the geostrophic approximation and adopting physical or geometrical parameters, such as the direction and strength of airflow and degree of cyclonicity, based on 16 grid points (Figure 2, points 1–16). A comprehensive explanation of the methodology can be found in the original work of Trigo and DaCamara (2000). Moreover, the application of the CWT methodology to the EMULATE daily historical European–North Atlantic SLP has been used with success and described for the Iberian Peninsula domain by the authors in Cortesi et al. (2013, 2013b). The method allows the identification of 26 different CWT (10 pure, 8 anticyclonic hybrids and 8 cyclonic hybrids). However, in order to work out a more practical, though reliable, statistical analysis scheme for the monthly frequency analysis, the 26 circulation types were re‐grouped into ten basic classes. This was done by simply adopting a similar procedure to that of Jones et al. (1993) and Trigo and DaCamara (2000): each of the 16 hybrid types was included with a weight of 0.5 into the corresponding pure directional and cyclonic/anticyclonic type monthly frequencies (e.g. one case hybrid like the A.NE is included as 0.5 in A and 0.5 in NE). Therefore, we obtain 10 circulation types, eight driven by the direction of the flow (NE, E, SE, S, SW, W, NW and N) and two by the shear vorticity (cyclonic or anticyclonic). Therefore, 10 distinct CWTs are considered, including 8 directional types dominated by strong non‐rotational flow (within 45° sectors), and two other CWTs dominated by high absolute values of geostrophic vorticity (cyclonic and anticyclonic types). The corresponding maps of each CWT can be found in Cortesi et al. (2013b).

Figure 2

Pressure grid points (1–16) used to compute the Lamb weather types for the Iberian Peninsula.

3 Intense floods and other precipitation events in the hydrological year 1855/1856

3.1 Documentary description

We have abstracted a great amount of newspapers of these dates looking for references to meteorology, floods, damages… The analysis of the documents makes clear that the most important damages occurred in January 1856. A general picture of the situation in this month is described by La Época newspaper (19 January 1856): ‘…the continuation of the storms, with the rivers overflooded, destroyed bridges, destroyed fields and livestock lost in the streams, villages and farms isolated and lacking everything. This is the general situation, especially in Castille, La Mancha (both in the central plateau) and Andalusia, especially in the areas close to the coast’. Other news in the diaries describe the continuous rainfalls and the melting of the snow in the mountains as the origin of the floods: ‘The rain storms, far from decreasing, have increased in the first fortnight of January 1856, this jointly with a high temperature, unknown in this season, has caused the melting of the snow in our mountains. Their waters have accumulated in the plains, with soils wet due to the previous rains, so they have increased the very high level in the rivers originating floods and ravages. We cannot see the end of this situation. (La Iberia, 23 January)’.

We have compiled a significant amount of news reporting problems due to the excessive rainfalls. Table 1 and Figure 3 summarize the damages occurred in different locations of Iberia.

Table 1. Damages caused by rainfalls during January 1856.

Location	Damages	Source
Santander (1) Bilbao (2) Barcelona (7) Valencia (23)	Important lost in the markets due to the intense rainfall	La Época, 19 January
Burgos (3)	The Arlanzon river overflows in Burgos, important damages in its left bank	Comisión Nacional de Protección Civil, 1988
Valladolid (4)	The Pisuerga river caused important damages	Comisión Nacional de Protección Civil, 1988
Zaragoza (5)	Huerva river arrived to 4.6 m in the San José bridge. The river overflows upstream and downstream	Comisión Nacional de Protección Civil, 1988
Mataró (6)	The railway was flooded and could not operate	La Iberia, 21 January
Zamora (8)	An important part of the city flooded. Boats were required to rescue the residents	Comisión Nacional de Protección Civil, 1988
Salamanca (9)	The Tormes river overflowed in Salamanca	Comisión Nacional de Protección Civil, 1988
Guadarrama (10)	Great snowfalls and blizzars	La Época, 19 de January
Torronteras (11)	A great chasm opened in the town affecting almost all houses, all the population were evacuated	La Esperanza, 29 January
Budia (12)	Important landslide and an important subsidence in a vegetable garden	La Esperanza, 29 January
Madrid (13)	Homes, vegetable garden, washing places and a train bridge were destroyed by the Manzanares river. Death of cattle and some people were evacuated	El clamor Público 8 January, Comisión Nacional de Protección Civil, 1988
Torrejon de Ardoz (14)	Some houses fell down	Oficial de Avisos de Madrid, 23 January
Alcantara (15)	The water level of the Tajo rose 30 meters and the volume of flow was 7000 m^3/s	La Iberia, 21 January
Sacedón (16)	An olive grove disappearedby subsidence	La Esperanza, 29 January
Escalona (17)	Alberche river destroyed the bridge and the town remianedcut off	Comisión Nacional de Protección Civil, 1988
Aranjuez (18)	17 people death, bridges lost (Tajo and Jarama rivers), damages in boats, death of cattle, numerous towns cut off	La Época, 19 and 28 January, El Clamor Público, 29 January; La Iberia, 29 January y La Esperanza, 28 January
Talavera (19)	Flooded homes, death of cattle, damages in the bridge	La Época, 28 January
Toledo (20)	Great infrastructure damaged in Toledo (Mills, dams, roads…)	Comisión Nacional de Protección Civil, 1988
Tembleque (21)	All the street were flooded	Comisión Nacional de Protección Civil, 1988
San Clemente (22)	Flood of the river Rus	El Balear, 6 February
Valencia (23) Carcagente (24) Sellent (25) Benegida (26) Alberique (27) Albaida (28)	Important damages in the Jucar basin (factory, cattle, buildings and fields)	Comisión Nacional de Protección Civil, 1988
Denia (29)	Boats in harbor were destroyed by a windstorm	La Iberia, 21 January
Alcocer (30)	The Guadiela river destroyed two arches of the bridge	La Esperanza, 29 January
Ribatejo (31)	The excess of rain and bad weather caused damages to the farmer sheds and confines them to extreme poverty. Luckily the fields were unseeded at that time	Ilustracao Luso‐Brazileira, 19 January
Santarem (32)	Dislodged people cattle and agriculture lost, communication blocked, supplies could not be delivered	Azevêdo et al. , 2004
Alhandra (33)	The Alhandra church is full of cracks and has collapsed partially	Ilustracao Luso‐Brazileira, 2 February
Lisbon (34)	Just a few days ago we could walk in Lisbon, now we sloshes, soon one will sail	A Illustraçao Luso‐Brazileira, 12 January
Cape Raso (35)	10 casualties in a shipwreck	Ilustracao Luso‐Brazileira, 19 January
San Pedro de Mérida (36)	On 10 January 1856 the town was flooded by the Guadiana river and other stream	Comisión Nacional de Protección Civil, 1988
Jaén (37)	The Guadalmedina and Guadalhorce rivers damaged the fields	La Época, 19 January
Écija (38)	Great flood, many people was evacuated	Diario Oficial de Avisos de Madrid 1 February
Algaba (39)	More than 80 houses fell down	Comisión Nacional de Protección Civil, 1988
Seville (40)	More than a third of the city was flooded during 25 days, from 6th to 30th of January. In some districts the water flooded houses through the balconies	La Esperanza, 29 January Comisión Nacional de Protección Civil, 1988
Algarve (41)	Important losses in agriculture due to bad weather and damages in fishing boats due to stormy condition of the sea	Ilustracao Luso‐Brazileira, 19 January
Granada (42)	Important rainfalls, the river Saleres overflowed	Diario Oficial de Avisos de Madrid 1 February
Orgiva (43)	Affected by floods	Diario Oficial de Avisos de Madrid 1 February
Velez‐Malaga (44)	The town was completely surrounded by water, the access was impossible	Oficial de Avisos de Madrid, 23 January
Malaga (45)	The Guadalmedina river flooded streets and houses, a stone bridge was destroyed, the city was cut off with the other towns of the province	El Balear, 3 January; Diario Oficial de Avisos de Madrid, 23 January
Jerez de la Frontera (46)	68 houses evacuated by risk of collapse	El Balear, 28 January
Cádiz (47)	Horrendous storms	La Época, 19 January
Estepona (48)	All the beach full of fruit trees that have been uprooted and transported by the water from the gardens	La Epoca, 2 January

Figure 3

Location affected by the excessive rainfalls during January 1856 (cited in Table 1).

According to the spatial distribution of the affected locations, the Inner Plateau and Andalusia were the regions mostly impacted by rains and floods. Nevertheless we can find description of damages in all the important hydrological basins of Iberia (Duero, Ebro, Tagus, Guadiana, Guadalquivir and Jucar) with prominent clusters close to Lisbon (Tagus) on the western coast and Valencia (Jucar) close to the Mediterranean.

As expected, the above described disturbances in crop fields and transport facilities caused widespread socio‐economic effects. The prices of the basic foodstuff increased and most humble people and day labourers could not afford them. Reports dated on 19 and 21 January describe their misery (El Clamor Público 21 January and La Iberia 19 January ).

3.2 The instrumental record. How anomalous was the event?

We have been able to obtain instrumental measurements in 14 Spanish locations for this period in (Dirección General de Obras Públicas, 1856) and one series from Portugal (Fragoso et al. , 2010) which allow quantifying the intensity of the precipitation.

Figure 4 shows examples of the monthly evolution of precipitation relative to the cities of Soria, Valladolid, Madrid and Seville and spanning from September 1855 to August 1856 (i.e. the 1855/1856 hydrological year used in Iberia) compared with the long‐term percentiles obtained for the common period 1972–2012. The records for September and October 1855 and January 1856 confirm the outstanding nature of these months with values above percentiles 90th and even above 98th. April 1856 shows high percentiles in some stations (particularly Valladolid) but, according to the documentary sources, this month was not characterized by important floods, nor social problems or fatalities. As we have seen, according to the newspapers, the most important damages occurred during January 1856 affecting almost all the large hydrological basins of Iberia. The instrumental record confirms that the precipitation of January was not the sole responsible for these damages, as we have seen in the newspapers. The extraordinary wet autumn of 1855 (especially September and October) filled at full capacity the aquifers, without causing important damages. Then a regular November and December maintained this condition, and finally, the high precipitation of January in combination with the contribution of the aquifers, induced large values of runoff in many watersheds, that caused all the damages described in the previous section.

Figure 4

Monthly precipitation values from September 1855 to August 1856 in Madrid, Soria, Valladolid and Seville and long‐term (1972–2012) 10th–98th percentiles.

Figure 5 shows the spatial pattern of the precipitation in percentiles obtained from the common period 1972–2012 during the three most extreme months (September, October and January) in the available stations. In September 1855 all the stations, with the exception of Zaragoza, Granada and Barcelona recorded values above the 70th percentile, Seville exceeded the 90th percentile and Madrid, Valladolid, Soria and Oviedo surpassed the 98th percentile. The outstanding monthly values were greater in the northern Inner Plateau. During October 1855, the largest values were recorded in all the Inner Plateau and in Granada, where the 98th percentile was reached. The values decreased in the Mediterranean area where Barcelona and Tarragona showed percentiles below 50th.

Figure 5

Precipitation in percentiles respect to the long‐term (1972–2012) for September and October 1855 and January 1856 for the 15 stations available.

In January of 1856, the high values continued in the northern Inner Plateau but were also recorded in the Ebro valley, Seville, Malaga and Lisbon (this station only started measuring on this month and no data are available for the previous months).

In summary during the most intense 3 months almost all the peninsula reached percentiles above the 70th being the northern Inner Plateau and the Seville region the most anomalous, researching percentiles above 90th and 98th percentiles.

3.3 Synoptic analysis

Next we analyse the underlying dynamics during the key months of September, October 1855 and January 1856. To achieve this, SLP monthly means and the 10th and 90th percentiles were computed taking into account the entire distribution relative to the 20th century (1900–1999). For the three key months (September, October 1855 and January 1856) monthly anomalies were computed after removing the corresponding monthly means. Figure 6 shows the SLP anomalies obtained with respect to the corresponding average for the long‐term period 1900–1999. In addition, we have highlighted the grid points that show SLP values above (X) and below (O) the 90th and 10th percentile monthly distribution, respectively. Table 2 provides the monthly frequency of the different CWTs for the 30 months assessed and the average and 10 and 90 percentiles during the reference 1900–1999 period. Figures 7-9 show selected daily SLP fields.

Figure 6

SLP anomalies of September and October 1855 and January 1856 with respect to the period 1900–1999. (X) SLP values above 90 percentile and (O) SLP values below 10 percentile.

Table 2. Percentage of different weather types.

Weather type	September				October				January
	1855	10th per	Mean	90th per	1855	10th per	Mean	90th per	1856	10th per	Mean	90th per
North	13.3	1.7	9.3	18.3	1.6	0.0	7.0	16.3	0.0	0.0	5.5	12.9
Northeast	28.3	4.8	14.1	25.3	8.1	0.0	9.3	21.3	3.2	0.0	6.6	14.7
East	10.0	3.3	16.7	33.3	3.2	0.0	10.7	22.6	0.0	0.0	5.9	16.1
Southeast	1.7	0.0	5.0	13.3	0.0	0.0	5.3	16.13	9.7	0.0	3.6	9.8
South	3.3	0.0	1.7	5.0	3.2	0.0	4.3	13.1	3.2	0.0	3.0	8.1
Southwest	0.0	0.0	2.4	6.7	8.1	0.0	5.9	14.7	30.6	0.0	6.9	16.3
West	10.0	0.0	5.2	13.3	29.0	0.0	9.4	21.0	16.1	0.0	13.1	27.7
Northwest	3.3	0.0	7.9	18.3	21.0	0.0	7.7	16.1	0.0	0.0	10.2	22.6
Cyclonic	13.3	0.0	7.5	18.5	4.8	0.0	7.6	17.7	8.1	0.0	5.8	12.9
Anticiclonic	16.7	8.3	30.9	55.2	21.0	12.7	32.8	61.4	29.0	19.4	39.5	61.3
C + W + SWa^a Sum of the percentages of the weather types associated with rainfall in Iberia (Cyclonic, West and Southwest).	23.3	3.3	15.1	26.8	41.9	6.3	23.2	47.1	54.8	3.2	25.8	55.0

• The table includes the long‐term means (1900–1999); the 10th and 90th percentiles and the percentage for the anomalous months October and September 1855 and January 1856 (Values in italic indicates above the 90th percentile).
• ^a Sum of the percentages of the weather types associated with rainfall in Iberia (Cyclonic, West and Southwest).

Figure 7

Daily sea level pressure for selected days of September 1855.

Figure 8

Daily sea level pressure for selected days of October 1855.

Figure 9

Daily sea level pressure for selected days of January 1855

According to the figures, the spatial configuration obtained for October and January is clear. In October, we observe negative departures located W‐NW of Iberia, which, according to the daily fields were associated to low pressure systems over the British Isles. This is evident in Table 2, with W and NW weather types frequency showing values much higher than their corresponding 90th percentiles. January was characterized by the occurrence of persistent low pressures located over the Atlantic just west of Iberia, as depicted by Figure 6 bottom panel and confirmed by Figure 9 and Table 2, which shows that the precipitation were mostly associated to wet types (W, SW and C), which accounted for more than 50% of the days (Table 2). The SW was especially abundant, with a frequency of 30.6%, well above its 90 percentile. It was the sixth highest frequency in the record just below 1851, 1866, 1875, 1937 and 1970.

However, the situation during September is much less clear. Figure 6 upper panel shows that positive departures dominated the Atlantic west of Iberia, a situation which usually leads to dry weather. According to the descriptions obtained from Manuel Rico Sinobas (Rico Sinobas, 1855) rains were frequent during the whole month, concentrated at the beginning and the end of the month in the North Coast, Ebro, Duero, Tajo and Guadalquivir basins, with 17 rain days in Soria, 15 in Valladolid and 13 in Salamanca (all of them located in the northern Inner Plateau, Figure 1). In the Mediterranean coast the rain was concentrated in the first 9 days. In addition the documents state that some of these precipitation events were associated to sudden storms, downpour and electric storms. Values in Table 2 show that the frequency of wet CWTs was high, but not above the 90th percentile. According to the SLP daily fields selected for Figure 7 Iberia was under the influence of the Azores high, which is usually associated to dry weather, during the first third of the month. The most likely explanation for the high precipitation of these days is that they were associated to high level intrusions of cold air or cut‐off low systems common in September (Nieto et al. , 2005). These situations are characterized by high instability and generate storms with frequent electric discharges (Ramos et al. , 2011) and are often not reflected in SLP fields. This mechanism should also explain the high precipitation in Oviedo (above the 98th percentile, Figure 5), since this is one of the most common entrance areas of these perturbations into Iberia. This type of anomalies is usually associated to low temperatures. So, to check the consistency of this interpretation, we looked for evidences of cold temperatures. In Cádiz, the September departure with respect to 1803–2004 was −1.5 °C (Barriendos et al. , 2002). We have found additional evidences in press reports from other parts of Iberia. According to El Clamor Público , by 8 September the cold temperatures and intense precipitation had stopped the open air activities in Madrid. According to a report from the same newspaper dated on 6 October, flies had disappeared much earlier than usual. This phenological reference is interesting, since the cold temperatures made that the Musca domestica should stay as larvae as early as September. So, cold‐air intrusions associated to upper level cut‐off‐lows are the most likely mechanism to explain the extremely wet September.

4 Conclusions

This article describes the spatial and temporal characteristics of the unusually wet year of 1855/1856 in the Iberian Peninsula. The extreme precipitation in September and October, followed by wet conditions in the following months and extreme precipitation in January, led to catastrophic floods in several parts of the Peninsula. These extreme rains affected most of Iberia, but were especially intense in the inner plateau, an area usually characterized by dry and cold winters. This analysis has been possible because of the availability of a fruitful mix of sources that we have used for this purpose, including (1) documentary sources (mainly newspapers reports with eyewitness accounts), (2) precipitation data from 15 stations scattered throughout the Iberian Peninsula and where 11 of these stations have been rescued and digitized for the first time in this paper and (3) the longest gridded SLP data available over the Atlantic‐European window (Ansell et al. , 2006). The use of this dataset was particularly relevant, as it allowed to obtain an equally long automated atmospheric circulation classification scheme over Iberia, which revealed the main features of surface atmospheric circulation during the considered year. We show that there is a good correspondence between SLP anomalies and precipitation during October and January, with the predominance of wet circulation types (W, SW and C). However, in September, surface circulation cannot alone explain so easily the occurrence of such extreme rain spells and we hypothesize that cold upper air intrusions were the main responsible of the anomalous precipitation.

This case illustrates how currently available tools such as early observations and gridded data are useful for the analysis of extreme events, but also that they are insufficient to characterize fully the associated dynamics. In this sense, the recovery of early upper air data would improve our understanding of extremes in this early instrumental period.

Acknowledgements

Angel Rivera and José Francisco Mediato discussed an early version of the manuscript. Alexandre M. Ramos was supported by the Portuguese Science Foundation (FCT) through a Post‐doctoral Grant (SFRH/BPD/84328/2012). Two anonymous reviewers contributed to improve the original manuscript. Fernando Domínguez‐Castro was partially supported by the Prometeo Project, Secretariat of Higher Learning, Science, Technology and Innovation (Ecuador government). This work was partially supported through projects AYA2011‐25945 (Ministerio de Economía y Competitividad of the Spanish Government), KlimHist PTDC/AAC‐CLI/119078/2010 (Fundacão para a Ciência e a Tecnologia), STORMEx FCOMP‐01‐0124‐FEDER‐019524 (PTDC/AAC‐CLI/121339/2010) by FEDER (Fundo Europeu de Desenvolvimento Regional) funds through the COMPETE (Programa Operacional Factores de Competitividade) Programme and by national funds through FCT (Fundacão para a Ciência e a Tecnologia, Portugal) and the support from the Junta de Extremadura (Research Group Grant No. GR10131).