GoKawiilDevelopment of the historical flood database
Flood evidence in Europe over the period 1341–1343 was collected from narrative sources (chronicles, annals) and legal and economic administrative documentation (charters, accounts) as well as other documentary source types (thanksgiving inscriptions/memorials, scientific discussion, flood marks) (Supplementary Data 1–3). Although chronicles provided most of the information in most parts of Europe, in some areas, such as the Carpathian Basin, charters were more relevant16. Chronicles and annals often captured the greatest floods in a systematic way, but did not cover smaller floods or floods that occurred shortly after a great one. The other source types provided non-systematic evidence on individual flood events and were usually more precise in location.
We applied critical historical source evaluation, removing evidence deemed unreliable, and separating contemporary and non-contemporary sources: around 77% of the flood data entries come from contemporary sources, the rest from medieval sources based on earlier, contemporary source evidence. All floods are covered by contemporary sources, and other medieval sources were only used as complementary evidence. For over 70% of the floods, several types of independent contemporary source evidence was available. Giving priority to contemporary sources sets this study apart from previous research and enhances its reliability, as contemporary sources are first-hand information. This process resulted in 168 flood entries in the database on 60 European rivers, and four large lakes and wetlands.
In the period 1341–1343, several major floods occurred on the same river, including the Rhine, Tisza, Po, Elbe, Danube, Axios, Main, Inn, Drava and the Altmühl. For example, in the upper sections of the Tisza River, three major and two notable independent floods were recorded. Inspired by the term ‘Magdalena Flood’ used to describe the flood that peaked on the day of Saint Maria Magdalena, we propose similar names for three other major floods: the Candlemas, Jacob and Bartholomew Floods (Extended Data Tables 1 and 2).
Magnitude class and return period estimation
We assess the magnitudes of each event in two ways. In a first step, we categorize each flood during 1341–1343 into one of three classes, consistent with the 500-year European database11 using the source texts, where class 1 represents notable floods, class 2 great floods and class 3 extraordinary floods. As more detailed information was available for the 1341–1343 floods than for the 500-year series, we also assign return periods in a second step. As expert-based, interpretative assessments, the return periods (Fig. 5) draw on a comprehensive collation of local and regional historical quantitative and qualitative water-level data, complemented by archaeological and sedimentary evidence. Even where quantitative indicators such as preserved flood marks are available, their interpretation necessarily reflects the substantial long-term modification of catchments and river systems. Accordingly, the estimates should be understood as contextually informed reconstructions that integrate several lines of evidence within an evolving environmental framework. Direct comparisons between medieval and modern floods are complicated by substantial anthropogenic changes in river morphology, floodplains and catchment hydrology since 1342. Accordingly, our return period estimates do not rely on direct comparisons of absolute flood levels or damage with modern gauged floods, but instead use a relative, rank-based approach that is standard practice in historical flood hydrology. We both identify most likely return periods and an uncertainty range. The method, detailed in Supplementary Information section 1, depends on the information available55:
1. Exact numerical information on the maximum water level or the maximum flood extent: these cases were compared with modern floods where the relationship between water level and return period are known, taking into consideration the possible uncertainties and changes that occurred in the landscape, hydromorphology, hydrology and river dynamics over time. Most of the maximum water-level estimations presented in Fig. 5 belong to this category. For example, based on flood marks and maximum water-level descriptions with exact localization, with additional information about changes in the environmental conditions such as the landscape, the town wall and the river morphology, the Magdalena Flood on the Main at Würzburg was estimated as a 1,000-year return period flood, and uncertainty ranges were identified6. Based on the report by local monk Johann von Winterthur56 that the Barfüsser Church was flooded up to the upper church (or podium), and on the account by Heinrich von Diessenhofen that, at roughly the same time, the town wall of Konstanz was overtopped only at the Dominican Friary, we estimated the return period of the Bartholomew flood in Lindau to be around 1,000 years57,58,59,60 (Supplementary Information section 1). This estimate accounts for lake sedimentation rates, differences in church floor elevations in Lindau and the known height of the medieval town wall in Konstanz Return period estimate with uncertainties for the same flood are also available for the Upper Rhine at Basel55,60 (Supplementary Information section 1). Contemporary chronicler Fritsche Closener58 wrote that the Bartholomew Flood was 2 feet higher than the Jacob Flood in Strasbourg, from which the differences in the return periods on the Rhine were estimated. Contemporary sources also report the height of water standing in churches (for example, in Vendôme and Luzern) and note that churches in Rouen and Luzern could be entered by large boat during the Candlemas and Jacob Floods, respectively. From these accounts water levels and return periods were inferred, considering elevation changes. In many cases hydro-morphological, and archaeological data were used to assist in the comparison between fourteenth century and modern conditions (Supplementary Information section 1). 2. Descriptive, quasi-numerical information related to the flood magnitude. In this case, the return period was inferred from the assessed frequency of how often a similar situation occurred over the centuries and how unusual it was in terms of infrastructure damage. For instance, during the Candlemas Flood in Prague59, where two-thirds of the famous Judith stone bridge were destroyed, we estimated a return period of at least 300 years. This estimate is based on the fact that the bridge was constructed in the twelfth century, taking previous floods into consideration. The Candlemas Flood was probably the most outstanding flood in several hundred years. The exceptionally severe damage to riverine infrastructure and town buildings, together with mass fatalities and the destruction of entire villages, supports this estimate (Franciscus Pragensis)59. Based on this and other information, such as a late medieval flood mark, and alluviation data from archaeological profiles, the return period of Candlemas in Prague was estimated as 500 years with a lower limit of the uncertainty range of 300 years. The upper limit of the uncertainty range was estimated as 1,000 years from the fact that the Roudnice bridge 50 km downstream of Prague was not damaged (Supplementary Information section 1). These estimates are similar to the Candlemas return periods for Vendôme estimated in step 1. Similarly, Heinrich von Diessenhofen60 reported that during the Jacob Flood all bridges along the Rhine between Schaffhausen and Strasbourg were completely destroyed and Johann von Winterthur56 noted that the vineyards (never placed in flood-prone areas) and entire districts of Lindau were destroyed, suggesting a return period of around or greater than 100 years (Supplementary Information section 1). 3. Information on the flood magnitude expressed by the choice of language and the broader environmental, legal or military implications: Here again, the return period was inferred from the assessed frequency of the event and, in some cases, from infrastructure damage, which provided indications of how far flood waters extended across the landscape and thus allowed estimation of water levels. The mid-March 1342 flood61 at the lower sections of the Upper Tisza obstructed the legal processes over debated land boundaries to the extent that survey participants could not even reach the region of the settlements in question, suggesting an event that does not happen more often than once or twice a century16. In the same area the medieval environment was described in this and other charters with exact dating, and detailed maps are also available in the eighteenth and nineteenth centuries, which make it possible to locate the flood and its approximate extension16.
The Strassburger Chronik of Fritsche Closener58 describes the Jacob Flood at the Rhine “Do man zalte 1343 jor, do wart (der) Rin also groß und ging also sere us, daz nieman do zemol lebete der üt gedohte oder ie hette gehoret sagen, daz er so groß würde.” (The flood level and extent was beyond imagination of anybody living or any reports), from which we inferred a minimum 100-year return period (three generations), based on an interpretation of communicative memory62. More broadly, the term ‘diluvium’ (deluge) appears only rarely in medieval Latin texts and is generally understood to denote events with a return period of several hundred years, and therefore related to cultural memory62. It is used for all four major floods, for example, by Heinrich von Taube63 for the Magdalena and Jacob floods in Eichstätt, and by Franciscus Pragensis for the Candlemas Flood in Prague59.
Further details are available in Supplementary Information section 1, where case studies of return period estimations are presented on the examples of the Candlemas, Magdalena, Jacob and Bartholomew Floods, and are based on refs. 64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157.
Development of historical weather database
... continue reading