Abstract
Snow events are not a rare episode in Mediterranean area, especially in northern and hilly areas of Italy. However, snowfall occurring quasi-simultaneously in the whole peninsula is extraordinary. This study collects, reconstructs, and analyzes the extraordinary snowfall episodes that occurred simultaneously in the whole Italian peninsula since 1709. This is the longest snowfall time series in the central Mediterranean area. The data, obtained by several documentary sources (from ancient archival to online databases), have been analyzed using different statistical tests, in order to explore normality, homogeneity, and stationarity. The results are characterized by a time-series stationarity with a quasi 60- and 100-year-dominant oscillation. No clear trend in the snowfall episode records is found. The 60-year cycle roughly matches with global-scale oscillations linked to natural forces, such as the Atlantic Multidecadal Oscillation and the winter North Atlantic Oscillation. Particular attention was directed to analyze the impact of snowfall on pre-industrial society, underlining the differences among northern and central regions, where snow was a more usual phenomenon, and its impact was mainly on transports of supplies or exceptionally on buildings, and southern regions, where it had a stronger impact also on orchards and cattle.
Introduction
Snow events have been always a fascinating phenomenon for people in the Mediterranean, and an important and essential water resource for the hydrological cycle locally as well as globally. Our understanding of the snowfall characteristics in several regions of the world is limited by the short record availability and by the poor knowledge of complex weather and climate interactions. These data are becoming increasingly important because of the high variability of snowfalls on a global scale (e.g. Davi et al., 2012) and the impact that snow can have on society, agriculture, and water resources. Therefore, a systematic and detailed knowledge can be of great support in the climate-related decisions of global public and scientific agendas, particularly to assess effective prevention and protection plans in case of extraordinary snowfall episodes.
Hitherto synthesis on snow change and its impact on society involved especially remote and mountain areas of the world (Huber et al., 2005). The issue still remains open for a multisite global research to explore both decadal to interdecadal climate variability scale and the departure from previous climate periods as the ‘Little Ice Age’ (LIA).
For the Italian regions, the longest snow recurrence time series goes back to the year 1741 for Rome (Mangianti and Beltrano, 1993, see also no. 23 in documentary sources list, from now on: doc. no.), 1788 for Turin (Leporati and Mercalli, 1993), 1866 for Naples (Andreotti Majo, 1958, doc. 41), and 1884 for Montevergine (Diodato, 1997, doc. 57).
The first aim of this paper is the reconstruction of long homogeneous time series of annual snow occurrence in northern, central, and southern Italy by collecting data from historical descriptive sources. The homogenized series of snow events is then documented and analyzed in order to investigate the presence of trends and cyclical patterns, and the results discussed in the light of published research. The phenomenon was also analyzed detecting the main impacts it had on the pre-industrial Italian society, compared with the present ones.
Data and methods
Reconstructing time series of extraordinary snowfall episodes may seem trivial; on the contrary, this type of data collection and analysis is far from easy. To this aim, documentary data provide more homogeneous information in space and time compared with instrumental data, which, going back in time to 18th century, become non-detailed, sporadic, limited in metadata, not homogeneous in terms of measurement methods, and recording only the number of snowy days and not the amount of precipitation (Eredia, 1910). In recent times (20th century), measuring parameters concerning snowfall have been established: accumulation of snow on the ground in the last 24 h, corresponding rain, thickness of the snow cover on the ground and its duration, and number of days with snowfall. All these information, under the qualitative point of view can be extracted also from documentary data; they have a high qualitative value because they give us detailed information about the development of the phenomenon and about its exact meteorological and anthropic context.
The snowfall may vary in duration and geographical location as a function of several factors, such as geographical latitude, altitude, topography, and other weather factors. Snowfall at low altitude is rare in the Italian peninsula. It usually occurs associated with episodes of cold spells, whose frequency and intensity tend to decrease as the cold increases due to the decrease in absolute humidity. Moreover, the spatial distribution of the snow event is strongly influenced by the trajectory of the minimum baric in transit; a few hundred kilometers away can create different orographic effects in terms of ‘snow shadows’ as mountain chains hinder air mass motion attenuating or amplifying the perturbation effects in a complex and not easily predictable way.
Taking this information into account, data were selected from historical documentary sources following strict criteria. The information extracted has been individually checked in terms of dating and evaluation of the reliability of described events. In the present snowfall time-series reconstruction, a temporal and spatial approach to the calibration of proxies against instrumental data is used. The temporal approach evaluates the strength of the climate signal in the documentary description (i.e. 1 = occurrence of extraordinary snowfall at low altitude; 0 = no occurrence) properly assessing the winter date from the context (whether the ‘old’ or the ‘new’ year is meant) and in comparison with the year recorded by instrumental series. The spatial approach determines the quality of the proxy climate signal by assembling proxy from a number of locations (extended geographical area) evaluating them properly after the overlap with instrumental records over the same geographic area.
Three main criteria have been followed in selecting data: extension in the whole peninsula (snowfalls documented simultaneously in different places all along the peninsula), low-altitude occurrence, and magnitude (intensity and/or long-lasting). For northern and central Italy, where snow events at an altitude >200 m a.s.l. are very frequent (Alpine and Apennine regions), only events that occurred at a lower altitude can be considered exceptional and therefore have been taken into account (Figure 1).
Geographical location of the main sites quoted in the sources. Main Italian plain in the peninsula is indicated as in the legend.
With respect to the magnitude, we have considered only events (episodes) described by the sources as relevant in terms of consistency (a declared exceptionally high snow layer) or duration (lasting more than 2 days, considering an ‘episode’ of an event lasting for more days when the occurrence was due to the persistence of the same meteorological factors); this choice aimed to exclude the common (normal) events of snowfall.
The selected time span covers the last three centuries; it begins with the winter of 1709, year of the ‘big cold’ in the whole of Europe, one of the (or the) best documented winter of the past. A few years later (in Italy, Rome, 1740), meteorological instrumental records began to be quite widespread and more and more regular, thus allowing – when necessary – cross-checking with documentary reports (Tables 1 and 2; Figure 2). Figure 2 highlights the comparison between ‘proxy’ climate records (upper plot) and the instrumental series of snowfall in Rome (lower plot). The reconstruction from documentary sources is reliable, the largest discrepancy being since 1940 the lack of certain documentary data showing snowfall events lower than 7 cm. This fault demonstrates how some events may escape from the documentary reconstruction as historical information often emphasizes extreme conditions or extreme severity.
Snowfall instrumental series collected and analyzed in this work for the main Italian cities located at an altitude below 200 m a.s.l.
Comparison of snow events that occurred in Rome as retrieved by documentary sources (top in Figure 2) and instrumental data (bottom in Figure 2) measured at the Osservatorio Romano in Rome (1740–2010). Measured snowfall amount is divided by classes of 5 cm each and reported in the first column. The second and third columns summarize, respectively, the frequency of snowfall events recorded and not recorded by documentary proxy. Numbers in bold highlight the 15-cm snowfall threshold below which documentary proxy may sometimes lose the event recording.
Comparison of the occurrence of snow events in the Italian plain as retrieved by documentary sources (top) and the historic instrumental data (bottom) measured at the Osservatorio Romano in Rome (1740–2010).
Data sources and collection
Data have been selected from historical sources contemporary to the quoted events. Types and characteristics of documentary sources change during the centuries; in the 18th century, the most common ones are local chronicles or diaries; from the end of this century and during the following one, scientific publications become more and more widespread, both the specialized and the popular ones, constituting our principal sources. Another interesting source is medical treatises studying the interactions between climate and human health, a scientific tradition particularly in fashion during the second half of the 19th century. Some 300 documentary direct sources have been individually checked, selected, and analyzed. For the episode concerning the winter of 1709, a specific modern monography, containing 91 selected ancient records, has been used (Finzi, 1986). A great amount of information, particularly in the 19th and following centuries, is also given by gazettes and agricultural journals.
When referring to the recent years, online literature also has been used. We used online databases that collect climate-related information, such as the Hydrogeological Disasters Prevention database, to collect snowfall events in the present time. This database was created compiling an inventory of sites historically affected by catastrophic events related to hydrological phenomena (e.g. landslides, floods, and heavy snowfall) for the period 1918–1994. In the AVI Project for Prevention of Hydrological Hazards, the completeness and reliability of this database was assessed within the project (Table 3). A total of 22 journals were systematically searched, 350,000 newspaper issues were screened, and 1000 published and unpublished technical and scientific reports were reviewed, testing the consistency and performing a validation of the information content. This modern database represents the most comprehensive source of information concerning extreme events that could affect population or isolate large areas of Italy.
Occurrence and snowfall amount for the main Italian cities located at an altitude below 200 m a.s.l. retrieved by the AVI database of the National Group for Prevention of Hydrological Hazards (GNDCI) of the National Research Council.
Cities in central and southern Italy with altitude between 200 and 700 m a.s.l. (i.e. Perugia = 494 m a.s.l., Campobasso = 701 m a.s.l., and Catanzaro = 320 m a.s.l.).
All these ancient and modern descriptive sources are of great interest for the study of the phenomenon considered here because they exceed a limit of historical instrumental data, that is, to quote only the total monthly amount of melted precipitation (rain plus snow) or, simply, the annotation of snow presence (Figure 3). The exclusive use of instrumental data does not allow to fully understand and evaluate the actual quantitative and qualitative entities of the event because they lack details useful for a global view of the phenomenon, such as the height reached by the snow or its duration on the ground.
An example of part of a 19th-century meteorological table (year 1864, January; Cappelli, 1864).
Data analysis
After the search of suitable documentary data, following the criteria expressed above, the snowfall time-series reconstruction starts with a critical analysis of sources. The comparison and cross-checking of data from different documents, sources, and involved areas allows an assessment of unequivocal extreme episodes at an altitude lower than 200 m a.s.l. The documentary information is then transformed into a time series. In this phase, the statistical transformation of the proxy data into ordinal data in the form of a time series requires both a broad statistical and a dynamical understanding.
The climatic tendency is generally expressed in the form of an intensity index that, in this paper, was fixed in terms of the number of snowy episodes attributed per year. This scaling choice was used depending on the low density of the basic information (i.e. extreme snowfall episodes). However, it could be to some extent affected by the subjectivity of the researcher; for this reason, we applied several quality-control tests on the data. This data validation is a stage in data analysis which is essential when data could be affected by bias, without it, data from different documentary sources cannot be combined in a long time series.
Results on data quality-control tests (e.g. Gandin, 1988) ensure that possible errors arising from data acquisition and check process, inconsistencies, and other anomalies in the data are apparent to the user, who has sufficient information to assess its suitability for the final time-series reconstruction. In short, quality control must detect both random and systematic errors.
Quality checks performed on the data are as follows:
Analysis of missing observations in order to assess whether missing values are distributed randomly.
Analysis of extreme observations to check whether outliers change the distribution.
Comparison and correction of differences in indexing through the comparison of information or variables external to the data set (e.g. instrumental data).
The analysis of systematic and random fluctuations of the time sequence of snowfall events provides relevant information for climate change studies and for statistical modeling and long-range climate exploration. Exploratory and time-series analyses with annual snowy data were performed by spreadsheet-based tools and online available software: STATGRAPHIC online (http://www.statgraphicsonline.com) for interactive statistical analysis, AnClim (http://www.climahom.eu/software-solution/anclim) tool for time-series analysis and homogenization (Štěpánek, 2005), and Free Statistics and Forecasting Software (Wessa, 2012, http://www.wessa.net).
Results and discussions
This section shows the results obtained by analyzing the time series of extraordinary snow recurrence. First, both the graphical and statistical analyses carried out to test for normality, homogeneity, and stationarity of the data are illustrated. Then, the results of a trend analysis are presented and discussed, with respect to the published literature. The time series is also analyzed for the presence of cyclical patterns, and the results are compared with previous studies. A specific paragraph is devoted to the observation of the impact of snow events on pre-industrial society.
Normality and homogeneity testing
All processes employed to generate the entire snow time series could have potentially affected the data for the time period investigated. In particular, inhomogeneities may have been caused by changes affecting the statistical properties of the observations through time. A set of statistical and graphical tests was used to determine whether the snow recurrence events on the reconstructed series can be considered homogeneous to a certain degree of accuracy. The distribution plot (Figure 4) indicates that the time series is highly asymmetrically distributed.
Italian Peninsula. Histogram of frequencies of annual snow recurrence for the 1709–2013 period.
The classical cumulative deviation test by Buishand (1982), applied to detect a possible change point of the mean, shows that the null hypothesis of homogeneity is not rejected for significance level, p = 0.05 (Figure 5a).
(a) Cumulative deviation test (critical value at p = 0.05) for checking time-series homogeneity (AnClim; Štěpánek, 2005). (b) Variogram showing the change in variance of snow occurrence with increasing time lag between pairs of yearly observations (analysis performed by the PAST Software).
Finally, the temporal correlation of snow recurrence was investigated using a variogram in the temporal domain, which is essentially a plot of dissimilarity over time distance. The range of a variogram is the lag distance beyond which data are independent of each other (no temporal autocorrelation). Modeling a variogram is similar to fitting a least squares line in regression analysis. It represents the internal structure (temporal stochastic memory) of the time series. A variogram is said to display a hole effect when its growth is not monotonic (nonstationarity). The hole effect model typically reflects pseudo-periodic or cyclic phenomena (Journel and Huijbregts, 1978). The shape of the experimental variogram obtained in this study (Figure 5b) appears to be close to a hole model, in which a periodic structure with initial high values of variance is at a relatively short distance.
Time-series reconstruction and cyclical patterns detection
Figure 6a shows the reconstructed extraordinary snow annual occurrence in time domain for Italy from 1709 to 2012. The average annual frequency is about 0.31 times per year with positive standard deviation equal to 1 time. It emerges from the reconstructed series in Figure 6a that peaks of snow occurrence (above line of one standard deviation σ) appear at intervals that are not dependent on time periods (around 1740, 1860, and 1960), although there is a decline in the last period. At the same time, in Figure 6a, lower snowy episodes (below line of one standard deviation) show a regular manifestation with a fairly uniform spread throughout both the end of the LIA (18th and 19th centuries) and the most recent warming period. However, this conclusion should be supported by the general notion (Field et al., 2012) that frequency and severity of extreme events are altered by changing climate means.
Analysis of cyclical patterns in the time series of annual snowy episodes (1709–2012) reconstructed for the Italian Peninsula. (a) Reconstructed series (histogram) and its low-frequency Fourier component signal for the 60-year cycle (solid gray curve). Horizontal gray line indicates bound of one the standard deviation (σ) around the mean. (b) Background spectrum cycles using the wavelet generation software provided by Torrence and Compo (1998), available at http://paos.colorado.edu/research/wavelets. Significant cycle values at p = 0.10 are given as oblique gray line.
Long-term changes in the snow series were investigated to uncover patterns of temporal variation. The nonparametric Mann–Kendall statistic was used in identifying a trend in the time series of snow (S > 0, increasing trend; S < 0, decreasing trend; Kendall, 1975). The absence of a trend is supported by the normal approximation of the calculated Mann–Kendall statistics (S = −655), for which the null hypothesis of no trend in the time series is not rejected (p = 0.71). This indicates that any change in snowfalls that occurred in Italy between 1709 and 2012 was minimal because the average yearly snowy episodes remained roughly constant, as reflected in the finding of non-significant trends in the year-to-year changes in snowfall semivariance patterns.
A Fourier transformation was performed to express the time function in terms of frequencies (harmonics) in order to test nonstationarity with cyclical patterns in the observed snowy episode variations. The result, displayed in the form of a quasi-60-year oscillation sine gray curve with an amplitude of 0.45 times the snowy episode per year, is superimposed in Figure 6a. Moreover, to avoid losing, with Fourier transformation, the wavelet transformation was also used to construct a time–frequency representation and detect possible changes in cyclicity characteristics over time (Torrence and Compo, 1998). The result of the wavelet analysis, in Figure 6b, emphasizes the power contribution of 60- and 100-year periodicity that exceeds the red noise of the global spectra.
Finally, the temporal correlation of snowy series was investigated using a semivariogram in the temporal domain, which is essentially a plot of dissimilarity over time distance. The range of a semivariogram is the lag distance beyond which data are independent of each other (no temporal autocorrelation). A semivariogram displays a hole effect when its growth is not monotonic. The shape of the semivariogram obtained in Figure 5b indicates a first steep slope and a change in slope at a time sample distance (around 60 years) at which snowy episodes manifest hole effect correlation (periodicity after Journel and Huijbregts, 1978). This cycle of 60 years roughly matches with the results observed both in the previous Fourier and wavelet analyses, providing sufficient evidences in recurrence of extraordinary snowy events. The 100-year cycle has been rejected according to the Nyquist criterion to avoid a predicted period as alias.
Circulation pattern influences
In order to relate the climate forcing to the snowy extreme episodes over the whole Italian peninsula, a study was carried out to assess the most common circulation patterns that gave rise to these episodes more frequently. The leading variability patterns of the sea-level pressure model were searched for by computing empirical orthogonal functions (EOFs) of winter anomalies. Figure 7a and b shows the third and fifth EOF pressure patterns, which explain, respectively, the 13% and 4% of the total variance during wintertime.
(a) Spatial patterns of the third (13% of variance explained) and (b) fifth empirical orthogonal function modes (4% of variance explained) of the sea-level pressure (hPa) in winter associated with extraordinary snowfall over all Italy (1750–2008). Red (blue) color indicates positive (negative) values. (The SLP database used is from Luterbacher et al., 2002.)
The third EOF, that is, the pattern dominating the high-frequency variability of snowfall episodes (Figure 7a) is characterized by a strong latitudinal gradient between western and south-eastern Europe, describing a north-easterly very cold circulation over Mediterranean area resembling the North Atlantic Oscillation (NAO) negative level. The sea-level pattern associated with the fifth mode EOF (Figure 7b) is characterized mainly by very low pressure cold polar air dragging over Mediterranean area and forming a close circulation.
Anomalies covering the Mediterranean area (Figure 7b) are always accompanied by significant, extraordinary, and extended snowfall episodes (humid and cold air masses circulation). However, the pattern, presented in Figure 7b, is not a pattern that persists for a longer time. On the contrary, the Siberian pattern (represented in Figure 7a) drives extraordinary snowfall that is less spatially homogeneous but temporally persistent with a very long and intense freezing phase across Italy due to the genesis of a blocking condition. These circulatory patterns are well documented by literary sources, handing down in their descriptions an actual picture of the situation (Table 4).
Snowfalls occurred in the whole Italian peninsula during the last three centuries. Please note that empty box in ‘Effects of the event’ column indicates that no particular effect has been registered by the sources.
An example of the pattern illustrated in Figure 7a is the 1755 winter: in central Italy (Marche Region), heavy snowfalls and intense cold occurred (Anselmi, 1989, doc. 13); between 31 January and 6 February, in the town of Catanzaro (southern Italy, Calabria Region), some 40 cm of snow fell and remained on the ground frozen for a long time:
Horrible freezing causing great damages to the cattle, and then, on 6 February, it fell so much snow that it was two span high and more, endangering the buildings that, due to the excessive weight, risked the collapse … (Moio and Susanna, 1977, doc. 11)
In the same year 1755, in January, Venice lagoon froze over; the ice layer was so thick that people and carriages could cross the distance between Venice and the mainland passing on it. Then, in the middle of the month, there was Sirocco wind lasting some days and then again, at the end of the month, icy and freezing wind: ‘Brenta river froze and it was possible to cross Adige river with cows and the chariots. On 28, 29, 30, 31 <January> and 1 February, […] icy air […]. On 6 […] in the night, snow’ (Benigna, XVIII sec., doc. 5).
During the month of December 1887, the exceptional snowfalls widespread in the whole peninsula were preceded in the southern part of Italy from rain and hailstorms, as described in the Journal of Meteorology and Agriculture of Rome:
Benevento (Campania region, southern Italy). In the night, very strong western wind, rainstorms, hailstorms with thunders and lightning, large hail and snow in the surrounding areas […] S. Bartolomeo in Galdo: snowfall lasting five days, south-western wind; no works in the country, everything is covered by the snow.
In Forlì (Emilia Romagna Region, northern Italy), the snow reached an exceptional height of 190 cm (Rosetti, 1894, doc. 44).
Impact of the snow on pre-industrial society
Snow events occurring in plain areas can impact society negatively in basically two fields: damage to the economy and direct damage to people. This latter is, of course, linked to the collapse of buildings and thus a risk to human lives. The first one approaches the effects of snow on crops and communications. Only the snow events are considered here and not their significance in terms of climate change and its impact on human society (Xoplaki et al., 2001, doc. 28).
The snow layer on the ground has an important biological function of protecting the underlying soil from freezing, while from the hydrological point of view, the slow melting allows greater water infiltration into the soil with large accumulation of water reserves – unlike liquid precipitation, which if too intense and lasting, leads to flocking of large quantities of water into the ground, which the soil is unable to absorb and therefore flows directly into streams, rivers, and lakes. Therefore, it follows that the snow also drastically reduces the hydrogeological risk in a given area in correspondence with intense precipitation events. Taking into account the wide time span and geographical area, only a general picture will be traced here, in order to understand whether and how snow events impacted society (Pfister and Brazdil, 2006); anyway, this picture is representative, due to the relatively slow changes in economical asset that occurred in the considered time range. Great differences due to climatic, historical, social, and political conditions existed between northern-central Italy and southern Italy. These differences, of course, also reflected the impact of snow events on society and are highlighted here.
In the past, as is well known, Italian economy was based on agriculture and was quite underdeveloped, when compared with the rest of Europe (Romano and Vivanti, 1973; Sella and Capra, 1984; Sereni, 1997; Villani, 1968; Zaninelli, 1990). Agricultural revolution in Italy took place slowly and late, in the second half of 19th century (Fumi, 2003). In the 18th century, abbot Giuseppe Toaldo, a famous meteorologist living in Padua, talking about snow and its effects on agriculture, claims,
Are these salts, these nitrates, this mucilage, this calcareous soil <contained in the snow flakes>, as already repeated, that constitute the nutritional substance for the plants. That’s why the grass, under the snow cover, grows green again soon and after a snowy winter it always follows an abundant harvest. Snow brings another benefit to the sown fields: it defends them against cold and frost. If the snow comes before the cold, nothing must be feared for the roots neither of crops, nor of trees. (Toaldo, 1775: 18–19)
These words of the scientist perfectly agree with popular, peasant knowledge: snow – unlike, of course, frost – was considered a favorable element, if falling at the due time: ‘Good the snow coming at the right time’, says a proverb; or another: ‘Under the snow bread, under the rain famine’.
Stock breeding, in central and northern Italy, consisting mainly of cattle and swine, was not so affected by the phenomenon because during the winter, animals were usually kept in stables and some stock of forage was provided. In some cases, snow could be disruptive: when occurring in a Siberian air pattern, accompanied by freezing, it caused damage to crops and particularly to arboriculture, or when occurring late in the season, that is, when crops were growing. Extreme cold could furthermore cause the death of cattle and thus shortage of dairy products and meat.
Conversely, in southern Italy, one of the problems caused by the snow coverage more frequently documented was the lack of animal feeding: the breeding consisted mainly of sheep and goats, pastured in grasslands, thus both the cold and difficulty to reach the food under the snow (Marino, 1988) often led to mortality of animals. It is well documented in the sources, for example, in the winter of 1828/1829: ‘During the last days of 1828 and the first ones of 1829, a high snow killed more than 300000 sheep in Puglia <southern Italy>’ (Various authors, 1844, doc. 27).
Of course, the production derived from the breeding activity was also damaged; a dramatic occurrence was in the winter of 1765/1766: ‘Horrible freezing and cold, and they lasted many days causing great mortality of animals, both bovines and ovine, so that no dairy products could be found’ (Moio and Susanna, 1977, doc. 11).
Generally speaking, snowfalls that impact the agricultural economy worst were – and are – the ones that occurred late in the season. This is mainly due to the direct damages they caused to the growing crops, and also due to their eventual melting, both in the plains and on the mountains, creating local stagnation of water or flooding due to the increase of the hydrological flow.
Arboriculture also could be damaged; sources describe some occurrence of mechanical fracture of branches due to the weight of the snow, but it was quite rare; furthermore, the production of fruits was mainly aimed to a limited utilization (e.g. for the family). The most quoted by the sources and impacting damages were those to the olive trees: they were more sensitive to the cold, and the shortage of olives – and consequently of olive oil – was very harmful.
Anyway, in the agrarian agreements of the 18th century, there was no mention of the snow among the so-called celestial accidents, that is, those natural accidents allowing the tenant to be compensated by the owner of the land. They were hailstorms, heavy rains, and floods. (Enzi et al., 2010; Vivanti, 1959).
In the past, as at the present time, other problems were caused by the blockage of communication – often lasting for longer periods than today, due to the minor efficiency of clearing technology – that in turn damaged the commerce; comparing the past with the present, apparently, the slowing down in trading created less damage to the population than today (Call, 2005), as movements across territories were less and transports too were slower and usually limited to shorter distances. Railways in Italy slowly developed only after the national unification, that is, after 1870; prior to that, commercial exchanges took place on dirt roads or on barges on rivers, and goods traveled mostly between the countryside and the nearest town, along short distances. When snowfalls were particularly heavy and a high snow cover lasted on the ground, difficulties in the transport of food and firewood supplies could cause an increase in the prices (Alfani et al., 2012).
An impact on society stronger than today is documented due to the occurrence of exceptionally abundant snowfall, blocking houses or villages: in these circumstances, rescuing people was slower and sometimes impossible. For example, a well-documented event occurred on the mountains of Piemonte (north-western Italy) in the winter of 1755, when the alternation of extreme cold and warm winds caused many avalanches, killing more than 200 people, and 3 women, blocked in their cowshed under the snow, survived there from 19 March to 25 April (Somis, 1758).
Conclusion
This study collects, reconstructs, and analyzes the extraordinary (in terms of intensity and/or long-lasting) snowfall episodes recurrence at different Italian plain sites since 1709, which was the year of the ‘big cold’ all over Europe. This is the longest homogeneously reconstructed snowfall time series in the central Mediterranean area. The reconstruction of extreme snowfalls was obtained by several documentary sources of different kinds (e.g. archives, diaries, manuscripts, journals, and web databases). These data provided detailed and complete information in terms of temporal and spatial extensions proving that, in recent times, the use of trusted descriptive sources to analyze extreme events is still a reliable option, for example, using database describing the areas historically affected by geological and hydraulic disasters. This methodological approach, supported by the comparison with instrumental data, allowed the reconstruction of a 300-year homogeneous snowfall intensity index and the understanding of the development of the weather phenomena with information on the forcing that lead the highlighted extreme events.
The results obtained from the analysis of the time series after the application of different statistical tests show that at the Italian latitude and at low altitude, the yearly recurrence of snowfalls has a low density (the average annual frequency is about 0.31 ± 1 event per year), and it is asymmetrically distributed (i.e. on the last 300 years more often, for 16 times, snowfall occurred with 1 event/year and only in two cases there were 8 events/year). The analysis of the time-series temporal stochastic memory appears to be close to a hole model in which a periodic structure similar to a sinusoidal wave forms peaks and troughs. This model fits quite well lower snowy episodes (below line of one standard deviation) that show a regular manifestation with a fairly uniform spread throughout both the end of the LIA (18th and 19th centuries) and the most recent warming period. On the long period, the temporal correlation of the time series points out a 60-year cycle that roughly matches with the results observed in the Fourier and wavelet analyses, providing sufficient evidences in recurrence of extraordinary snowy events.
The analysis of the documents allowed to study the cause–effect relationship and to trace the two most common synoptic configurations which led to extreme snowfall events over the Italian plains. These two recognized synoptic configuration were responsible for past extreme events as well as recent snowfalls that paralyzed the whole peninsula. The first recognized synoptic pattern (i.e. Siberian pattern) drives to temporally persistent snowfall episodes due to the genesis of a blocking condition. It is characterized by a strong latitudinal gradient between western and south-eastern Europe, describing a north-easterly very cold circulation over Mediterranean area. A recent example of such a pattern was the 28 January/15 February 2012 snowfall when two cold waves with powerful irruptions of cold air of continental origin (continental arctic air) from Russia and Siberia reinforced by the union of the Azores anticyclone with the Russian–Siberian anticyclone (i.e. the Wejkoff’s Bridge) persisted over the Italian peninsula for a quite long time. These air masses, due to the contrast with the mild and moist air over the Mediterranean basin, generated two cyclogenesis that led to a drop in temperatures with heavy snowfall in the plains, especially on the Adriatic coast and over the Tyrrhenian Sea. The second recognized synoptic configuration is characterized by very low pressure dragging cold polar air over Mediterranean area and forming a close circulation. This pattern usually does not persist for a long time.
The analysis of the described effects on the human life and society suggests that exceptional wintertime snow (in itself, without considering the association with other extreme events) had a possibly lower impact during the pre-industrial era than today. Few scientific literature has been found on this specific topic; most of the scholars focus their interest in the wider field of social vulnerability to climate change or climate fluctuations (Juneja and Mauelshagen, 2007; Pfister, 2005, 2010). Vulnerability is the key word in approaching the analysis of the impact of meteorological events on past societies; for the event considered here, the present industrialized society seems more vulnerable than the pre-industrialized one. Some differences exist between northern and southern Italy.
In agriculture, the wintertime snow had mainly a positive impact. Late snowfall could, on the contrary, damage the crops; anyway, the sources never mention a direct correspondence between snow and famine, the last one occurring only when snow was associated with an extremely cold period. In 2 years (1887 and 1901), the sources note that agricultural works were slowed down. No correspondence has been found between occurrence of snowfalls and increase in food prices.
In southern Italy, the snow was more harmful, having a worst impact on arboriculture, characterized by less resistant (Mediterranean) species such as olive and citrus trees. In breeding, snow had a very low impact in the northern part (if not accompanied by intense cold), but a higher one in the southern part of the peninsula, impeding, when its persistence on soil was prolonged, the feeding of the animals usually bred on the ground in open spaces (e.g. years 1789 and 1828/1829).
Communications, involving merchandises and workers, were stopped or slowed for a longer time than today, but they were naturally slower, less developed, and mostly covering short distances, so it seems that the damage was limited to delays in the usual flow of exchanges; furthermore, considering that until the late 19th century, the economy was almost entirely local and self-sufficient, holding up in exchanges was not so impacting (Armiero and Hall, 2010).
Social vulnerability was more evident in the occurrence of heavy snowfall: they could be very harmful for the population when causing collapses and/or isolating people in houses or entire villages; this is due to the backwardness of rescue technology and because the exceptional slowing of food and firewood transports could lead to an increase in prices and some hunger of cold-linked disease, although this effect is never described in the sources as dramatic.
Documentary sources
Please note that ancient source references cannot easily follow the modern citation systems. In particular, early journals were written by more authors, not individually quoted; they were a literary genre between books and modern scientific journals (see Table 4).
Corradi A (1865–1894) Annali delle epidemie occorse in Italia dalle prime memorie fino al 1850. Bologna: Forni.
Fiandrini B (sine data) Annali ravennati, Ravenna XVIII cent., Classense manuscript Mob. 3. 4. C, t. III. Ravenna: Classense Library.
Riva G (Cronaca di Modena dal 1702 al 1713) Soli-Muratori manuscript, filza 40, fasc. 2-3. Modena: Estense library.
Capezza V and Morici R (2004) Il Clima di Senigallia: Aspetti storici e profili attuali. Ancona: Sagraf.
Benigna A (sine data) Libro de memorie, Marciano manuscript cl. VII, cod. 1620, n. 7846. Venice: Marciana Library.
Giordano G (manuscript chronicle). In: Pisano G (ed.) Un’inedita cronaca Irpina del Settecento, Economia Irpina, 7/8, 1968.
Giornale de’ letterati pubblicato in Firenze (1744) book III, part I, Florence: Giovannelli.
Spanò Bolani D (1857) Storia di Reggio dai tempi primitivi sino all’anno di Cristo 1797. Naples: Stamperia e Cartiere del Fibreno.
Various authors (1857) Commentari dell’ateneo di Brescia per gli anni 1858–1861. Brescia: Apollonio.
Spallanzani L (1841) Lettere di vari illustri italiani al celebre abate Lorenzo Spallanzani. Reggio: Torreggiani & C.
Moio GB and Susanna G (1977) Diario di quanto successe in Catanzaro dal 1710 al 1769. Chiaravalle Centrale: Effe Emme.
Anonymous (1747) Almanacco per l’anno 1747. Florence: Paperini.
Anselmi S (1989) L’ampliazione di Senigallia, 1747–1762. Senigallia: Comune di Senigallia.
Toaldo G (1784) Opuscoli scelti sulle scienze e sulle arti. Milan: Marelli.
Arago F (1860) Sämmtliche Werke. Leipzig: Wigand.
Toaldo G (1770) Saggio meteorologico. Padova: Stamperia del Seminario.
Anonymous (sine data, 18th century) Freddi e geli in Venezia, Codex Gradenigo 200 XXI. Venice: Correr Museum Library.
Gennari G (1982–1984) Notizie giornaliere di quanto avvenne specialmente in Padova dall’anno 1739 all’anno 1800. In: Olivato L (ed.) Collana Scrittori Padovani. Cittadella: Rebellato.
Toaldo G (1797) Della vera influenza degli astri sulle stagioni e mutazioni di tempo. Padova: Bettinelli.
Levi MG (1845) Dell’inverno e di alcuni suoi effetti. Il Gondoliere XII 14: 105–108.
Various authors (1815) Memorie dell’Accademia d’agricoltura, commercio ed arti di Verona, vol. VI. Verona: Mainardi.
Gallicciolli GB (1795) Delle memorie venete antiche profane ed ecclesiastiche. Venice: Fracasso.
Mangianti F and Beltrano MC (1993) La neve a Roma dal 1741 al 1990. Rome: Ministero dell’Agricoltura e delle Foreste – UCEA.
Various authors (1808) Giornale d’agricoltura, vol. V. Milan: Silvestri.
Various authors (1836) Messaggiere Tirolese, n. 8. Rovereto: Marchesani.
Zerella G (1810) Quadro statistico sulla forma e produttività del suolo (manuscript). Ariano Irpino Library.
Various authors (1844) Annali civili del regno delle due Sicilie, vol. 34–36. Naples: real Ministero dell’Interno.
Xoplaki E, Maheras P and Luterbacher J (2001) Variability of climate in meridional Balkans during the periods 1675–1715 and 1780–1830 and its impact on human life. Climatic Change 48: 581–615.
Various authors (1830) Gazzetta Piemontese 4: 17.
Various authors (1836) Biblioteca italiana ossia giornale di letteratura. Scienze ed Arti 81: 350–354.
Fusinieri A (1847) Memorie di meteorologia. Padova: Sicca.
Tedeschi G in Laudato A. Un libro di memorie, Campolattaro, Centro Culturale per lo Studio della Civiltà Contadina nel Sannio 1989, p. 50.
Bresciani A (1869) Lettere familiari, descrittive ed erudite. Rome: Civiltà Cattolica.
Perrella A (1890) Effemeride della Provincia di Molise. Isernia: De Matteis.
Various authors (1848) Giornale arcadico di scienze lettere ed arti, 114: 185 (Rome: Tipografia delle belle arti).
Lombardini E (1846) Importanza degli studi sulla statistica dei fiumi e cenni intorno a quelli finora intrapresi. Giornale dell’I.R. Istituto lombardo di scienze lettere ed arti e biblioteca italiana compilata da varj dotti nazionali e stranieri, 42, I: 257–302.
Bellumé M, Maugeri M and Mazzuchelli E (1998) Due secoli di osservazioni meteorologiche a Mantova. Milan: CUSL.
Various authors (1860) Bullettino delle scienze mediche, IV, 13. Società medico chirurgica di Bologna. Bologna: Gamberini e Parmeggiani.
De Bosis F (1869) Meteorologia anconitana. Ancona: Cherubini.
Pasquale GA (1867) Catalogo del real orto botanico di Napoli. Naples: Ghio.
Andreotti Maio E (1958) Le nevicate a Napoli (1866–1957). Bollettino Società Naturalistica in Napoli, LXVII.
Ragona D (1871) Sui principali fenomeni delle variazioni diurne del calore atmosferico. Modena: Soliani.
Various authors (1887) Rivista Meteorico-Agraria, VIII: 561–576 (Rome: Ufficio Centrale di Meteorologia e Geodinamica).
Rosetti E (1894) La Romagna, Geografia e Storia. Milan: Hoepli.
Guerra C. http://www.nimbus.it/eventi/2012/120204NeveAggiornamento.htm.
Borzini G. http://forum.meteo4.com/showthread.php?14553-La-grande-ondata-di-gelo-del-Gennaio-1907.
http://www.frontedelpiave.info/public/modules/Fronte_del_Piave_article/.
http://www.aeris.toscana.it/articles.asp?id=5&mese=1&anno=2009.
Guerrieri E (1930) Il freddo straordinario dell’inverno 1928-29. Rivista di Fisica, Matematica e Scienze Naturali, II, IV, 6–7: 14–19.
Bassani G (2012) Cinque storie ferraresi. Milan: Feltrinelli.
http://it.wikipedia.org/wiki/Inverni_freddi_in_Europa_dal_XV_secolo.
Iamiceli D (1996) Sassinoro di ieri di oggi. Kissimmee, FL: Osceola.
Progetto AVI. Database of the areas historically affected by geological and hydraulic disasters. The archive contains historical information that took place during the 20th century in Italy. Database open source: http://sici.irpi.cnr.it/.
Diodato N (1997) Paesaggi d’inverno. Aspetti naturalistici e climatologici delle nevicate sulla Campania interna. Benevento: La Provincia Sannita.
http://www.centrometeoitaliano.it/l-ondata-di-freddo-del-7-8-aprile-2003/.
Various sources (newspapers) published in the dates of the events: Aeronautica militare, Corriere della Sera, Il sole 24 ore, Daily Telegraph, ANSA, La Repubblica, Meteoweb.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.