Analysing the Data

To reconstruct climate trends, historical climate research requires continuous, long and preferably homogeneous time series containing information that can be converted into quantitative data. Historical climate impact research, on the other hand, requires high-resolution temperature data and precise information on precipitation conditions, as both are important for societal activities.

Bio-physical proxy data make it possible to estimate temperatures for periods of several months. However, these multi-month estimation periods can be based on very different monthly temperature and precipitation patterns. For example, an early rye harvest can follow either an extremely warm month of May and a normal month of June, or a cool month of May and a very warm month of June, or even an extremely warm April a normal May and a June with only above-average temperatures. Weather descriptions in chronicles and weather diaries make it possible to reconstruct the conditions for individual months. However, these vary greatly depending on the author and are therefore difficult to categorise.

The balancing act between climate reconstruction and climate impact research is made possible by monthly or seasonal temperature and precipitation indices.

Indices are derived from all the historical data available for a given period. For this purpose, the messages are sorted chronologically and then interpreted for each month/season on the basis of a comparative assessment of all the material: A temperature index and a precipitation index are assigned to each month/season based on this comparison. The so-called Pfister indices (Mauelshagen 2010) are divided into seven classes.

a) Temperature

• -3: extremely cold
• -2: very cold
• -1: cold
• 0: without obvious positive or negative tendency
• +1: warm
• +2: very warm
• +3: extremely warm

b) Precipitation

• -3: extremely dry
• -2: very dry
• -1: dry
• 0: without obvious positive or negative tendency
• +1: humid
• +2: very humid
• +3: extremely humid

It is important to emphasise that these Pfister indices contain information about the ranking of a characteristic, but not about its magnitude. Nor can they be assigned units (e.g. specific temperatures or precipitation amounts).

Example: A month with a temperature index of -3 is colder than a month with a temperature index of -2, but no statement can be made about the magnitude of the difference. A basis is always needed for such comparisons. For the Pfister indices in Euro-Climhist, this basis for comparison is the period 1901-1960, as this is after the "Little Ice Age" but before the period of rapid global warming from the 1980s onwards. A month with a temperature index of -3 is therefore categorised as very cold according to the statistics for the 1901-1960 reference period. Temperature indices of levels 3, -2, 2 and 3 must also be based on proxy data that is statistically verified.

Indices for seasons are derived from the average of the monthly indices. Example: A seasonal index of -1.6 can be derived from monthly indices of -1, -1 and -3.

In addition, researchers are faced with the task that individual criteria are decisive for each climatic area in order to carry out a classification according to the Pfister indices. For example, a report that a lake was frozen over for 30 days in winter indicates an extremely cold winter for the Swiss Plateau (temperature index -3), but this finding would be more normal for northern Sweden and would therefore result in a rating with a temperature index of 0 or -1.

The derivation of temperature and precipitation indices from historical documentary data follows a methodology that has been increasingly refined in recent years in order to deal with a variety of individual sources whose spatial and temporal information varies in accuracy. Most of these sources do not contain quantitative data on weather conditions; they are not standardised in form or content and tend to be incomplete. Nevertheless, some of the individual observations can be quantified. This applies in particular to daily weather observations, which can be counted on a monthly basis. In addition, descriptions of extreme temperatures are often based on quasi-objective observations that can be compared over time. Such observations include indications of the - premature or delayed - state of vegetation, the duration or absence of snow cover, the freezing of bodies of water or the appearance of spring vegetation in winter. Descriptions of high or low water can be used as evidence of extreme precipitation or long periods of drought.

Climate-sensitive data is also available in the bookkeeping of institutions. In some archives, for example, the time of the grain harvest or grape harvest is recorded year after year, often over centuries. Statistical methods can be used to estimate temperatures for periods of several months from corresponding time series, as is done for data from natural archives, such as tree rings.

In order to estimate temperature and precipitation indices, all available data for a month or a season must be included. They should be mutually supportive and show a plausible meteorological trend. With the increase in Swiss data material from 1550 onwards, a temperature and precipitation index can be assigned to almost every month.

A good example of the derivation of monthly temperature and precipitation indices is provided by the weather in April 1731 in Switzerland:

Euro-Climhist: Query result for April 1731 (without weather damage)
1731-April 1-10 / snowmelt: complete / Canton Nidwalden
1731-April 11-20 / permanent snow cover / Canton of Appenzell-Innerrhoden
1731-April 21-25 / cold / Canton of Nidwalden
1731-April 21-30 / large amounts of snow / Winterthur (ZH)
1731-April / permanent snow cover: several days / Canton of Nidwalden
1731-April / cold / Canton of Geneva
1731-April / easterly wind / Bätterkinden (BE)
1731-April 30 / extremely long snow cover / Winterthur (ZH)
1731-May 7 / Cherry trees blossom 127 days after New Year, extremely late / Central Plateau
1731-April / Pfister temperature index: -3 extremely cold / Central Plateau
1731-April / Precipitation total 102 (mm): wet / Zurich (ZH)
1731-April / 11 days with precipitation: rather dry / Winterthur (ZH)
1731-April / Pfister precipitation index: 0 average / Central Plateau

In its entirety, this data field corresponds to an extremely cold April. This can be derived from several indicators: an extremely long snow cover in Winterthur and in Nidwalden (Stans), the prevailing very cold Bisen situation (north-easterly wind) and the extreme delay of the cherry blossom in the Central Plateau. This is mainly dependent on the April temperatures (Rutishauser, Studer 2007).

A team of authors led by the Czech climate researcher Petr Dobrovolný (Dobrovolný et al. 2010) has estimated the monthly temperatures since 1501 from the temperature indices available for Germany, the Czech Republic and Switzerland from 1501 and the subsequent instrument data from the 18th century onwards. Their data is available in Euro-Climhist as monthly, seasonal and annual averages for the period from 1500 onwards. It is also important that the size of the potential estimation error can be specified.

Societies are particularly sensitive to extreme weather conditions and their consequences. Modern examples are often attributed, without reflection, to man-made climate change. Historical cases, on the other hand, are often used to deny or trivialise man-made climate change on the grounds that such extremes have "always" existed. However, individual events generally have little to do with long-term climate development because they cannot be analysed statistically. A Siberian winter does not an ice age make, and even the hot summer of 2003 was surpassed by the unique eleven-month heat and drought in 1540 (Wetter et al. 2014). However, the year 1540 remained a singular event of the century, whereas most summers since 2003 must be categorised as extremely hot (+3) or very hot (+2) compared to the reference period 1901-1960.

The history of weather and climate does not end with the establishment of today's MeteoSwiss monitoring network in 1864. Conversely, questions about weather and climate history often arise from current weather events. For this reason, selected long series of instrumental and proxy data are continued in Euro-Climhist up to the present day wherever possible.

For a climate-historical interpretation, the results are presented on the one hand in the form of time series, and on the other hand they can be compared with other long series based on data from natural archives, preferably tree rings and known reconstructions of Alpine glaciers.

However, caution is required when interpreting certain proxy data. Since 1808, for example, the leaf sprouting of a horse chestnut tree in the centre of Geneva has been recorded annually. Since the late 19th century, there has been a strong trend towards earlier leaf emergence and earlier flowering. However, this is mainly due to the extreme warming caused by urban development, as illustrated by the comparison of leaf emergence in Geneva with the first flowering of the horse chestnut in rural Hallau (SH) (Wetter, Pfister 2014).